Is it only the university they are satisfied with? – Foreign student satisfaction and its effect on loyalty

The internationalization of higher education has become the center of research interest in the past years (Buckner & Stein, 2020; Garwe & Thondhlana, 2021; Ghazarian, 2020). Even though there are certain contradictions – according to which higher education institutions (hereinafter HEIs) emphasize the importance of mobility and studying abroad, but at the same time they are eager to keep their students – the number of foreign students has been on the rise in recent years all over the world (Restaino et al., 2020; Van Mol et al., 2020). With the upsurge in international student numbers, there has been an increasingly renowned interest in investigating international students’ service quality expectations, satisfaction, and loyalty (Lovemore et al., 2020; Rehman et al., 2020), as satisfying students’ needs is of key importance at today’s higher education environment for retaining students and ensuring positive word-of-mouth recommendations (Alsheyadi & Albalushi, 2020; Landrum et al., 2021).

Some studies have been uncovered, which are concerned with foreign student country-, and institution-specific expectations and satisfaction (de Souza Câmara et al., 2020; Williams, 2020). In the majority of these research papers, examined factors are closely related to university service quality and include factors such as academic services, academic facilities, administrative services, HEI performance, and employee orientation (Alfy & Abukari, 2020; Moslehpour et al., 2020).

Non-university-related – potentially satisfaction and loyalty altering – factors beyond service quality are rarely studied or categorized (Alfano et al., 2021; Faizan et al., 2016; Mihanovic et al., 2016) but are not unprecedented. Amaro et al. (2019) revealed that the perception of the country, the environment and location of the HEI influence satisfaction, while the global perception of the HEI influences recommendations and increases student satisfaction and loyalty. As several factors beyond HEI service quality influencing student satisfaction and loyalty have been revealed, the topic seems crucial to investigate further for more successful strategic marketing decisions for HEIs to increase international student numbers (Amaro et al., 2019).

Loyalty of students has been a widely studied phenomena. It has mostly been examined from the service quality perspective (Alves & Raposo, 2009) and was found to be influenced by satisfaction (Alfano et al., 2021; Huybers et al., 2015). Pedro et al. (2020) found that loyalty of students included strong positive feelings, sense of belonging and pride of being part of the university. There has also been evidence that students are willing to give back to the university (Pedro et al., 2020). This suggests that student loyalty – and most importantly the meaning behind student loyalty – is a phenomenon to be further studied and investigated, so that student commitment – and thereby loyalty – can be enhanced.

Based on the above, the aim of this paper is to reveal what university- and non-university-related factors influence foreign student satisfaction and how satisfaction influences student loyalty. Furthermore, the study investigates if different types of student loyalty can be differentiated between, on which satisfaction might have an influence. Therefore, the paper intends to add to the higher education marketing literature by determining factors that may influence student satisfaction and contribute to loyalty, thereby further effecting student retention, positive word-of-mouth recommendations, student commitment, more effective student recruitment, and the creation of a successful marketing strategy.

In the paper, a research model is developed and analyzed with structural equation modeling (SEM), with the method of partial least squares (PLS) and with cluster analysis. These methods are widely used in previous higher education marketing literature for the measurement of expectations, satisfaction, and loyalty (Faizan et al., 2016; Savitha & Padmaja, 2017; Amaro et al., 2019).

The current paper investigates the connection between several notions. Therefore, the main theoretical aspects (expectations, satisfaction, and loyalty) used and examined throughout this paper are defined in the higher education marketing context specifically related to foreign students. Firstly, foreign students are determined as those students who come from foreign countries (except for neighboring countries) and have a foreign nationality. The definitions of expectations, satisfaction and loyalty are investigated and determined from the viewpoint of and are relevant to foreign students exclusively.

The primary research method of the paper is an online questionnaire, which was available to fill between March and April 2019. Constructs were measured with Likert-scale questions, as their usage in higher education marketing studies is an internationally accepted methodology (Gronholdt et al., 2000; Turkyilmaz et al., 2018). Expectations scales were measured based on Østergaard and Kristensen (2005), while university-related and non-university-related satisfaction measurement relied on studies from Owlia and Aspinwall (1996), Mihanovic et al. (2016) and Machado et al. (2011). Foreign student loyalty was measured partly by scales used in previous studies (Østergaard and Kristensen, 2005; Alves & Raposo, 2009), and scales developed by the authors. These scales’ reliability was tested with Cronbach-alpha and were deemed suitable for further analysis.

