Towards Sustainable Rural Tourism: An SEM Study of Tourist Satisfaction in Thai Nguyen’s Tea Craft Villages

In the context of developing rural tourism and experiential tourism in Vietnam, traditional craft villages play an important role in preserving cultural values, creating livelihoods for the community, and contributing to diversifying tourism products [1-3]. Thai Nguyen is known as the “tea center” of the country, with dozens of craft villages growing, processing and trading in tea. Thai Nguyen tea products have built a strong brand in the domestic and foreign markets, considered a symbol of Vietnamese tea culture [4].

However, in reality, the number of tourists visiting Thai Nguyen tea villages is still limited compared to its potential. Although tea is a well-known specialty, the number of tourists choosing to experience tea villages is not high; in particular, the rate of visitors returning for a second or multiple times remains low. Some studies show that factors such as infrastructure quality, services and community participation have an important impact on the development of craft village tourism, but in Thai Nguyen, there are still many limitations in building experiential services [5, 6].

Specifically, tea villages in Thai Nguyen currently still tend to the traditional production model: planting, harvesting, drying tea and packaging. Tourists mainly come to buy products, take a few pictures at the tea hills, and then quickly leave. Meanwhile, tea cultural experience services such as tea tasting ceremonies, visiting the processing process, participating in product packaging, or ancillary services such as cuisine, accommodation, and souvenirs have not been properly invested [1, 5]. The lack of rich experiential activities and synchronous support services reduces the attractiveness of the destination, thereby directly affecting the satisfaction level and intention to return of tourists [4, 7].

From a research perspective, this raises the question: What factors determine the satisfaction of tourists in Thai Nguyen tea craft villages? Answering this question not only helps explain the reasons for the limited number of visitors and low return rate, but also provides a practical basis for orienting policies and solutions to develop tea craft village tourism in a sustainable direction. Recent studies on rural tourism and craft village tourism often focus on a number of aspects such as authenticity, destination image, service quality, experience quality and perceived value [1, 3, 5]. However, few studies apply structural equation modeling (SEM) to simultaneously analyze these factors in the specific context of tea craft villages, which both have agricultural economic value and are unique cultural and tourism spaces.

The selection of the topic "Determinants of Tourist Satisfaction in Tea Craft Villages: An SEM Approach from Thai Nguyen, Vietnam" comes from two main reasons. First, in terms of practice, the study aims to accurately reflect the current situation of Thai Nguyen tea craft villages, pointing out gaps in tourism product development and key factors affecting customer satisfaction. The research results will help localities, tourism businesses and craft village communities have a scientific basis to redesign products, diversify activities, and improve service quality, thereby attracting more tourists and increasing the return rate. Second, academically, the study adds empirical evidence to the theory of tourist satisfaction in the context of rural tourism in developing countries, where the relationship between authenticity, experience, destination image, perceived value, service quality and satisfaction has not been fully tested [3-5].

3.1 Descriptive statistics of respondents

Based on n = 320 valid samples, the characteristics of the respondents show a fairly balanced sample structure and are suitable for SEM analysis, while suggesting some potential differences that need to be tested.

Table 2. Respondents’ demographic profile (n = 320)

According to Table 2, in terms of gender, women account for 58.1% and men 41.9%. This structure suggests the possibility of gender differences in expectations/experiences in constructs such as experience quality, service quality and perceived value.

In terms of age, the 30–45 age group accounts for the largest proportion (41.6%), followed by over 45 (31.9%) and under 30 (26.6%). The age structure is biased towards the mature and middle-aged groups that often have higher expectations of authenticity and perceived value, but are also sensitive to service factors.

In terms of place of residence, guests from Thai Nguyen province account for 57.8%, guests from other provinces 31.9%, and international guests 10.3%.

In terms of occupation, the distribution is quite dispersed: business employees 27.8%, students 22.5%, civil servants 13.1%, and others 36.6%.

In terms of education level, the group below university level accounts for 54.4%, while university level 41.6% and postgraduate level 4.1%. This structure suggests that it is necessary to pay attention to the communication/interpretation language at the destination at an understandable level, while controlling for education when analyzing perceived value and destination image, as education is often related to cultural sensitivity.

In terms of income, the distribution is relatively balanced: <10 million (24.7%), 10–20 million (38.4%, middle group–possibly median), ≥ 20 million (36.9%). There is a basis to test the differential effect by income on perceived value and satisfaction.

