Data collection

The study focused on domestic and foreign tourists aged 18 and above participating in rafting at Antalya Köprülü Canyon. The questionnaires were applied immediately after rafting in-person, and it was thought that the tourists’ experiences were reflected. In order to accurately measure tourist motivations, the literature was reviewed and scales were selected from the literature. In the process of selecting the scales, previously experienced ready-made scales were used, however, the scales were preferred from ready-made scales with high values in terms of validity and reliability. The aim here is to measure the constructs measured in the study in the most reliable way and in a way that can be distinguished from other constructs. For this purpose, scales with high Cronbach α or composite reliability values and AVE (average variance extracted) values were preferred. Then the convenience sampling method was used as the sampling method because there was no random selection. It is a statistical fact that the convenience sampling method does not represent the whole population because it is not random. However, it is easier to apply than random sampling in terms of reaching individuals with new experiences. In addition, as a result of studies that can be carried out in other countries or regions, although it is not a random sample, new literature becomes more debatable and converges to a scientific reality with the literature obtained with the convenience sampling method together with the developing literature.

Participants voluntarily participated in the survey after the rafting experience. Thus, it can be stated that the participants’ responses to the questionnaire were not influenced by any incentives. This situation causes the participants’ views on the subject to be more sincere. The questionnaires were collected in 2021. From 327 questionnaires, 31 were excluded for incomplete data, leaving 296 for analysis. The demographics included 68.1% Russian, 21.5% EU citizens, 9.5% Turkish, and 1% from other nationalities, reflecting general tourism trends in Turkey as reported by the World Travel and Tourism Council (2021). According to the World Travel and Tourism Council (2021) report, 12% of those who came to Turkey in 2019 were Russian, and 8% were German tourists, while in 2020, this rate was 13% for Russians and 7% for Bulgaria, Germany and Ukraine. In this case, it is predicted that the data and results obtained from the target audience will provide correct inferences. Therefore, bias in the study poses as much risk as bias that can occur in real life.

Although 50% of the participants have visited Antalya before, the rate of those who have visited Köprülü Canyon before is 24.3%. In this case, it can be said that individuals who have visited before have returned home without rafting in Köprülü Canyon. The rate of those who have rafted before is 29.7%. The fact that the rates of those who have visited Köprülü Canyon and those who have rafted are close may indicate that individuals tend to do it again after the first experience. While 62.4% of the participants were female, 37.6% were male. In this case, it can be stated that women are more oriented towards adventure tourism. 12.2% of the participants are high school graduates, 21.3% are associate degree graduates, 57.4% are bachelor’s degree graduates, and 9.1% are master’s and doctorate graduates. The average age of the participants was 33.36, while the median was 33.

Measures

Five-point Likert-type scales assessed all constructs. The scales covered “experiencing nature” (Perić et al., 2019), “escape” (Carvache-Franco et al., 2019), and “joy” (Pestana et al., 2020). The “WOM” influence (Sirakaya-Turk et al., 2015) and “self-image congruence” (Sirgy et al., 1997) were also measured, along with “revisit intention” (Zhang et al., 2018).

Data analysis and results

The data analysis validated the measurement model and evaluated relationships between the constructs.

Measurement model

The confirmatory factor analysis (CFA) indicated a satisfactory fit, with chi-square/df at 2.81, CFI at 0.92, SRMR at 0.059, and RMSEA at 0.078 (Hu and Bentler, 1999), as detailed in Table 1.

Construct validity was confirmed, with convergent and discriminant validity assessed and meeting established thresholds (Hair et al., 2014) as shown in Table 2.