The method used for analysis included PLS-SEM and cluster analysis. The examined theoretical concepts of expectations, satisfaction and loyalty were studied as latent variables. To examine the hypotheses, the relationship between the latent variables was investigated with structural equation modeling (SEM), as this methodology can be applied higher education marketing research (El-Hilali et al., 2015; Giner & Rillo, 2016; Kazár, 2014). In this study, the partial least squares (PLS) technique (Hair et al., 2014) can be applied, as the variables are not normally distributed (in case of both Kolmogorov–Smirnov and Shapiro–Wilk tests and in case of each variable it is p < 0,01). SmartPLS 3 software was used for the PLS path analysis (Ringle et al., 2015). Additionally, to classify foreign students participating in the survey into separate sub-groups, cluster analysis was conducted, as it is a methodologically accepted way to classify university students into sub-groups in higher education marketing studies (Saenz et al., 2011; Caliskan et al., 2013; Gallyamova et al., 2018; Bennasar-Veny et al., 2020). The aim of the cluster analysis was to determine whether students could be classified based on their loyalty.

The sample of the study was drawn from international students studying full-time at the chosen university in Hungary. The University of Szeged was the subject of the current study, as it has been welcoming foreign students for more than 30 years. With a continuously growing number of study programs available for foreign students, this university presents a good opportunity for the current research, as results might be of key importance in the marketing directions the institution intends to take to attract and retain more foreign students. At the time of the research, about 2505 foreign students were studying at the university, while the sample consisted of 188 students. The sample size can be considered relatively small due to no direct contact to foreign students and their low willingness to fill in surveys. However, regardless of the sample size, the current study can provide us with a better understanding on their satisfaction and loyalty, based on which further research could be conducted on a bigger sample. Respondents’ country of origin was varied, they arrived at the destination from more than 40 countries and studied almost on every faculty of the examined university. Due to the nature of the survey, it cannot be deemed representative.

The constructs’ validity was examined with Cronbach-Alfa and CR (composite reliability) indicators in the outer model. Results suggest that each examined construct reaches the minimum value (> 0.6 Hair et al., 2009). Convergence validity was examined with standardized factor weights and AVE (average variance extracted) indicators. Factor loadings above 0.6 and below 0.7 were also accepted (Hair et al., 1998; IRM, 2014), as AVE indicators exceeded the minimum value (> 0.5 Hair et al., 2014) in case of each latent variable, after two variables were excluded from the analysis. Therefore, the existence of all remaining constructs is validated. Discriminant validity was examined based on the test of Fornel and Larcker (1981) and the heterotrait-monotrait ratio of correlations (HTMT) (Henseler et al., 2015). Each variable met the Fornel and Larcker criterion, as variables’ AVE values were higher than the squares of the correlation coefficients between the construct and the other constructs. The HTMT of each examined variable was smaller than one (Henseler et al., 2015). Therefore, discriminant validity is established. The outer model suggests the existence of the latent variables and each indicator represents the same phenomenon.

The bootstrap algorithm helped to test the path coefficients’ significance regarding the inner model (Hair et al., 2014). The number of iterations were 5000. The results show that 7 paths can be considered significant, while 8 paths are non-significant in the model (H1, H2, H3a, H3d, H3e, H3f, H4c, H4e). The latter ones mean that expectations do not have an effect on either university-, or non-university-related satisfaction (H1, H2); tangibles, attitude and delivery (H3a, H3d, H3e, H3f) do not affect university-related satisfaction; public facilities and job opportunities (H4c, H4e) do not influence non-university-related satisfaction.

After leaving out the non-significant effects from the model, the PLS algorithm was run again, and each remaining path has a significant effect at a five percent significance level. The significant paths can be seen in Table 1. In case of the outer model, running the PLS algorithm did not affect the criterion values (Cronbach-alpha, CR, AVE, standardized factor loadings, Fornell-Larcker criteria and the HTMT).