3.2. Descriptive analysis of factors

Table 3 shows that the average (Mean) of the factors in the study ranges from 3.32 to 3.55 on a 5-point Likert scale. The factor with the highest average value is TS = 3.55, reflecting that, in general, tourists are quite satisfied when participating in tourism in Thai Nguyen tea villages. Other factors, such as AU = 3.53 and PV = 3.52, also reached relatively high levels, showing that tourists positively evaluate the authenticity and value received. Meanwhile, SQ = 3.32 has the lowest score, reflecting the limitations in service at the destination, which may be an issue that needs priority improvement.

Table 3. Descriptive statistics for tea craft village factors and tourist satisfaction

In terms of dispersion, most factors have standard deviations (SD) from 0.61 to 0.79, demonstrating that the level of consensus among tourists is at an average level, not too dispersed. Notably, TS has the lowest SD (0.61), indicating high consistency in tourists' satisfaction assessments. In contrast, EQ = 0.79 and SQ = 0.78 have larger standard deviations, reflecting differences in perceived experiences and services.

In summary, the descriptive results show that Thai Nguyen tea village tourism is quite positively evaluated by tourists, especially in terms of satisfaction, authenticity and perceived value. However, service quality is still limited and there are significant differences between tourist groups, which is an important suggestion for the management and development of village tourism products in the future.

3.3 Reliability analysis

The results of reliability analysis using Cronbach’s Alpha presented in Table 4 show that all scales have values above 0.8, exceeding the acceptable threshold of 0.7. This proves that the scales used in the study have high reliability and ensure internal consistency between observed variables.

Table 4. Cronbach’s Alpha for factors

Specifically, SQ has a Cronbach’s Alpha coefficient = 0.891 and EQ = 0.87, reflecting high stability in measuring service quality and tourism experience. Similarly, the factors DI = 0.843, PV = 0.854 and AU = 0.874 also show a solid level of reliability, ensuring that statements measuring these concepts are consistently evaluated by tourists.

Notably, TS has Cronbach’s Alpha = 0.951, the highest among the scales, although it only includes 3 observed variables. This shows a very high consensus of tourists in assessing the level of satisfaction when participating in tea village tourism in Thai Nguyen. This result also confirms that tourist satisfaction is a stable and less variable measured construct.

In summary, all scales ensure high reliability, allowing the study to continue to use them in the next analysis steps such as EFA, CFA and SEM to test the research model.

3.4 Exploratory Factor Analysis (EFA)

The EFA analysis results in Table 5 show that the KMO coefficient = 0.871, exceeding the threshold of 0.8, proving that the data is suitable for factor analysis. The Bartlett test has a Chi-square value = 4721.862 with p < 0.001, confirming that the difference correlation matrix is statistically significant. Six factors were extracted with eigenvalue > 1, consistent with the proposed research model. The total explained variance reached 74.948%, exceeding the standard of 50%, showing that the six factors explained most of the variation in the data. Thus, the EFA results confirmed the initial validity of the scale and created a solid foundation for the next CFA step.

Table 5. Results of Exploratory Factor Analysis (EFA)

3.5 Confirmatory Factor Analysis (CFA)

The results of the model fit test (Model Fit) show that all indices meet the recommended level, confirming that the measurement model has a good fit with the survey data. Specifically, the Chi-square/df index (CMIN/DF) = 1.485 < 3, reflecting an acceptable level of fit. The comparison indices such as NFI = 0.934; IFI = 0.977; TLI = 0.973; CFI = 0.977, all exceed the threshold of 0.9, showing that the model has a high level of explanation and compatibility with the observed data. In addition, the calibration indices such as PNFI = 0.794 and PCFI = 0.831 are both acceptable, confirming the stability of the model. In particular, the RMSEA index = 0.039 < 0.05 with PCLOSE = 0.982 > 0.05 shows that the model has a very good fit, and there is no significant difference between the hypothetical model and the actual model. Thus, the entire fit index results allow to confirm that the measurement model is appropriate, ensuring the basis for continuing the steps of analyzing convergent value, discriminant value (CR, AVE, HTMT) and testing the SEM structural model.

Table 6. Summary of CFA results

The CFA results presented in Table 6 confirm that the scales in the study all achieved reliability and convergent validity. Specifically, all factors had CR > 0.7 and AVE > 0.5, exceeding the acceptable threshold according to international standards. This proves that the observed variables measure latent concepts well, ensuring internal consistency and explanatory power of the scale. In particular, TS achieved CR = 0.954 and AVE = 0.873, the highest among the factors, reflecting the outstanding stability and reliability of the satisfaction scale. Other factors such as service quality (0.892; 0.674) and experience quality (0.873; 0.633) also showed very good measurement ability. Even factors with lower values such as destination image (0.844; 0.575) and perceived value (0.858; 0.603) still meet the standards, confirming their validity. Thus, the entire measurement model in the study of tea village tourism in Thai Nguyen is confirmed to be reliable and valid, creating a solid foundation for the SEM analysis step to test the research hypotheses.

