Introduction

Rural tourism, known for its natural beauty, cultural richness, and low population density, has become an important dimension in global tourism discussions. It offers tourists a chance to experience both natural landscapes and indigenous cultures (Lu et al. 2017). While rural tourism has been a long-standing attraction in countries like France, Italy, and Spain (Bel et al. 2015; Campón-Cerro et al. 2017; Galluzzo 2022), it has gained significant momentum in China as well. Since the implementation of the Rural Revitalization Strategy in 2017, the Chinese government has actively promoted it as a means to stimulate local economies, preserve heritage, and bridge the rural-urban divide. Official statistics indicate that, in 2019 (pre-pandemic), rural tourism in China recorded approximately 3.09 billion visits—over half of all domestic trips—and generated about RMB 1.81 trillion in revenue (Ministry of Culture and Tourism of China, https://www.gov.cn/xinwen/2020-09/13/content_5543050.htm, accessed 26 August 2025). This scale highlights both the sector’s market significance and the need to understand evolving tourist behavior. At the same time, the pandemic period accelerated shifts toward digital reliance, heightened health and safety considerations, and sustainability values (Robina-Ramírez et al. 2023; Zhong et al. 2024; Liu et al. 2024), trends that are particularly relevant for rural destinations and inform our subsequent analysis. Recent studies suggest that tourists, especially younger cohorts, increasingly rely on digital tools, social media, and smart technologies when planning rural getaways (Torabi et al. 2023; Maltese and Zamparini 2023).

The sustainability and high-quality development of this sector are closely linked to reliable land planning, which influences infrastructure location and characteristics (Nunkoo and Smith 2014; Saarinen et al. 2017; Var and Gunn 2020). Innovation and product differentiation may be realized through thoughtful land resource allocation, thereby enabling a broader range of activities (Pechlaner 2006; Volo 2006; Chen and Chen 2012; Li 2021; Pásková et al. 2021). Infrastructure that is well-suited to its spatial context may better align with visitor expectations, serving both existing and emerging markets (Benur and Bramwell 2015; Gardiner and Scott, 2018; Cooper et al., 2019). From the user perspective, factors such as accommodations, dining, leisure, shopping, and entertainment evoke distinct emotional responses and psychological attachments (Cho et al. 2014). Jiang et al. (2023) found that well-designed rural environments can enhance tourist interest and activity levels. Similarly, Qi et al. (2022) argue that understanding these psychological cues is crucial for managing services and land use decisions.

Prior research has often examined how particular attributes—such as local food, culture, and outdoor recreation—affect destination choice. For instance, An and Alarcón (2021b) emphasize the importance of staff hospitality, outdoor activities, additional facilities, and location. Likewise, Vujko et al. (2017) and Liang et al. (2021) highlight the role of food and outdoor activities in improving visitors’ satisfaction and encouraging return visits. Other studies (e.g., Kim 2015; Chaminuka et al. 2012; Albaladejo and Díaz 2021; Wang et al. 2024) explore tourist behavior from different disciplinary perspectives. Nonetheless, fewer works have investigated how these features interact with demographic and motivational factors, or how they vary across consumer groups. Although some segmentation studies exist (Mesić et al. 2022; Li et al. 2023; An and Alarcón 2021a; Lwoga and Maturo 2020; Eusébio et al. 2017), they leave scope for further development in translating such insights into market strategies.

China’s rural tourism boom reflects global patterns, yet poses unique challenges due to its vast geography, cultural diversity, and socio-economic variation. As it increasingly contributes to local economies and heritage protection, understanding what drives demand becomes more important. While global literature is abundant, research tailored to China remains relatively sparse—especially on how local factors shape travel decisions. Li et al. (2023) examined urban residents’ willingness to pay, and Douglas et al. (2024) emphasized cultural identity. However, these studies fall short in addressing broader consumer trade-offs—such as balancing tradition with modern amenities or integrating sustainability. A more comprehensive and behaviorally informed approach is warranted in this context. This also includes paying closer attention to digital behaviors and post-pandemic risk sensitivities (Torabi et al. 2023; Zhong et al. 2024), all of which now shape rural travel decisions in subtle but meaningful ways.

This study seeks to address these gaps by employing a choice experiment to investigate how tourists in China evaluate different tourism attributes. The method provides an established framework for capturing attribute-level trade-offs. We utilize both the Mixed Logit Model (MLM) and the Latent Class Model (LCM) to uncover heterogeneity. The MLM estimates average preferences while accounting for variation across individuals; the LCM segments tourists into behaviorally distinct groups—offering both a population-level and segment-specific analysis of rural tourism demand.

The contributions of this study are fourfold. First, it adds to the empirical understanding of Chinese rural tourism by applying a choice modeling approach. Second, it reveals nuanced behaviors within subgroups, enabling more tailored product development. Third, it offers indicative guidance on how to optimize rural resource use and support local planning. Finally, the findings can inform stakeholder strategies to promote sustainability, cultural retention, and inclusive development.