After the exclusion of non-significant paths, the final model can be seen on Fig. 2. There are positive paths observed in each case. Standardized path coefficients (β) show that:

Results show that H3b, H3c, H4a, H4b, H4d, H5 and H6 can be accepted. Competences and content of curriculum (H3b, H3c) affect university-related foreign student satisfaction, while living in the city, international environment, and places to spend free time at (H4a, H4b, H4d) influence non-university-related foreign student satisfaction. Both satisfaction types have an effect on foreign student loyalty (H5, H6).

As factors influencing foreign student satisfaction and loyalty are in the center of the current research, indirect and total effects should also be examined in the model. The indirect effect of competences on foreign student loyalty (β = 0.225) materializes through university-related foreign student satisfaction (= 0.312*0.720), similarly to the effect of content of curriculum (β = 0.349) on loyalty (= 0.485*0.720).

Places to spend free time at has an indirect effect on foreign student loyalty (β = 0.037) via non-university-related foreign student satisfaction (= 0.275*0.135). Living in the city has a similarly weak but significant indirect effect on foreign student loyalty (β = 0.078) via non-university-related foreign student satisfaction (= 0.578*0.135). Based on the R² values in the ellipses, the exploratory forces in the model could be considered moderately strong. It is important to highlight that even though the path coefficient is higher between university-related foreign student satisfaction and loyalty, than between non-university-related foreign student satisfaction and loyalty, the R² value is higher in case of non-university-related foreign student satisfaction (R² = 0.712) than that of university-related foreign student satisfaction (R² = 0.541).

It is also important to highlight the significance of the effects between the variables in the model. This is based on the f² indicator, which examines the change in the determination coefficient of the endogenous variable when an exogenous variable is omitted (Hair et al., 2014). Based on Table 2 it can be stated that weak, medium, and strong effects can be observed in the model. The effects of university-related foreign student satisfaction on foreign student loyalty and living in the city on non-university-related foreign student satisfaction can be considered strong. Medium effects can be observed between places to spend free time at and non-university-related foreign student satisfaction, and competences and university-related foreign student satisfaction. Additional and less strong effects should also be considered, as all these paths can be considered significant at a five percent significance level (Table 3).

In order to understand the examined institution’s foreign student loyalty deeper and to see if foreign student loyalty types can be differentiated between, factor and cluster analysis were conducted based on the loyalty scale items previously added to the survey as a result of previous in-depth interviews. Creating new variables was possible based on the reliability analysis. Therefore, foreign student loyalty scales were examined with Principal Component Analysis. The KMO value was 0.907 and according to the Bartlett’s test of sphericity, factor analysis could be conducted. After Varimax rotation a two-factor result was concluded and factor loadings above 0.5 were accepted (Hair et al., 1998; IRM, 2014). The first factor could be named as foreign student university loyalty, as items related to foreign student university were grouped together. The second factor included items connected to the foreign student experience of studying abroad, so it is named foreign student experience loyalty.

On the basis of the factor analysis, cluster analysis was conducted to determine the loyalty of respondents. Hierarchical cluster analysis was conducted with Ward method and Euclidean distance. The dendrogram suggested several clusters but based on the scatter plot diagram (Fig. 3), a three-cluster solution was accepted. The first cluster includes foreign students who seem to be more loyal towards the university, while the second involves those who are more loyal to the study-abroad experience. Therefore, the first cluster is named “university loyal foreign students” (41.5%) and the second is “experience loyal foreign students” (46.8%). Foreign students in the third cluster do not seem to be loyal towards either university or experience, so they are named “Not loyal foreign students” (11.7%).

These findings on clusters reveal those foreign students participating in the research can be categorized regarding their loyalty based on what they are loyal to. As the current study is not representative in nature, conclusions can only be drawn to the sample. Further cross tabs findings show that most university loyal students are fee payers and participate in undivided 5-year degree programs (e.g.: in the field of: medicine, law, dentistry), while experience loyal students come from bachelor’s, PhD, and undivided programs with scholarships (e.g.: in the field of music, economics, philology, or IT). Non-loyal students can be found on almost every program level, except for PhD programs. Foreign student loyalty was found to develop over time spent in a higher education institution, as foreign students tend to reflect on themselves as loyal, when they are in the middle or towards the end of their studies.

