1. Introduction
Rural settlements are important places for rural population life and production and are a space phenomenon. The rural settlement space has been the research key of geographic studies concerning rural settlements [
1]. With progress in urbanization and industrialization, many rural settlements face or have entered rapid transformation stages [
2]. Rural transformation development is the reconstruction of rural settlement and mainly involves economic, social morphology, and spatial changes [
3,
4]. In the transformation process, the quality of many rural settlements has been improved, but ‘rural hollowing and planned chaos’ can coincide. These problems are particularly prominent for rural settlements in mountainous areas with complicated geological environments [
5]. In previous studies, Woods (2005) emphasized the reconstruction of socio-economic formations of rural regions resulting from changes to subjects in the transformation process [
6]. Li Hongbo et al. (2012) focused more on spatial transformation in rural spaces. They believed that the removal, decline, and disappearance of villages should also be considered reconstruction of rural settlements [
7]. Rural space reconstruction is an important manifestation of the reconstruction of rural settlements [
8], and the spatial reconstruction of rural settlements is the outlook of reconstructed spatial forms of rural settlement [
9]. In the present study, the spatial reconstruction of rural settlements was defined in a narrow sense. That is, to adapt to the development of urban and rural areas, the process of change in the spatial distribution of rural residences is caused by changes to peasant households, which are crucial subjects of rural settlement. Whether spatial pattern changes of rural settlements are reasonable directly determines whether the rural settlement can realize comprehensive, coordinated, and sustainable development. Therefore, promoting the suitable reconstruction of rural settlement spaces in mountainous areas can improve the quality of rural settlement [
10].
Exploring the driving forces for the spatial reconstruction of rural settlements is necessary and helps guarantee the effective spatial reconstruction of rural settlements. The spatial reconstruction of rural settlements is accomplished under the collaborative promotion of internal and external driving forces [
11]. These include place attachment, historical culture, environment fitness, housing condition and quality, natural disasters, economic levels, infrastructure, government policies, and peasants [
12,
13,
14,
15,
16,
17,
18,
19,
20,
21,
22]. The government has been the primary driver of rural reconstruction [
23,
24,
25]. Although such top-down planning is characteristic of high efficiency and rapid construction [
26,
27], it can ignore the intentions and needs of peasant households. Thus, the outcomes of the reconstruction are not ideal [
28]. As the subject of rural settlement, peasant households have the most direct and deepest needs relating to the spatial reconstruction of rural settlements. Reconstruction intention, family structure, policy cognition, risk perception, and the neighborhood of peasant households can influence reconstruction progress [
29,
30,
31,
32]. During the spatial reconstruction of rural settlements, it is necessary to combine local, practical situations [
33] and consider the subjective demands of peasant households [
34]. Previous studies on the driving forces for the spatial reconstruction of rural settlements have mainly concentrated on plain regions; however, there are few that have involved mountainous regions and hills [
35,
36,
37,
38]. The most common measurement models that have been used to study the driving forces for the spatial reconstruction of rural settlements are the Probit and Logit models [
29,
39,
40]. Recently, structural equation modeling (SEM) [
41] has also been applied to this area. The Probit and Logit models are used in traditional linear regression analysis. Linear regression analysis defines dependent and independent variables in the model, but it can only provide direct effects between variables and cannot show possible indirect effects. Unlike traditional regression analysis, the structural equation model can handle multiple dependent variables simultaneously and replace multiple regression, path analysis, factor analysis, covariance analysis, and other methods. This model can analyze the effect of individual indicators on the overall outcome and the relationship between individual indicators, which overcomes the limitations of the Probit and Logit models in being unable to explore internal relations among factors intuitively [
42]. To date, there have been few studies from the perspective of peasant households on the driving forces for the spatial reconstruction of rural settlements in mountainous areas with complicated geological environments based on the SEM model.