Materials and methods

Choice experiment design

To investigate the key factors influencing rural tourism preferences, this study adopts a choice experiment approach. The selection of relevant attributes and their corresponding levels followed a structured, multi-stage procedure to ensure conceptual clarity and contextual appropriateness. First, a comprehensive review of rural tourism literature and policy documents helped identify frequently cited and policy-relevant features (e.g., Li et al. 2023; Vujko et al. 2017; Albaladejo and Díaz 2021). Second, semi-structured online interviews were conducted with stakeholders connected to representative rural destinations across Guangdong Province, involving tourism officials, industry practitioners, and rural tourists to validate and refine attribute relevance. Third, we conducted a pilot survey with 30 participants to test attribute comprehension, level clarity, and overall response reliability. This iterative process ensured both scientific validity and practical resonance in the final choice experiment design.

Based on this process, we selected eight core attributes for rural tourism products: distance, natural scenery, local cuisine, cultural experiences, outdoor activities, accommodations (hotels & homestays), hospitality, and price (see Table 1). Natural scenery consistently emerged as a top priority for tourists, especially for urban residents seeking relief from industrialization and urbanization in tranquil rural landscapes (Lane 1994; Li et al. 2023). Homestay tourism, highlighted by Dai (2020), also plays a crucial role in supporting rural economies. Price levels (200, 400, 600 RMB per person per day) were set to span commonly observed ranges in the pre-test and policy documents, facilitating identification of price sensitivity (approximately USD 30–60 at end-2022 exchange rates, for reader convenience). Other common rural tourism features, such as local cuisine, cultural experiences, outdoor activities, and hospitality, were also included in the analysis. We categorized distance, natural scenery, and price into three levels, while the remaining attributes were divided into two levels.

Table 1 Rural tourism attributes and level design.
Full size table

The combination of these attributes and levels resulted in 864 possible product profiles. To achieve estimation efficiency under a large design space, we used Ngene 1.2.1 and implemented a D-optimal fractional factorial design, reducing the design to 36 choice sets suitable for estimating preferences. Respondents were divided into six groups based on birth month, with each group completing six choice sets. Each task presented three alternatives: Destination 1, Destination 2, or a “Neither” (opt-out) option. Figure 1 illustrates a sample choice set.

Fig. 1
figure 1

Presents a representative choice set.

Full size image

The inclusion of a “Neither” option allowed respondents to opt out of unrealistic or unappealing bundles, thereby improving the ecological validity of the experiment and mitigating forced-choice bias. This feature also allows the models to capture baseline participation/rejection tendencies (via the opt-out constant).

Survey design and implementation

Selection of research location and rationale

The decision to select Guangdong Province as the research site was based on multiple considerations and a clear rationale.

Firstly, located along the southeastern coast of China, Guangdong is at the forefront of the country’s reform and opening-up initiatives. In 2021, the province reported a GDP of 12.44 trillion yuan, ranking first nationally for 33 consecutive years (https://www.thepaper.cn/newsDetail_forward_16392210#/_blank, accessed on 26 August 2025). This economic scale indicates a high level of market dynamism and a broad consumer base for rural tourism.

Secondly, Guangdong has diverse natural landscapes and a rich cultural and historical heritage. For instance, elements of Lingnan culture offer a wealth of cultural resources and tourism products that support the development of rural tourism. The province’s geographical characteristics and cultural context, combined with its strategic position within the Guangdong-Hong Kong-Macao Greater Bay Area, contribute to its tourism appeal and market potential. Initially focused primarily on the Pearl River Delta region, rural tourism in Guangdong has expanded to include extensive development in the eastern and northern areas of the province, gradually forming a multifaceted product system (Yuan and Kong 2020).

Furthermore, Guangdong’s well-developed tourism infrastructure and service industry provide supportive conditions for promoting and developing rural tourism. In recent years, the sector has grown rapidly, with wider geographic coverage and increasingly diverse offerings, evolving from traditional sightseeing agriculture to leisure vacations and cultural experiences. Available market information and our pre-survey testing suggested that residents of Guangdong have notable awareness of and demand for rural tourism, along with a strong willingness to spend on tourism-related activities.

Accordingly, Guangdong serves as a relevant and informative setting for this study. To ensure regional breadth, the survey included respondents from all four major subregions of Guangdong—namely the Pearl River Delta, Northern Guangdong, Eastern Guangdong, and Western Guangdong. Sampled cities with relatively higher response volumes included Guangzhou, Shaoguan, Shantou, and Zhanjiang. While the sampling was not strictly stratified by region, this distribution reflects substantial variation in economic development, cultural context, and tourism infrastructure across the province, contributing to the contextual relevance of the findings.