Finally, the study assessed common method variance (CMV) to ensure data quality. Harman’s single-factor test results indicated that the first factor accounted for 32.48% of the total variance, which is well below the recommended 50% threshold. Furthermore, a Common Latent Factor (CLF) analysis was conducted, showing negligible differences in standardized regression weights when comparing the baseline model with the CLF model. These findings confirm that CMV is not a significant concern in this research.

Table 7. HTMT ratios among constructs

The results of the discriminant value analysis using the HTMT index (Table 7) show that all coefficients are smaller than the threshold of 0.85. This confirms that the concepts in the model have clear differences and do not overlap in measurement. Specifically, the factor pairs with the highest HTMT values are EQ - TS = 0.631 and SQ - TS = 0.637, reflecting a close relationship but still ensuring discriminant value. The remaining pairs such as DI - SQ = 0.091 or DI - AU = 0.103 have quite low values, showing a strong distinction between these constructs. Thus, all scales in the study met the standards of discriminant validity, ensuring the validity of the measurement model and providing a solid basis for testing the SEM structural model in the next step.

3.6 Structural model and hypothesis testing

In this section, the study conducts structural equation modeling (SEM) to assess the relationship between factors and tourist satisfaction at tea villages in Thai Nguyen. SEM allows for the simultaneous consideration of both measurement models and structural models, thereby providing comprehensive evidence on the suitability of the research hypothesis. Hypotheses H1–H5 are tested based on estimated coefficients and statistical significance levels. The test results are illustrated through the SEM diagram below.

According to Figure 2, the model fit indices show that the SEM model achieves very good quality. The Chi-square/df index = 1.628 is less than 3, reflecting the acceptable fit of the model to the survey data. The comparison indexes such as CFI = 0.970; TLI = 0.965; IFI = 0.970; NFI = 0.926 all exceed the threshold of 0.9, confirming that the model is capable of reproducing experimental data at a high level. In particular, the RMSEA index = 0.044 (< 0.05) together with PCLOSE = 0.867 (> 0.05) show that the model is not only suitable but also optimal, that is, the difference between the hypothetical model and the actual data is not statistically significant.

From a practical perspective, these results confirm that the research model on the relationship between factors such as service quality, experience quality, authenticity, perceived value and destination image on tourist satisfaction in Thai Nguyen tea craft villages is built in a reasonable and reliable way. This creates a solid foundation for analyzing hypotheses, thereby providing scientific evidence to help plan policies and strategies for developing craft village tourism in a sustainable direction and improving tourist experiences.

Table 8. Standardized regression weights, standard errors, critical ratios, and hypothesis testing

The results of hypothesis testing presented in Table 8 show that all five hypotheses (H1–H5) were supported with high statistical significance (p < 0.05, with which the majority achieved p < 0.001). This proves that the factors in the research model all have positive and significant impacts on tourists’ satisfaction when visiting tea craft villages in Thai Nguyen.

First of all, SQ has the highest standardized impact coefficient (β = 0.295, C.R. = 5.186, p < 0.001), showing that service quality is the most important factor determining satisfaction. This is consistent with the characteristics of craft village tourism, where tour guides, support and customer service play a key role in enhancing the experience.

Next, DI also showed a strong influence (β = 0.271, C.R. = 5.679, p < 0.001). This result confirmed that destination image – including the popularity, cultural identity and brand reputation of Thai Nguyen tea village – has a positive impact on tourists' feelings and satisfaction. When the destination image is communicated and built effectively, tourists' attachment and trust are strengthened. EQ has an impact coefficient of β = 0.249 (C.R. = 4.432, p < 0.001), reflecting that the quality of experience at the destination (participating in tea processing, learning about local culture, interacting with artisans, etc.) also plays an important role in creating satisfaction. This result is consistent with the trend of experiential tourism, where tourists value direct participation and the uniqueness of activities.

Meanwhile, AU also has a significant impact (β = 0.180, C.R. = 3.798, p < 0.001). The authenticity of production activities, cultural spaces, and the way people preserve traditions have contributed to improving customer satisfaction. This factor shows that tourists are not only looking for simple tourism products, but also want experiences associated with authentic cultural and historical values.

Finally, PV has an impact coefficient of β = 0.136 (C.R. = 2.519, p = 0.012), the lowest influence among the five factors, but still statistically significant. This implies that when comparing the costs and benefits, tourists still evaluate positively and are satisfied, although this factor is not the most decisive.

Notably, the model explains 54.5% of the variance in tourist satisfaction (R² = 0.545), an acceptable level in social studies, confirming that the factors selected in the study have high practical significance.

