Antecedents of loyalty intentions among young adult tourists: a survey

One of the attributes of tourism demand is its restitution [Dziedzic and Skalska, 2012], caused by the constant renewal of tourists’ needs. An emerging tourist need may be associated with a desire to repeat a similar trip or may manifest itself in a preference to visit the same location but in a different form or for a different purpose (e.g. different accommodation, meals, different trip organization). Given how widespread this phenomenon is, the issue of loyalty and its determinants becomes particularly relevant in the tourism sector.

Customer loyalty is a broad and complex concept. It refers to the overall behavior towards a specific offer, company, product, facility, institution, person or place. Oliver [1997] believes that full loyalty occurs when a customer is deeply convinced that he will buy a particular product again in the future and recommend it, regardless of the market conditions and other contextual factors that may cause a change in their behavior.

Seweryn [2012] pointed out that current trends in consumption (hedonism, sublimation, and individualism) may result in a lower inclination to stay loyal. All manifestations of loyalty behavior can have a positive impact and bring benefits to vendors [Reichheld et al., 2000]. Retaining existing customers may be cheaper than acquiring new ones [Fornell and Wernerfelt, 1987]. Loyal customers are also more likely to share positive information about the product with their friends, relatives, and other potential consumers [Foscht et al., 2009]. Meleddu et al. [2015] indicated that the tendency to recommend a tourist product is a result of being satisfied with staying at a destination. At the same time, different elements of the destination offer may diversely affect the level of satisfaction, and consequently revisit intention (RVI) and share positive information. Therefore, understanding the reasons for tourist loyalty to destinations may help regions and locations compete in an increasingly challenging market [Chen and Gursoy, 2001].

One of the factors differentiating loyalty behavior in tourism may be the age of travelers. Members of different generations tend to demonstrate different willingness to return to places already visited. In this article, the preferences of young adults – the so-called Generation Y – will be investigated.

The manuscript presents the results of the author’s own research conducted on a sample of young adults. Constructed on the basis of a literature review, concepts used in the research tool allowed for the estimation of a structural model showing the relationships that exist between the novelty and variety seeking (NVS), uncertainty and risk perception, social bonding (SB), and loyalty inclination expressed through its two dimensions: RVI and intention to recommend the destination.

Having identified the research gap, the author decided to conduct a survey to collect data from Generation Y members to test the six research hypotheses.

In this study, the author models the relationships between destination loyalty dimensions and their antecedents with a framework that involves five composite variables (or constructs): 1) RUP, 2) NVS, 3) SB, 4) RI, and 5) RVI. The links among the variables and pertinent hypotheses are illustrated in Figure 1.

Figure 1

The model indicates that motivations and attitudes of tourists may influence their loyalty behavior. The research purposefully omits the specific features of the visited destination and the tourist’s satisfaction with the stay, which are most frequently studied by other researchers in the context of loyalty. The author’s intention was to determine general factors that might be important when returning to or recommending a destination, having more to do with individual attributes of the tourist rather than the specifics of a particular destination.

In order to collect empirical data, an Internet questionnaire was used. It was divided into three parts. The first part included questions about previous tourist experiences and motivation for travels. Questions in this part were constructed using nominal and Likert scales. The second part of the questionnaire referred to the tendency to revisit and to share opinions about the visited place, as well as the reasons for (not) returning to the visited places. Questions were designed with the use of Likert scales. Likert statements used for measuring the constructs are presented in Table 4. The questionnaire concluded with demographical items.

The survey was conducted on a group of 305 respondents recruited from the students of the SGH Warsaw School of Economics and the WSB University in Gdańsk from October 2018 to April 2019. All respondents were aged 18–32 years. Groups of students are a common subject of research in the tourism sector (e.g. Michael et al., 2003; Gallarza and Saura, 2006; Kim et al., 2006; Xu et al., 2009; Kim and Park, 2016). This is because this group of young adults is very active in tourism and involved in many tourist trips. It is partly because they have more free time to spend on traveling. They are also involved in many different tourism activities, e.g. active, cultural, educational, and leisure tourism. So, their experiences are very differentiated and potentially interesting for researchers. In addition, most students are unmarried and have no children, which allows them to make purchasing decisions based on their own preferences and likes, without the need to adjust to the needs of other family members. Thus, it can be assumed that the opinions expressed in the research reflect their own needs and not those shaped by the circumstances they find themselves in.

Table 1 describes demographical characteristics of the sample.

Because of a filter question at the beginning of the questionnaire, all respondents had traveled for tourist purposes at least once in the five years preceding the interview. The vast majority declared traveling both in Poland and abroad (89.2%). A total of 5.6% of respondents traveled only abroad, while 5.2% visited only Polish destinations.

The initial statistical analysis was performed using SPSS 25. In the second phase, to explore regression paths among variables, SmartPLS 3.2.8 was used.