Questionnaire design and sample collection

We used a questionnaire as the primary data collection tool, administered via an online survey platform (Wenjuanxing) to recruit respondents across Guangdong Province, and we also invited additional participants via personal networks to broaden participation. A preliminary (pilot) survey was conducted in multiple cities across Guangdong in January–February 2022 to refine both the questionnaire and the choice experiment design. Feedback from this phase helped fine-tune the research instrument. During the pre-test phase, some respondents reported that certain questions were difficult to understand or ambiguous. In response, we simplified and clarified the questionnaire wording to improve readability and survey flow. After these revisions, the final version was fielded on 22–24 April 2022.

To improve response quality and motivation, all participants received modest incentives upon completion—Wenjuanxing platform respondents were compensated through the platform’s built-in reward system, while personally invited participants received small cash rewards (typically 3–5 RMB). Incentive amounts were modest and not expected to create undue influence.

The revised questionnaire consisted of four main sections: First, it surveyed the respondents’ travel habits to understand their general behavior patterns. Second, it focused on their attitudes and expectations regarding tourism. Third, it employed a choice experiment to capture preferences for various rural tourism attributes. Finally, the questionnaire collected demographic information such as age, gender, and income. To minimize potential social desirability bias, participants were assured of anonymity and informed that there were no right or wrong answers and that their responses would remain confidential. Additionally, all attitudinal questions were carefully phrased in a neutral and non-leading manner to avoid priming effects.

In total, we collected 394 responses. To ensure data integrity and high quality, this study adopted completeness and quality of information as screening criteria, excluding invalid questionnaires that were missing key information or contained logical inconsistencies. We also excluded patterned responses (e.g., straight-lining) and other clear signs of inattention. After screening, 354 responses met the validation criteria, resulting in an effective response rate of 89.85%. These responses provided an analytic dataset of 6372 observations for empirical analysis.

While the sampling process aimed to reach a broad demographic, the approach—based on Wenjuanxing and supplemented by personal networks—may still favor younger, urban, and digitally literate respondents. However, this aligns with the emerging profile of rural tourism consumers in China, who tend to be mobile, educated, and actively engaged in leisure travel planning. The final sample provides a reasonable basis for capturing key preference structures within this target population.

Measurement models

This study employs an empirical framework grounded in Random Utility Theory (RUT) to investigate rural tourists’ decision-making processes. To capture preference heterogeneity, we utilize three models in sequence: the CLM, the MLM, and the LCM. These models allow for a progressively detailed understanding of how tourists’ preferences vary.

We begin with the CLM to establish a baseline understanding of how key rural tourism attributes—such as price, distance, and cultural experiences—influence decision-making. The utility Unit derived by tourist n from choosing option i in scenario t is expressed as:

$${{{U}}}_{{{nit}}}={{{\beta }}{{X}}}_{{{nit}}}+{\in }_{{{nit}}}$$
(1)

where Xnit is the vector of observed attributes, β is the vector of fixed coefficients, and ϵnit represents unobserved factors. The CLM assumes that all tourists have the same preferences and that the error term follows an independent and identically distributed extreme value distribution. However, this homogeneity assumption is restrictive and may not reflect real-world variation.

Building upon Eq. (1), the MLM relaxes the assumption of fixed coefficients by allowing them to differ randomly across individuals, thus capturing unobserved preference heterogeneity. Specifically, in the MLM, the coefficients \({\beta }_{n}\) vary according to a specified distribution (e.g., normal or log-normal), thereby capturing individual-level heterogeneity:

$${{{U}}}_{{{nit}}}={{{\beta }}}_{{{n}}}{{{X}}}_{{{nit}}}+{\in }_{{{nit}}}$$
(2)

The MLM was estimated by simulated maximum likelihood with a panel specification. Each respondent answered six choice tasks, forming a panel structure. To account for within-respondent correlation from repeated choices, we treated the data as a panel during estimation, so that uncertainty estimates reflect this dependence. The MLM thereby allows a richer description of preference variation across the population.

To further explore the structure and sources of this heterogeneity, we applied the LCM. The LCM segments the population into distinct latent classes, each characterized by homogeneous preferences within the class but with variation across classes. The probability of tourist n belonging to class s is modeled as:

$${{{P}}}_{{{ns}}}=\frac{exp ({{{\lambda }}}_{{{s}}}{{{Z}}}_{{{n}}})}{{\sum }_{{{k}}}exp ({{{\lambda }}}_{{{k}}}{{{Z}}}_{{{n}}})}$$
(3)

where Zn includes demographic and psychographic variables (e.g., age, income, environmental awareness), and \({\lambda }_{s}\) represents class-membership parameters. The overall probability of choosing an option is then a weighted sum of class-specific choice probabilities:

$${{{P}}}_{{{nit}}}=\mathop{\sum }\limits_{{{s}}}{{{P}}}_{{{ns}}}\,\cdot\,\frac{exp ({{{\beta }}}_{{{s}}}{{{X}}}_{{{nit}}})}{{\sum }_{{{j}}\in {{C}}}exp ({{{\beta }}}_{{{s}}}{{{X}}}_{{{njt}}})}$$
(4)