This research introduces a new hypothetical model for understanding what factors influence foreign student satisfaction and loyalty. The novelty of the model lies in the fact that non-university-related factors influencing foreign student satisfaction are distinguished between. The model was tested with PLS path analysis and bootstrapping, which are widely accepted methods in HEI studies (El-Hilali et al., 2015; Giner & Rillo, 2016; Lee, 2010). Moreover, the study uncovered the categorization of foreign student loyalty at the examined university with the help of cluster analysis (Bennasar-Veny et al., 2020; Caliskan et al., 2013; Saenz et al., 2011; Sultan & Wong, 2013a).

The results show that foreign student expectations do not have a significant effect on neither university- (H1), nor non-university-related foreign student satisfaction (H2). These results contradict previous literature, which revealed significant effects thereof (Alves & Raposo, 2009; Shahsavar & Sudzina, 2017; Gregory, 2019; Pezeshki et al., 2020). Interestingly, tangibles had no effect on university-related foreign student satisfaction (H3a), as previously seen in the literature, where Mekic and Mekic (2016) found the opposite. Similarly to this, attitude did not influence university-related foreign student satisfaction (H3d), even though prior studies showed significant effects (Ahmed & Masud, 2014; Lenton, 2015). Neither reliability (H3e) nor delivery (H3f) affected university-related foreign student satisfaction, even though both Yousapronpaiboon (2014) and Chui et al. (2016) found otherwise. The same tendency appeared in connection with public facilities (H4c) and job opportunities (H4e). As contrary to previous findings (Aldemir & Gülcan, 2004; Mihanovic et al., 2016; Smith, 2020), no significant effects were found between them and university-related foreign student satisfaction.

Significant path findings correspond to previous higher education marketing literature, as teachers’ competences (H3b) and the content of the curriculum (H3c) have a significant effect on university-related foreign student satisfaction (El-Hilali et al., 2015; Lenton, 2015) and they also have an indirect effect on foreign student loyalty. Additionally, living in the city (H4a), the international environment (H4b) (Machado et al., 2011; Schertzer & Schertzer, 2004) and places to spend free time at (H4d) (Abdullah, 2006; Mihanovic et al., 2016) also have a significant effect on non-university-related foreign student satisfaction. Living in the city and places to spend free time at both have an indirect effect on foreign student loyalty as well. All in all, both non-university (H5) and university-related satisfaction (H6) have a strong effect on foreign student loyalty, the latter of which was already discussed in previous studies mostly related to service quality (Alves & Raposo, 2009; Shahsavar & Sudzina, 2017).

Foreign student loyalty was studied in depth, but contrary to certain previous literature (Douglas & Davies, 2008; Sultan & Wong, 2013a), not in a qualitative way. Even though previous studies revealed different loyalty clusters and categorizations of students (Gallyamova, 2018; Bennasar-Veny et al., 2020), new light is shed on foreign student loyalty with the differentiation of university-, experience foreign student loyalty and lack thereof. These findings could specifically be useful for the chosen university and might encourage other HEIs to investigate their foreign student loyalty deeper.

This research was undertaken to reveal what factors influence university- and non-university related foreign student satisfaction, and whether foreign student satisfaction has an effect on foreign student loyalty. The second aim of this paper was to investigate foreign student loyalty deeper. Returning to the aims declared at the beginning of this study, this research has identified that both university- and non-university-related foreign student satisfaction exist as latent variables, competences and content influencing university-related foreign student satisfaction; living in the city, international atmosphere, and places to spend free time at affecting non-university-related foreign student satisfaction. Both types of satisfaction were found to have an effect on foreign student loyalty. The second major finding was revealed by cluster analysis showing that university-, experience and lack of foreign student loyalty can be differentiated between in case of the examined foreign students.