It seems that out of the two dimensions of destination loyalty, respondents are more likely to share opinions (both positive and negative) about visited places than to return to them (see Table 2). This may be related to the popularity of Internet tools and social media among Generation Y, and social media are among the most commonly used channels for sharing opinions about trips and destinations [TripAdvisor, 2016].

Regarding frequency of travel to the same places and not returning to previously visited ones, the average level of indications was similar and relatively low (2.18 and 2.32, respectively; there were no statistically significant differences between the mean values). This should be interpreted as a lack of a dominant preference in this regard – somewhat surprisingly, the young people were equally likely to travel to the same locations as new ones.

What is more, the respondents did not consider traveling to familiar places as overly boring (2.69). Curiosity about new destinations probably limits the tendency to return to those already visited, but with the mean value of 3.08, this is not a very strong sentiment. On the other hand, respondents recognized traveling to well-known places as somewhat less risky (3.09) but rather did not agree with the opinion that visiting a new place may be stressful (2.06) (see Table 3).

The main statistical method for testing the research hypotheses was a structural model estimated by the partial least squares (PLS) method. According to the best practices outlined in the literature, the first step in investigating a partial PLS model is to establish the quality of the so-called measurement model, which defines relationships between latent constructs and their indicators.

According to the procedures recommended by Hair et al. [2017], the metrics for average variance extracted (AVE) and composite reliability (CR) were calculated in order to verify the correctness of the investigated constructs (Table 4). For each factor, CR was greater than 0.7 and AVE exceeded the 0.5 threshold. These figures, combined with the absolute values of factor loadings exceeding 0.6, demonstrate the high reliability of all the extracted factors [Hair et al., 2017].

For further confirmation of the quality of the constructs, discriminant validity was investigated to check if sufficient differences exist between latent variables. Fornell and Larcker [1981] suggested that discriminant validity could be established when the square root of AVE for each construct is higher than the correlation coefficients between the focal construct and all the other constructs in the model (Table 5). Since all the correlations are smaller than the square roots of AVEs, it can be concluded that all latent variables show appropriate separation from each other and there are no big overlaps in their meaning.

To provide additional insights about the quality of the model, Stone–Geisser’s Q² metrics were computed for the two endogenous constructs, using the blindfolding procedure with an omission distance of 7. Their respective values for RVI and RI were both positive and equal to 0.196 and 0.061, respectively, which point to predictive relevance of the structural model [Hair et al., 2016, p. 212].

An extension of R² values are f² coefficients that capture changes in R² when specific exogenous constructs are omitted from the model, drawing on the drop in R² before and after the omission. The metric is standardized with three recommended thresholds conventionally used for interpretation [Cohen, 1988]: 0.02 (small effect), 0.15 (medium effect), and 0.35 (large effect). In the current analysis, the strongest effect found was the link between novelty seeking and RVI (f² = 0.391). Other predictors, showing small but significant effects, were found for the regression paths: risk and uncertainty avoidance – RVI (f² = 0.050), emotional solidarity – RI (f² = 0.052), incomes – RI (f² = 0.023), and incomes*risk and uncertainty avoidance – RI (moderating effect, f² = 0.035).

Even though the main object of the PLS method is maximizing the predictive accuracy of endogenous constructs, and not – unlike the covariance-based structural equation modeling (CB SEM) – minimizing errors between empirical and model-implied covariance matrices, it is useful to see how well a PLS model performs in this regard. Accordingly, if it shows small errors (residuals) between the empirical and theoretical covariance matrixes, it can be assumed to not only work well as a predictive tool but also to accurately reflect hypothesized relationships embedded into the model. To this end, one can use the metrics of standardized root mean square residuals (SRMR) and root mean square residual covariances (RMS theta). It is generally considered that for CB SEM, a good fit with empirical data is indicated by SRMR <0.08 and RMS theta <0.12. At the moment of this writing, there were no widely acceptable thresholds for PLS SEMs, but many researchers were in agreement that the values of 0.08 and 0.12 were too restrictive [Henseler et al., 2015; Hair et al., 2016, p. 204]. In the current study, the coefficients of SRMR and RMS theta were almost exactly at the cutoff point (0.083 and 0.189, respectively), which corroborates the adequate quality of the analytical framework.

Once the quality of the measurement model was found acceptable, standard regression weights and p-values between constructs in the structural model were used to test the hypotheses (Table 6).

From Table 6 three estimated regression paths were statistically significant, which gives support to three hypotheses: H.2, H.4, and H.5.

Specifically, there is a statistically significant relationship between RUP and RVI. The greater the RUP level, the more tourists are likely to return to the places they have already visited.

The standardized regression weight values for NVS and RVI indicate a negative correlation, which means that higher levels of NVS tend to coincide with a lower inclination to revisit already known places.

The analysis also shows that RI is significantly influenced by SB so that higher SB levels are linked to greater RI.