By using the LCM, we can identify distinct segments within the tourist population and provide practical insights for designing tailored rural tourism strategies. The model is estimated using the Expectation-Maximization (EM) algorithm, which is commonly used in latent class modeling due to its efficiency in handling incomplete data and unobserved heterogeneity. Although alternative methods such as Maximum Likelihood Estimation (MLE) are also applicable, EM is generally preferred in this context for its convergence stability and computational advantages. The optimal number of classes was determined by criteria such as the Bayesian Information Criterion (BIC) and Consistent Akaike Information Criterion (CAIC). These criteria are widely used in latent class modeling because they impose stronger penalties for model complexity compared to the Akaike Information Criterion (AIC), thereby reducing the risk of overfitting and favoring more parsimonious solutions.

Together, the CLM, MLM, and LCM form a comprehensive analytical framework for examining rural tourists’ preferences. The CLM provides an initial, homogeneous view of preferences, the MLM accommodates individual-level heterogeneity, and the LCM identifies distinct preference-based segments. This suite of models offers a structured basis for interpreting behavior and for deriving segment-specific managerial implications.

All statistical analyses were conducted using Stata 17.0, whose advanced simulation routines and panel specification features supported robust estimation of both the MLM and LCM models.

Empirical results

Descriptive statistics

From the sample data in Table 2, 57.63% of respondents are female and 42.37% are male. The average age is 35.81 years. Notably, 64.69% have completed undergraduate education or higher. Monthly income predominantly falls within the 5000–10,000 RMB range (41.81%), placing the modal income group in this bracket. Additionally, 75.71% of respondents have children. This demographic feature is pertinent to family travel contexts commonly discussed in rural tourism. Such offerings often include outdoor excursions, opportunities to experience nature, and interactions with local communities. The majority of respondents (83.33%) prefer self-guided tours, with a significant portion (76.55%) reporting rural experience during childhood (ages 0–12). Moreover, the sample is predominantly composed of young and middle-aged adults, with limited representation from older populations such as retirees. This age profile is consistent with a digitally aware, economically active traveler base, but it limits generalizability to older populations.

Table 2 Respondent characteristics.
Full size table

Further exploration of respondents’ environmental awareness, perceptions of rural tourism, and views on government rural tourism initiatives is presented in Table 3. We measured these attitudes using multi-item Likert scales (1–5), and all constructs demonstrated strong internal consistency (Cronbach’s α > 0.7). Respondents reported high environmental awareness and positive attitudes toward rural tourism’s potential benefits, such as job creation and income diversification for villagers. However, government promotional efforts received comparatively lower ratings, particularly regarding the comprehensiveness and frequency of campaigns, suggesting potential areas for improvement.

Table 3 Respondents’ environmental awareness, rural tourism cognition, and views on government promotion.
Full size table

Consumer appeal of rural tourism attributes

Table 4 presents the regression results from both the CLM and the MLM, offering insights into respondents’ preferences for various rural tourism attributes. The two models exhibit broadly similar outcomes, though the MLM offers a more nuanced understanding by accounting for individual-level heterogeneity.

Table 4 Regression results highlighting tourist preferences for rural tourism attributes.
Full size table

Starting with the results from the CLM, tourists show strong preferences for key rural tourism attributes. Shorter driving distances, attractive natural landscapes, local cuisine, cultural experiences, and hospitality are all significant factors influencing decision-making. Specifically, respondents favor excellent natural scenery (β = 0.630), local cuisine (β = 0.509), and immersive cultural experiences (β = 0.457). However, at 4 h the preference for the destinations declines significantly (β = –0.273), indicating lower attractiveness at this distance level.

Next, the MLM builds on the CLM by allowing for individual-level variation in preferences. The MLM results confirm the robustness of the CLM findings but also highlight significant heterogeneity among tourists. For instance, natural scenery remains a key driver of preference, with both good (β = 0.676) and excellent (β = 0.766) scenery receiving strong positive coefficients. Local cuisine (β = 0.609) and cultural experiences (β = 0.569) continue to be highly valued. Furthermore, the standard deviations for several attributes—especially natural scenery, local cuisine, outdoor activities, and accommodations—are statistically significant at the 1% level, confirming the existence of preference heterogeneity across respondents. Additionally, the MLM indicates stronger sensitivity to long-distance travel (β = –0.390). Hospitality (β = 0.266) and the presence of locally styled rural accommodations (β = 0.488) are also positive and statistically significant.

Importantly, the results confirm that respondents consider multiple attributes simultaneously rather than focusing on a single feature. The combination of attributes—for example, good scenery together with cultural activities and welcoming service—is associated with higher stated utility within the choice tasks. Overall, the CLM provides a baseline understanding of tourist preferences, while the MLM allows for deeper exploration into preference heterogeneity, confirming that rural tourism preferences are not uniform across all tourists. These findings justify the use of the MLM to better capture the nuanced ways in which different attributes influence individual tourists’ decisions.