These findings have significant implications for the understanding of how foreign student expectations could be defined and measured in higher education institutions. If defined as a basis for subjective comparison (Oliver, 1980) or a forecasting factor (Churchill & Surprenant, 1982), unbiased expectations could be asked from freshmen. Therefore, the measurement of foreign student expectations prior to or upon student arrival in the target country (Gerdes & Mallinckrodt, 1994) is suggested. However, if examined as recalled assumptions – as it was done in this paper – expectations can be biased and regarded as a constantly changing and developing phenomenon (whether asked from a second- or fifth-year student) and can further depend on the study phase and study-field of the foreign student (e.g.: expectations from the field of medicine or business) (Woodruff et al., 1983).

Taken together, the findings suggest that foreign student satisfaction relates to the whole study abroad experience and both university- (Alfy & Abukari, 2020; Moslehpour et al., 2020) and non-university-related satisfaction (Alfano et al., 2021; Mihanovic et al., 2016) can be defined and distinguished between. Even though in the current study only five variables (competences, content, international environment, living in the city, places to spend free time at) influenced foreign student satisfaction, non-significant path-related variables can be included in the overall study of foreign students, as results may vary based on foreign student study track, level, and phase.

The data of this study suggest that loyalty can be achieved through both university- and non-university-related satisfaction. Moreover, foreign student loyalty can be comprehended according to the complex approach, as it includes emotional attachment, commitment, and possible word-of-mouth recommendations (Oliver, 1999; Reichheld, 2003). The results of the cluster analysis further implies that there are different underlying factors of foreign student loyalty. As opposed to the suggestions on expectations and satisfaction, the research results on loyalty imply the possibility of measuring foreign student loyalty on an institution-wide sample, based on which similarities and differences between study fields and levels can be found.

The findings of this research provide insights for higher education institutions into foreign student expectations, satisfaction, and loyalty. The principal implication of the current study for higher education institutions is that the continuous measurement of these notions has to be supplemented with specific study track- and study phase-related research potentially examining different levels of studies (bachelor, master, PhD) separately. A joint analysis of foreign students was proven to be useful and provides grounds for further studies but has raised several questions to be investigated.

Finally, a number of important limitations need to be considered. First, the research included a lengthy online questionnaire. Another limitation lies in the fact that the paper relied on a convenience sample including foreign students from different study tracks and phases. Moreover, the results of this survey are limited solely to foreign students at a Hungarian university. Therefore, additional studies are recommended to test the applicability of the examined model more in depth and in different HEIs as well.

Although the current study is based on a small sample of participants, it offers insight into foreign student satisfaction and loyalty. The paper establishes a quantitative framework and model for the measurement of several constructs. This approach is hoped to prove useful in expanding our understanding of foreign student satisfaction and loyalty at higher education institutions in the future.

The main strength of this study is that it represents a comprehensive examination of foreign student satisfaction and loyalty. However, this research has also raised many questions in need of further investigation. Firstly, the study of a less complex cohort of foreign students is needed in case of expectations and satisfaction, including only one specific faculty or department, where foreign students may have similar backgrounds and interests (e.g.: in job opportunities). The current study investigated a university with 12 different faculties, where willingness to work besides studies may vary deeply among foreign students of medicine and business (Gallyamova et al., 2018). Secondly, significant paths – especially the novel one between non-university foreign student satisfaction and foreign student loyalty – should be investigated with foreign students from different academic programs and years separately to determine any differences that might arise from these factors mentioned. Thirdly, foreign student loyalty may either be studied as three separate latent variables (university loyalty, experience loyalty, and lack of loyalty), on which university-related, non-university-related and overall foreign student satisfaction’s effects could be measured, or an institution-wide sample should be drawn to highlight any differences between study fields and levels. The conceptual model set and tested in the current research could be the basis of studies conducted in specific faculties of the institution, and it could be further adapted to other universities in Hungary and Europe as well. Another direction could be to further research the reason why some foreign students are not loyal. This would enable the HEI to enhance service quality, possibly satisfy foreign student students’ needs to a greater extent and thereby retain them, as retaining a customer may be less expensive than attracting new ones (Reichheld, 1996). As mentioned above, the measurement of the constructs included in this study can be adopted at a faculty level and conducted at a regular basis to ensure continuous feedback and potential service improvement (Chui et al., 2016).