To account for the possibility of dissimilar regression patterns in distinct demographical groups, the author investigated possible moderation effects of the focal relationships with gender and income. The significance tests for all possible two-way interactions showed that males and females are no different in terms of the investigated regression weights. Indeed, the only significant moderation effect was found for income and the way it affects the association between RUP and RI. It transpired that for average levels of income, RUP and RI are not significantly correlated. However, for very large and very small incomes (about 2 standard deviations from the mean), RUP has a significant influence on RI. Interestingly, the direction of the relationship is not constant and is positive for respondents with low incomes and negative for those with high earnings. This moderating effect of income provides partial support for hypothesis H.1, which can be deemed validated for those tourists whose incomes are substantially lower than the average. In this group, RUP has a positive impact on RI.

To conclude this overview of the research results, it should be noted that the estimated structural model explained 11.1% of variance in RI and 36.4% of variance in RVI. The diagram of regression paths between research constructs is presented in appendix (see Figure A.1 in Appendix).

The aim of the study was to address the existing research gap concerning the loyalty to tourist destinations of young adults. The hypotheses and the conceptual model were developed from previous research by other authors, in particular the works by Kozak [2001], Boksberger and Craig-Smith [2006], Kyle et al. [2005], Chi and Qu [2008], and Kim and Kim [2015]. The unique aspect of the study was investigating loyalty intentions in general and not toward a particular place, as was the case in most previous research.

The study demonstrates that members of Generation Y are split between the willingness to travel to familiar places and getting to know new destinations. It seems that a factor that can be a potent influence on them in trying to encourage loyalty behavior is SB. It was found that the social context represented by SB has a significant positive impact on RI. Han et al. [2019] in their research on community-based tourism came to a similar conclusion. Although a broader context of the sense of belonging was analyzed, of which SB is one aspect, it was observed that higher levels of this category correspond with a greater willingness to make recommendations.

In the present research, traveling to new destinations is not generally perceived as a stressful situation, but RUP is not a negligible factor and could play a role in consumer decisions. This was validated by a significant positive regression link demonstrating that high uncertainty and risk association with new places can lead to the choice of well-tested destinations. Rittichainuwat and Chakraborty [2009] also found that tourists may be less interested in destinations with high-risk perceptions. On the other hand, it should be stressed that the most important discouraging factor from visiting the same place is NVS. Similar observations can be found in Kim and Kim [2015], who report that novelty seeking decreases the overall level of satisfaction with travel and thus reduces the tendency to revisit.

An interesting finding was the role of income in the investigated conceptual framework. It turns out that the tourist’s income should be considered in anticipating their intentions to recommend familiar destinations. This effect is twofold: first, there is a significant positive main effect of the income variable, implying that those with higher incomes are more willing to share opinions with all the other variables kept constant (ceteris paribus rule). Second, there was a significant moderating effect of income on the regression path from RUPs to RIs. If an individual’s level of perceived risk and uncertainty is high, those with low incomes will share their opinions more willingly, while more affluent young tourists will demonstrate the opposite tendency.

The moderating influence of gender on factors affecting loyalty intentions has not been confirmed. Apparently, there are no differences between young men and women in terms of the explored processes leading to the formation of loyalty intentions. This might suggest that when catering to Millennials, tourist offers do not have to be differentiated by gender. A similar conclusion was reached by Chi et al. [2009] who concluded that male and female tourists displayed comparable loyalty toward the destination analyzed by them.

It should be stressed, however, that the author’s study did not investigate destination preferences of tourists in depth. So even though factors like RUPs seem to work in the same way for both genders, the specifics of what attributes of a tourism destination make it more risky or exciting could be different for males and females. This is one of the limitations of this study and also a suggestion for further research.

This study has a number of limitations. First, the collected sample allows to generalize findings to the populations of students from only two Polish universities located in two large urban centers of the country. The findings from a broader sample, encompassing universities in other parts of Poland, especially in less populous areas may give different results. Also, the Millennials from countries other than Poland could possibly yield different results. One of the possible directions for future research could be a comparison of different generation cohorts to see if the same factors contribute to loyalty intentions. In addition, the study could benefit from the inclusion of other control variables, to better model interpersonal differences between respondents; as a result, new interesting interaction patterns could be revealed. A replication of the study in other country settings could demonstrate if the relationships found here are only specific to Poland or could be considered universal for the whole Generation Y.

The developed model can be used as a guide for tourist enterprises and institutions managing tourist destinations to prepare more successful products and more effective promotional initiatives. For example, when preparing a tourist offer for young adults who are familiar with the destination from previous visits, it is advisable to introduce new activities in response to their expectation of variety and novelty. Because of the significance of RUP, it is instrumental to design marketing communications so that they provide sufficiently exhaustive and persuasive information to lower this sentiment. The important role played by SB points to possibly effective tropes and motives in the content of promotional campaigns, as well as suggests that social media could be a viable communication channel. To further leverage the positive effects of SB, social media could be used to create online communities around tourist destinations to foster loyalty intentions.