Latent class analysis of tourists

To segment rural tourists based on their preference structures, we applied a Latent Class Model (LCM) using individual-level responses from the choice experiment. The LCM identifies unobserved heterogeneity by analyzing how respondents evaluated different combinations of rural tourism attributes across choice sets. Each individual’s selection behavior—for example, consistently favoring destinations with excellent natural scenery or with local cuisine—was used to estimate class-specific utility coefficients. The model assumes that individuals with similar patterns of choice behavior are probabilistically grouped into the same segment.

Furthermore, class-membership probabilities were modeled as a function of socio-demographic characteristics such as age, education, income, and family structure. This joint modeling approach enables the segmentation to reflect not only preference-based differences but also demographic tendencies associated with class membership.

To determine the optimal number of classes, we compared the BIC and CAIC values. As shown in Table 5, both criteria achieve their lowest values with a three-class solution, leading to three distinct segments.

Table 5 Basis for latent class classification.
Full size table

The classification output provides not only probabilistic groupings of tourists but also distinct preference structures across classes. Based on the LCM results, respondents were categorized into three latent classes, each reflecting a unique pattern of sensitivity to rural tourism attributes. These segment-specific estimates are presented in Table 6, alongside the effects of socio-demographic covariates on class membership. Segment labels are descriptive mnemonics based on the sign and magnitude of class-specific attribute coefficients (and the opt-out constant) and are used for ease of reference rather than definitive typologies.

Table 6 Latent class model’s estimated results.
Full size table

Senior Nature Seekers (Class 1)

This class, demographically older, on average, exhibits strong positive coefficients for natural scenery (β = 0.948–1.003) and a negative coefficient at the longer distance level (4 h: β = –0.743). Price is also negative and statistically significant (β = –0.002). Other attributes (e.g., local cuisine, cultural experiences, outdoor activities, hotels & homestays, hospitality) are smaller in magnitude or not statistically significant for this group. For the class-membership covariates, higher education is negatively associated with membership (β = –1.043), and higher ratings of government promotion are negatively associated with membership (β = –1.012). The scenery-dominant pattern with sensitivity to distance and price motivates the label.

Young Quality Explorers (Class 2)

This segment is characterized by pronounced price sensitivity (β = –0.003), with large positive and significant coefficients for natural beauty (β = 1.117–1.470), local cuisine (β = 1.293), outdoor activities (β = 1.227), hotels & homestays (β = 0.918), and hospitality (β = 0.664). The opt-out constant is positive (β = 2.417), indicating a higher baseline propensity to reject offered bundles that do not meet expectations. On the class-membership side, older age (β = –1.978) and having children (β = –0.801) are associated with lower odds of membership; higher environmental awareness shows a similar negative association (β = –1.306). The multi-attribute, quality-attentive pattern with screening behavior motivates the label.

Broad-acceptance Experience Pursuers (Class 3)

This category shows positive coefficients across a broad set of attributes—natural scenery (β = 0.253–0.464), local cuisine (β = 0.268), cultural experiences (β = 0.322), hotels & homestays (β = 0.234), and hospitality (β = 0.124)—with lower price sensitivity (β = –0.001). The opt-out constant is negative (β = –2.427), consistent with a higher overall acceptance of offered bundles. The label reflects this broad, multi-attribute acceptance profile.

To further profile the latent classes, Table 7 reports within-segment percentages for socio-demographic and attitudinal indicators (gender, age, education, income, children, environmental awareness, rural tourism cognition, and evaluation of government promotion), together with Dunn’s pairwise tests.

Table 7 Descriptive statistics of three tourist segments and Dunn’s pairwise tests.
Full size table

Senior Nature Seekers (12.43%)

Higher share in the high age category (88.64%) and female (61.36%). The share in higher education (27.27%) is lower than in the other segments, with significant pairwise differences. Income shows a lower high share (34.09%), but pairwise differences are not significant across segments. Environmental awareness is high (High = 63.64%): it differs significantly from Class 2, but not from Class 3. Government promotion ratings are the lowest (High = 20.45%), significantly lower than Class 2 and Class 3. Rural tourism cognition (High = 68.18%) shows no significant pairwise differences.

Young Quality Explorers (35.03%)

Predominantly low age (91.94%), with significant differences relative to both other segments. The share with higher education is the highest (81.45%), with significant pairwise differences. Without-children is more prevalent (44.35%), significantly higher than Class 1 and Class 3. Environmental awareness is lower (High = 32.26%), significantly lower than Class 1 and Class 3. For government promotion, the high share (42.74%) is higher than Class 1, while the difference vs. Class 3 is not significant. Income distributions are broadly comparable (no significant pairwise differences). Rural tourism cognition (High = 58.06%) shows no significant pairwise differences.