According to Tables 1 and 2, the AVE values are greater than 0.50 and the correlation between the variables. Therefore, convergent and discriminant validity is provided. After this stage of the analysis, common method bias (CMB) or common method variance (CMV) was examined. According to Podsakoff et al. (2003), CMB analysis refers to the bias that emerges from external factors on the data set and occurs when the majority of the variance is explained by a single factor (Podsakoff et al., 2003; Gaskin and Lim, 2016). To measure whether the majority of the variance was gathered under a single factor, the single factor Harman test was performed, and the explained variance rate was calculated as 43%. Since the single factor Harman test is a weak analysis, CMB was re-examined using the Controlling for the effects of an unmeasured latent methods factor analysis suggested by Podsakoff et al. (2003), and each regression path was calculated as 0.64, and the explained variance rate was found to be 41%. Since CMB does not exceed 50% (Eichhorn, 2014: p. 8), it can be said that CMB does not exist or is insignificant (Büyükdağ and Kitapci, 2021).

Structural model

The structural model’s evaluation produced the following results: chi-square/df value at 2.81, CFI value at 0.92, SRMR value at 0.059, and RMSEA value at 0.078. These indices satisfy the criteria set by Hu and Bentler (1999), indicating a good fit between the theoretical model and the observed data.

Table 3 shows that push factors significantly and positively influence self-image congruence (β = 0.66), WOM (β = 0.55), and revisit intention (β = 0.32). Self-image congruity also significantly enhances WOM (β = 0.35) and revisit intention (β = 0.30), while WOM positively impacts revisit intention (β = 0.28). The model explains 44% of the variance in self-image congruity, 68% in WOM, and 66% in revisit intention (Fig. 2).

Multi-group structural equation modeling (SEM) and analysis results

Multi-group structural equation modeling (SEM) was employed to compare the regression paths between two variables based on socio-demographic and field-specific characteristics. Various studies have utilized this approach: Yada et al. (2018) to understand teachers’ attitudes and self-efficacy, Al-Swidi and Al Yahya (2017) to examine educational intention and work behavior differences by gender, and Babin et al. (2016), Huang and Ge (2019), Murray et al. (2017), and Aka and Buyukdag (2021) to analyze factors such as culture, household characteristics, store design, and marital status. In this study, multi-group SEM was applied to explore the effects of rafting experience (first-time vs. repeated) and gender (female vs. male model).

According to the multi-group SEM related to rafting experience, the model showed good fit indices with a chi-square/df value of 2.19, a CFI of 0.90, an RMSEA of 0.064, a GFI of 0.79, and an AGFI of 0.73. The comparative analysis between unconstrained and constrained models revealed a chi-square difference of 35.06 and a df difference of 25, indicating no significant variation between the effects of rafting experiences (p = 0.087). Consequently, the research model is applicable to both first-time and repeated rafters. The significance of each path’s rafting experience was further analyzed and is detailed in Table 4.

According to the multi-group structural equation modeling focused on gender, the model demonstrated good fit indices with a chi-square/df value of 2.17, a CFI of 0.90, an RMSEA of 0.063, a GFI of 0.79, and an AGFI of 0.73. This suggests that the multi-group SEM adequately represents the gender-based differences in the data. Comparative analysis between unconstrained and constrained models showed a chi-square difference of 24.83 and a df difference of 25, indicating no significant variance in gender effects (p = 0.472). Therefore, the research model is equally applicable to both female and male categories. Further analysis was conducted to determine if significant differences exist in local paths based on gender, with detailed results presented in Table 4.

Table 4 shows that push factors affect self-congruence differently for first-time versus repeated rafters. Rafting experience moderates how these factors influence self-image congruence, with a more pronounced effect on first-timers. While push factors significantly impact WOM for both groups, the effect is stronger for newcomers, but rafting experience doesn’t moderate this relationship. Similarly, push factors notably influence revisit intention for first-time rafters, but less so for experienced rafters, where experience doesn’t act as a moderator (Fig. 3).

The influence of self-congruence on WOM is significant for both novice and seasoned rafters, more so for the latter. This suggests that rafters with prior experience, and with higher self-image congruity, are likelier to share their experiences. Self-congruence significantly affects intention to revisit among experienced rafters, but not for newcomers. However, rafting experience does not moderate these relationships in either case.