In summary, the results of hypothesis testing show that tourist satisfaction at Thai Nguyen tea village is simultaneously influenced by many factors, in which service quality, destination image and experience quality play a key role, while authenticity and perceived value play a supporting role. This suggests that managers and tourism businesses need to focus on improving service quality, while promoting cultural and traditional values, in order to increase satisfaction and attract tourists to return.

This study tested the SEM model of tourist satisfaction with tea village tourism in Thai Nguyen with 320 observations. The results showed that five factors–service quality, destination image, experience quality, authenticity and perceived value–all had positive and statistically significant impacts on tourist satisfaction, in which service quality, destination image and experience quality were the strongest influencing factors. The model explained 54.5% of the variation in satisfaction (R² = 0.545), confirming its relevance and high application value in the practical context.

The study has contributed to expanding empirical evidence on craft village tourism – a field that is less studied than heritage and cultural tourism. The results emphasize the combined role of service quality, unique experiences, destination image and authenticity in enhancing satisfaction, thereby contributing to sustainable tourism development in the locality.

The findings of the study have several important academic implications:

First, the study reinforces the causal relationship among quality, value and satisfaction, which has been demonstrated in the context of heritage tourism [18, 19].

Second, the finding that destination image has an impact as strong as service quality suggests the unique role of agricultural-craft village destinations, as opposed to leisure tourism, where service is often dominant [16].

Third, the study adds evidence to the role of authenticity, in line with recent studies on heritage tourism [20], which emphasize that “living heritage” has a special value in enhancing experiences and satisfaction.

Finally, the results expand the application scope of SEM in the field of community and craft village tourism, contributing more empirical data from a province with a long-standing cultural tradition like Thai Nguyen.

Managerial Implications: From a practical perspective, the study offers some important management implications:

For businesses and tourist attractions:

Standardize service quality: Tour operators and tourist attractions need to develop a unified set of service standards (from reception, guidance, service, hygiene, to food safety). Regular training activities are needed for employees and collaborators to ensure stable service provision.

Innovate experiential products: Experiential service packages should be diversified, such as tea hill tours, handmade tea-making classes, tea tasting, and tea souvenir making. These products need to be designed into a closed-loop experience chain to increase visitors' length of stay and spending.

Apply digital technology: Promote online communication, ticket booking and payment via applications, provide digital guides, and smart maps to enhance convenience and increase perceived value.

For local authorities:

Building a destination brand: There needs to be a strategy to promote the image of “Thai Nguyen – Vietnam’s Tea Capital” associated with local cultural identity. This can be implemented through international tea festivals, social media campaigns, and cooperation with the press and media agencies.

Tourism support infrastructure: Authorities should prioritize investment in roads to craft villages, bilingual sign systems, public restrooms, and parking lots to improve convenience and destination image.

Conservation and development policies: Issue a mechanism to support artisans and households participating in tourism activities; encourage community tourism models (homestay, farmstay) to both preserve culture and increase income.

For the community and artisans:

Preserving authenticity: People and artisans need to continue to maintain traditional processing procedures, organize ceremonies and tell stories about tea as part of the tourism product. This helps increase authenticity and differentiate from other destinations.

Improving tourism participation capacity: Communities need to be trained in communication skills, welcoming international guests, and tourism safety knowledge. The active and professional participation of the community is a decisive factor in the sustainability of the craft village tourism model.

Connecting local products: Encourage the development of additional tea-based ancillary products (cosmetics, creative drinks, tea-based cuisine) to diversify choices for tourists, increase perceived value and extend the consumption chain.

In summary, the above administrative and policy implications show that to improve satisfaction and sustainable development of tea village tourism, there needs to be synchronous coordination among businesses, government and community. When these three subjects act consistently, service quality, experience, destination image and authenticity will be strengthened, thereby increasing the perceived value and satisfaction of tourists, contributing to the sustainable development of tourism in Thai Nguyen province.

Limitations and Future Research:

Despite the positive results, the study still has some limitations. Firstly, the survey scope is limited to Thai Nguyen, not compared with other craft villages in the country. Secondly, the model does not consider the mediating role of emotions or behavioral intentions, which have been shown to be important in recent studies. Therefore, further studies should expand the survey space, add mediating and moderating variables (e.g., demographic characteristics, travel experience), and conduct inter-regional comparisons to have a more comprehensive. Thirdly, while the total sample size (n = 320) is sufficient for the aggregate model, the sample sizes of specific subgroups (e.g., Male: n = 134; Non-local tourists: n = 102) were not large enough to conduct a robust Multi-Group Analysis (MGA). Therefore, this study could not comprehensively compare differences between demographic segments. Future research should aim for larger, stratified samples to explore these potential moderating effects.