Broad-acceptance Experience Pursuers (52.54%)

A balanced age profile; the with-children share is high (86.02%) and significantly higher than Class 2, while the difference vs. Class 1 is not significant. Environmental awareness (High = 61.29%) is significantly higher than Class 2, and not different from Class 1. Government promotion (High = 46.77%) is significantly higher than Class 1 and not different from Class 2. Income distributions are similar across segments (no significant pairwise differences). Rural tourism cognition (High = 61.83%) shows no significant pairwise differences.

These descriptive profiles complement the LCM preference patterns by indicating which socio-demographic and attitudinal traits are more prevalent in each segment, while highlighting where differences are—and are not—statistically significant.

Theme preferences of different tourist groups for rural tourism

Building upon the segmentation analysis in the previous section, this part explores the specific theme preferences of the three identified tourist groups: Senior Nature Seekers, Young Quality Explorers, and Broad-acceptance Experience Pursuers. Our exploration of rural tourism primarily encompasses eight themes:

  • Farm experiences (e.g., farms, fruit and vegetable picking gardens);

  • Local cuisine (e.g., farm meals, signature cuisine villages);

  • Ecological experiences (e.g., ecological and pastoral landscapes);

  • Cultural experiences (e.g., historic villages and towns, cultural museums, ancient post routes);

  • Sports and cultural activities (e.g., greenways, rafting, outdoor hubs);

  • Health and wellness (e.g., resorts, hot springs, recuperation spots);

  • Science and educational outreach (e.g., flora and fauna education, labor training);

  • Patriotic education (e.g., revolutionary memorials, monuments, foundational bases).

These themes were developed through a mixed-method approach combining prior literature review (e.g., Li et al. 2023; Fichter and Román 2023; Douglas et al. 2024; Pahari 2024), stakeholder interviews during fieldwork in Guangdong, and feedback from pre-survey testing. This triangulated process ensured that the final list reflects both theoretical grounding and practical relevance in the Chinese rural tourism context. Understanding how different tourist groups prioritize these themes is useful for designing products that align with diverse market segments.

To facilitate visual comparison across segments, we grouped selection rates into three interpretive tiers: A (60–80%), B (40–60%), and C (20–40%). Thresholds were chosen based on the empirical distribution in our data to provide equal-width bins for the radar charts. These tiers are interpretive aids for visualization only and are not statistical tests.

Figure 2a–c shows descriptive differences in the within-segment ordering of theme selection. For Senior Nature Seekers (see Fig. 2a), ecological experiences and local cuisine have comparatively high selection rates within the segment (72.73% and 63.64%, respectively), while activity- and education-oriented themes are lower—particularly sports/cultural activities (29.55%) and science and educational outreach (34.09%). This pattern is descriptively consistent with a more contemplative, scenery- and food-oriented style of rural visitation in this class.

Fig. 2: Presents the theme preferences of three types of rural tourists.
figure 2

a Theme preferences of Senior Nature Seekers for rural tourism. b Theme preferences of Young Quality Explorers for rural tourism. c Theme preferences of Broad-acceptance Experience Pursuers for rural tourism.

Full size image

For Young Quality Explorers, Fig. 2b indicates relatively high selection for local cuisine, farm experiences, ecological excursions, health and wellness, and cultural immersions. By contrast, science and educational outreach (33.87%) and patriotic education (34.68%) register at the lower end of the within-segment distribution. This ordering is descriptively in line with the class’s multi-attribute, quality-attentive profile identified in the LCM.

For Broad-acceptance Experience Pursuers, Fig. 2c suggests a balanced pattern: ecological activities and local cuisine rank relatively high, with farm and cultural experiences also prominent, while education- and patriotism-related themes are moderate within the segment. Overall, the selection ordering points to broad-based preferences rather than a focus on a single theme.

In summary, across segments, ecological scenery and local cuisine tend to sit toward the higher end of the within-segment distributions, whereas health/wellness, patriotic education, and science/educational outreach show segment-specific variation in relative priority. These distinctions underscore the importance of tailoring offerings to segment-specific expectations while maintaining attention to widely valued core themes.

Discussion and implications

Discussion

Rural tourism has gained significant momentum, becoming an important segment of the global tourism industry. The unique combination of cultural engagement and natural beauty has increasingly attracted tourists seeking more than just passive sightseeing. The findings of this study are consistent with these trends, revealing a pattern of preferences in our sample for experiences that blend ecological preservation with cultural authenticity. This aligns with existing research by Park and Yoon (2009), who observed that Korean tourists value a combination of agricultural and cultural activities, and Albaladejo and Díaz-Delfa (2009), who highlighted Spanish tourists’ interest in rural accommodations. These results suggest that, in prior studies across different contexts, the core appeal of rural tourism appears to be its authenticity and immersive nature. Recent literature also suggests that post-pandemic tourists may increasingly seek experiences emphasizing well-being, safety, and digital accessibility—dimensions particularly relevant to rural destinations (Robina-Ramírez et al. 2023; Zhong et al. 2024).