Rural development is the key to achieving the new goal of sustainable development. Nearly 45% of the global population lives in rural areas of developing countries that face issues such as hunger, poverty, and youth unemployment [
43]. Poverty eradication is the primary goal of the
Agenda for Sustainable Development in 2030. Countries worldwide are working towards this goal and trying not to leave anyone behind [
44]. Nevertheless, the number of residents in mountainous regions exposed to the risk of food shortage is increasing due to the worsening of mountainous environments, and the poverty problem in mountainous rural areas is particularly serious [
45]. As a global agricultural and population power, China still had about six million rural residents across millions of rural settlements in 2017—despite the continuous acceleration of urbanization since the 21st century began. Therefore, rural settlements are still a fundamental residential form for Chinese people [
46]. Mountains are extensively distributed in the continents of Eurasia and the Americas. China is also a mountainous country, with mountainous land accounting for nearly 70% of the total land area of China and the location of one-third of the population. The development of mountainous regions is related to poverty eradication for nearly 50% of China’s population. Therefore, the development trend of rural settlements in mountainous regions directly influences the national development situation [
47].
Against this background, Sichuan Province—a classical mountainous region in China characterized by poverty—was chosen as the research object. The driving forces for the spatial reconstruction of rural settlements in this mountainous area were explored by combining PRA (participatory rural appraisal) and SEM (structural equation modeling). The specific aims of this study were: (1) to combine information on the practical situations and intentions of peasant households in the study area to construct an SEM model about the factors influencing these households’ intentions relating to spatial reconstruction of rural settlements; (2) to identify the major driving forces of peasant households’ intentions relating to reconstruction; (3) to inform the further smooth reconstruction of rural settlement spaces in mountainous areas.
3. Results
3.1. Descriptive Statistical Analysis of Respondents
Among the 266 peasant households, 165 were willing to participate in the spatial reconstruction of rural settlements, accounting for 62.03%. The remaining 101 were unwilling to participate in the spatial reconstruction of rural settlements, accounting for 37.97%. This reflected that most peasant households had a strong inclination towards the spatial reconstruction of rural settlements for villages in the sampled regions (Xichang City, Miyi County, Puge County, and Yanyuan County). Among the 266 peasant households, most had an educational background of either illiteracy or primary school. Most respondents were in good physical condition. Males accounted for 61.28%, and the average age was 44.35 years. Minority groups accounted for 67.67%, with the highest proportion being the Yi ethnicity. Most peasant households (85.34%) were engaged in agricultural activities. Overall, the village populations were mainly middle-aged males from minority groups with low educational backgrounds who were engaged in agricultural production activities (
Table 2).
3.2. Validity and Reliability
First, reliability analysis was carried out on different layers of the questionnaire survey by using Cronbach’s alpha. An α value greater than or equal to 0.6 indicates acceptable reliability [
83]. It can be seen from
Table 3 that Cronbach’s α value for the general scale was 0.692, which indicates good reliability. Furthermore, Cronbach’s α values for the five scale layers were mainly ≥0.6, indicating that the questionnaire had some consistency and stability. The structural validity of the questionnaire was verified by factor analysis. The KMO (Kaiser–Meyer–Olkin) value for the general scale was 0.812, and the KMO values for the five layers of the scale were all higher than 0.5, and the
p-value was 0.000 < 0.01. The Bartlett test of sphericity indicated that factor analysis was applied to the survey data. It can be seen from
Table 4 that the principal factors that were screened from the factor analysis conformed to the theoretical structure entirely, and that the cumulative variance contribution rate was relatively high. This demonstrates that the questionnaire has good structural validity.
3.3. Fitting and Adaption of Models
The initial model (
Figure 2) was verified by survey data and revised according to correction indices. After three paths were added successively, the revised SEM paths were obtained (
Figure 4). The two paths e6 ↔ e7 and e1↔e20 passed the significance test and had positive values, indicating that transportation accessibility was positively correlated with the convenience of getting water, selection of living mode, and development opportunities. The new path e10 ↔ e18 passed the significance test and had a negative value. This indicates that a stronger community cultural atmosphere made peasant households easier to contact and accept different dietary cultures. As a result, their dietary habits were more readily changed.
During the fitting evaluation of an SEM model, a higher degree of fitting indicates that the model construction is more reasonable. In this study, the SEM of factors influencing peasant households’ reconstruction intentions was verified by confirmatory factor analysis. The results (
Table 5) demonstrate that among the absolute adaptation indices for the SEM model, χ
2/df = 1.592 (<3), RMR = 0.024 (<0.05), and RMSEA = 0.047 (<0.05). Furthermore, the SEM fitting indices GFI, NFI, RFI, IFI, TLI, and CFI were all greater than 0.90, and PGFI, PNFI, and PCFI in the simple adaptation index were all higher than 0.50. Therefore, all model indices conformed to the requirements, indicating the good fit of the model.