The impact of WOM on revisit intention was significant for first-time rafters but not for repeat rafters, with rafting experience not moderating this relationship. Table 4 shows variance differences between these groups. For first-timers, the explained variance is 51%, while only 26.9% for repeat rafters. For WOM, the variance is 70.5% for first-time users and 66.5% for repeat rafters. Regarding revisit intention, the variance is 65.3% for novices and slightly higher at 66.1% for experienced rafters.

The model showed no significant gender-based moderating effects, but coefficients highlight important relationship nuances. Both genders experience a positive, significant effect of push factors on self-image congruence, with males showing a higher coefficient. The impact of push factors on WOM is significant for both, yet stronger for males. Females, however, demonstrate a greater influence of push factors on revisit intention. The effects of self-congruence on WOM are similar across genders. Males exhibit a more substantial influence of self-image congruence on revisit intention. WOM’s impact on revisit intention is marginally higher in males. While gender doesn’t significantly moderate these paths, the data suggest males typically have higher values in consumer experiences involving adventure and risk-taking.

Self-congruence significantly influences WOM for both first-time and repeat rafters, more so for the latter. This suggests experienced rafters, likely with higher self-image congruity, are more prone to sharing their experiences. Self-congruence also impacts revisit intention significantly among experienced rafters, but less for novices. In both cases, rafting experience does not moderate these relationships.

The study shows gender differences in variance rates for self-image congruity, WOM, and revisit intention. Self-image congruity explains 49.3% of the variance in males and 38% in females. For WOM, the variance is 81.6% in males and 57.7% in females. Regarding revisit intention, males have a variance rate of 68.1% compared to 65.7% in females. These results imply that self-image congruence is more prominent in male first-time rafters, who also tend to discuss their adventurous experiences more, indicating higher communication about risk-taking and adventure among males.

Study 2

A multiple correspondence analysis examined relationships between push factors, self-image congruence, and demographics in adventure tourism for greater insight into consumer behavior dynamics.

Multiple correspondence analysis

Multiple Correspondence Analysis (MCA) is a robust multivariate technique used to examine relationships among nominal data. This method allows researchers to analyze data, interpret findings, and develop perceptual maps, facilitating a deeper understanding of the data structure (Hair et al., 2010; Hair et al., 2014). In this study, MCA was employed to analyze the relationships between individuals’ perceptions of push factors, self-image congruence, WOM, and revisit intentions, alongside demographic or social factors such as gender, nationality, rafting experience, and visiting status. The objective was to conduct in-depth research and derive meaningful inferences. The graphical representation from the Multiple Correspondence Analysis is provided in Fig. 4.

According to the multiple correspondence analysis (MCA) results, repeat visitors to Köprülü Canyon are predominantly Turkish, male, and have prior visits to Antalya and rafting experience. These individuals are notably influenced by push and pull factors and demonstrate high self-image congruity, WOM, loyalty, and satisfaction, indicating a strong intention to revisit. Conversely, first-time visitors to Antalya and Köprülü Canyon are primarily Russian and female tourists, characterized by their pursuit of excitement, unique experiences, and experiential pleasure in adventure and risk-taking activities. Despite showing a high intention to revisit, the likelihood of Russian and female tourists returning is relatively low. This pattern suggests that while tourists enjoy adventure tourism as part of their sea, sun, and sand vacation, it is not the primary purpose of their visit. The findings imply that although tourists have significant rafting experiences and entertainment, they are more inclined to explore different geographical regions rather than revisit the same location. Consequently, it is expected that these tourists will likely choose alternative destinations for their next vacation.

Therefore, emphasizing promotions targeting first-time visitors in rafting or adventure tourism is anticipated to yield significant benefits. Consequently, catering to the preferences of Russian and female tourists with diverse adventure and risk-taking tourism options is projected to create a vital market segment. However, the analysis indicates that European tourists exhibit lower levels of self-image congruity, WOM, revisit intention, and satisfaction with push and pull factors related to rafting. As such, understanding the specific expectations of tourists from the European Union and offering varied tourism alternatives could become a significant source of revenue. Addressing these preferences may lead to enhanced tourist experiences and increased revisit rates.