An important aspect of this study is the segmentation of rural tourists into three distinct categories: Senior Nature Seekers, Young Quality Explorers, and Broad-acceptance Experience Pursuers. Using the LCM, this study offers additional insights into the heterogeneity of tourist preferences, contributing to a deeper understanding of how different segments prioritize attributes such as natural scenery, local cuisine, and cultural experiences. Previous studies, such as Eusébio et al. (2017) and Mesić et al. (2022), have recognized the diversity of rural tourists, and this study complements those findings by offering an empirical breakdown of preference patterns in a stated-preference sample from Guangdong. Furthermore, these segments are broadly consistent with diversification trends identified in recent rural tourism literature, where younger groups have been reported to demand value-added services such as wellness tourism, digital information access, and integrated cultural storytelling (Douglas et al. 2024; Liu et al. 2024).

Additionally, our analysis of theme preferences indicates that different tourist groups are drawn to distinct rural tourism themes. For instance, Senior Nature Seekers show comparatively higher within-segment selection for ecological experiences and local cuisine, while Young Quality Explorers are relatively more attracted to activities involving adventure, farm experiences, and cultural immersions. Broad-acceptance Experience Pursuers display a balanced interest in nature, culture, and family-compatible activities. These findings suggest value in tailoring rural tourism offerings that cater to the diverse expectations of different tourist segments.

One notable finding is the consistently higher emphasis on natural scenery and local cuisine across segments in our sample. Senior Nature Seekers, for instance, prioritize scenic beauty, while Young Quality Explorers value both scenic and culinary experiences. This suggests value in rural tourism offerings that emphasize these core attributes, as noted in the works of Kim (2015) and Vujko et al. (2017), who emphasized the importance of local gastronomy and outdoor experiences in shaping tourist satisfaction. Additionally, the Broad-acceptance Experience Pursuers’ interest in a balance of ecological, cultural, and family-compatible activities is consistent with demand for multifaceted rural tourism experiences. These observations accord with the broader tourism literature, which highlights the importance of providing tailored experiences that meet both emotional and practical needs (An and Alarcón 2021a). More recent studies reaffirm these insights and further suggest a shift in preferences toward health-conscious, digital, and sustainability-linked rural experiences. For example, Torabi et al. (2023) and Maltese and Zamparini (2023) show how smart tourism technologies and green transport preferences influence travel intention and destination image among rural visitors.

The role of government promotion also appears as a differentiating factor in our results: the Senior Nature Seekers segment reports comparatively lower ratings of government promotion than the other segments in this sample. Taken at face value, this pattern suggests that the salience of policy-driven initiatives may vary across segments and contexts rather than being uniformly decisive. While Liu et al. (2023) report positive effects of government interventions in some settings, our evidence is correlational and context-specific and should not be read as a direct test of promotional effectiveness. For younger and more digitally connected consumers, prior studies indicate that the influence of government promotion may be secondary to digital channels—such as online reviews, influencer recommendations, and social media campaigns—which we did not measure directly in this study (Torabi et al., 2023; Rodrigues et al., 2023). Accordingly, any practical implication is provisional: destination managers could consider multichannel approaches and evaluate them locally rather than treating one channel as universally superior.

This study documents the heterogeneity in tourist preferences through both the MLM and LCM, offering a more nuanced understanding of the differences in how tourists perceive rural tourism. The findings indicate that tourists have diverse motivations, which align with Gao et al. (2019), who stressed the significance of preserving rural authenticity. By focusing on specific tourist segments, rural tourism stakeholders may better align their offerings with market expectations within the scope of this study’s setting and design.

However, several limitations of this study must be acknowledged.

First, the study relies on a non-probability, online sample from Guangdong Province (via an internet survey and personal networks). While Guangdong is economically dynamic, this sampling frame—together with the single-province scope—may not capture the full diversity of rural tourism behavior across China due to variations in geography, cultural practices, and tourism maturity levels across regions. Expanding geographic coverage in future research could enhance the external validity of the findings.

Second, although the sample was drawn from a diverse population, it skews younger and more digitally literate due to the online distribution method. While this aligns with the profile of emerging rural tourism consumers, older adults—particularly retirees—were under-represented. Self-selection into an online survey also cannot be ruled out, which may limit population representativeness. Future studies could adopt stratified sampling or targeted outreach to improve demographic balance.

Third, as with most self-reported surveys, social desirability bias may influence responses—particularly for attitudinal items such as environmental concern or institutional trust. We sought to mitigate this through anonymous responses and neutral wording, though some residual bias may remain.

Fourth, while the discrete choice experiment enables respondents to evaluate multiple attributes simultaneously, it cannot fully mirror the complexity of real-world decision-making. Moreover, the current model focuses on main effects and does not incorporate interaction terms—such as the potential moderating effect of scenery quality on the perceived value of accommodations. In addition, the number of tasks per respondent, while modest, may still introduce some simplification relative to in situ choices. Future studies could improve behavioral realism by integrating observational methods and exploring interaction effects. In the segment-profile comparisons, Dunn’s post-hoc tests were reported without multiple-comparison adjustment; pairwise inferences should be interpreted cautiously.

Finally, although this study incorporates recent literature on post-pandemic tourism dynamics, the data were collected in April 2022 and do not directly model behavioral drivers such as risk perception (including perceived safety), digital engagement, or sustainability values. These emerging factors, along with the potential influence of gender on attribute-based preferences, present important avenues for future research. In this regard, future studies could also experiment with alternative ways of framing attribute information in choice sets—for instance, tailoring textual prompts to different tourist segments—to examine how communication strategies affect stated preferences. Seasonal timing may also matter; future research might explore how preferences shift during peak periods such as national holidays or summer vacations. Taken together, these limitations underscore the importance of cautious interpretation and contextualization of the findings, especially when informing national-level policy or broad generalizations.

Implications

The success of rural tourism is often linked to broader policy objectives, such as cultural preservation, environmental stewardship, and rural revitalization. In light of our Guangdong sample, this study’s findings—particularly the segmentation results—may offer useful guidance for different stakeholders, including government planners and tourism operators, in crafting differentiated strategies.

For government agencies, aligning tourism planning with China’s Rural Revitalization Strategy can be helpful. Investment in infrastructure could be tailored to support diverse tourist segments—for example, enhancing accessibility and short-distance transportation options for Senior Nature Seekers, or developing cultural heritage sites and outdoor recreation zones to align with interests observed for Broad-acceptance Experience Pursuers and Young Quality Explorers. Local governments might also consider policy support and subsidies for eco-friendly accommodations, digital marketing training, and cultural experience programming. Strengthening community engagement and protecting rural identity—through initiatives such as festival preservation, traditional crafts, or local storytelling—may enhance both authenticity and sustainability.

For tourism enterprises, customized product offerings and refined marketing approaches are likely to be effective. Young Quality Explorers, for instance, appear to value immersive, high-value experiences that blend local cuisine, cultural depth, and hands-on activities. Tour operators targeting this group could consider bundling farm visits with culinary workshops or heritage performances, promoted through digital media and influencer partnerships. Broad-acceptance Experience Pursuers show broad acceptance of multiple features; given this segment’s higher with-children share in our sample, family-friendly bundles may perform well for this group (e.g., parent–child eco-camps or multi-generational cultural routes). While Senior Nature Seekers account for a smaller portion of the sample, they may represent a valuable market for low-season tourism and wellness-oriented travel. Calm environments, scenic walking trails, and restorative accommodations may appeal to this group, especially when combined with nostalgic or memory-driven narratives.

Moreover, digital marketing can be an important lever for expanding rural tourism reach. Personalized content on social media platforms, virtual previews of destinations, and online feedback channels can enhance tourist engagement. Equipping rural businesses with digital tools and training may help tourism offerings remain visible and competitive in a rapidly digitizing marketplace.

Finally, building a sustainable and inclusive rural tourism system typically benefits from coordinated effort across sectors. Partnerships between government, businesses, and local communities can support investments in green infrastructure, circular economy initiatives, and responsible tourism education. By tailoring actions to segment patterns observed in this study—and recognizing that these implications are context-specific rather than universal—stakeholders may help rural tourism serve as a dynamic driver of sustainable prosperity.

Conclusions

Rural tourism is often discussed as an element of China’s rural revitalization and cultural preservation agenda. This study applied a discrete choice experiment (analyzed with CLM and MLM, with an LCM for segmentation) to explore how Chinese tourists evaluate rural tourism attributes. We identified three segments in our sample—Senior Nature Seekers, Young Quality Explorers, and Broad-acceptance Experience Pursuers—each with differential preferences, constraints, and behavioral tendencies. While all groups value ecological landscapes and local cuisine, their responses to price, distance, accommodations, hospitality, and opt-out propensity vary. These findings support the importance of differentiated product and marketing strategies and suggest that integrating psychographic segmentation may be helpful in rural tourism planning.

The study contributes to the literature by linking heterogeneity in tourism preferences to demographic and attitudinal factors within this Guangdong sample, and provides evidence-informed insights for stakeholder-specific strategy design. Policy implications could include tailoring digital marketing for younger groups, promoting eco-accessible options for older tourists, and integrating family-compatible educational content. Although the analysis is limited to Guangdong, the patterns may be informative for similar contexts; generalization should be made with caution. Interpretation should also consider the non-probability, online sample; the cross-sectional design; reliance on stated-preference data; and that theme rankings were descriptive rather than inferential. Future research could explore digital trace data, behavioral prompts (e.g., safety cues), and cross-regional comparisons to deepen understanding of rural tourism behavior in the post-pandemic era.