3.4. Recognition of Driving Forces
The non-standardized regression coefficient was calculated by the maximum likelihood method (
Table 6). The standard deviations and critical values of the probability of geological disasters, poor traffic conditions, farming culture, and place attachment are blank because these five factors were set as fixed parameters in the initial modeling.
(1) Among the external attractions, the standardized regression coefficients of good medical conditions, sufficient water and power supply, convenient information acquisition, and a suitable living environment were all higher than 0.960. This reflects that these factors had significantly positive effects on the reconstruction intention of peasant households. To seek better living conditions, peasant households were more willing to accept the spatial reconstruction of rural settlements. Among external attractions, the path coefficient of more development opportunities was fixed at 1, and the standardized regression coefficient was 0.448, indicating that development opportunities (e.g., employment and education) had positive impacts on the reconstruction intention of peasant households. However, development opportunities were not highly related compared with other external attraction factors.
(2) Among geological disasters, the path coefficient of the probability of geological disasters was fixed at 1, and the standardized regression coefficient was 0.774. The standardized regression coefficients of the frequency of geological disasters, the influence of geological disasters on crops, and economic loss caused by geological disasters were 0.678, 0.841, and 0.682, respectively. Of these, B3 (the influence of geological disasters on crops) was the top factor that influenced the reconstruction intention of peasant households, while the other three factors also had relatively significant positive impacts on the reconstruction intention of peasant households. Most peasant households in surveyed villages had experienced geological disasters. They were mainly engaged in agricultural production, with crops forming their primary income source and survival foundation. They believed that geological disasters could affect crop outputs and even their houses, greatly influencing their current residence.
(3) For internal impetus, the standardized regression coefficients of a weak community cultural atmosphere, poor communication network, power shortages, and water shortages were 0.438, 0.489, 0.845, and 0.357, respectively. The path coefficient of poor traffic conditions was fixed at 1, and the standardized regression coefficient was 0.341. All these factors significantly influenced the reconstruction intention of peasant households. Specifically, power shortage was the primary influencing factor, followed by a poor communication network. This reflects that the imperfect communication networks and electronic devices in current residences brought great inconvenience to the daily life of peasant households. Peasant households had a stronger reconstruction intention if they had lower satisfaction with their current residential environment. Furthermore, the community cultural atmosphere had a slightly positive influence on the spatial reconstruction of rural settlements. Water shortages and poor traffic conditions had positive effects on the reconstruction intention of peasant households; however, these effects were minimal.
(4) Among the production cohesion factors, the path coefficient of farming culture was fixed at 1, and the standardized regression coefficient was 0.980. The standardized regression coefficient of pasture culture was 0.903. This indicates that farming culture and pasture culture were significantly positively correlated with the reconstruction intention of peasant households. Peasants prefer places where they can plant crops and feed poultry, as they depend on the planting and breeding industries. When there was a more robust farming culture and pasture culture, peasant households were worried that there was not enough land for agricultural and poultry industries and had a stronger reconstruction intention.
(5) In the life cohesion factors, the path coefficient of place attachment was fixed at 1, and the standardized regression coefficient (0.350) was lower than for the other three factors. This indicates that place attachment was not highly correlated with the reconstruction intention of peasant households. The standardized regression coefficients of dietary habit, language, and living mode were 0.530, 0.471, and 0.404, respectively. Of these, language, living mode, and place attachment positively affected the reconstruction intention of peasant households; however, these effects were small. Dietary habits was the top influencing factor of the reconstruction intention of peasant households. This was because most peasant households are minorities with unique dietary habits and are highly unwilling to change their current staple food. Dietary habits had a considerable positive influence on the reconstruction intention of peasant households.
4. Discussion
This study combined participatory rural appraisals with a structural equation model. It analyzed the influence of people’s willingness on behavior choices from the perspective of peasant households to explore the driving force of behavior results. The SEM method is an important analysis tool in quantitative research. For conceptual indicators that are difficult to directly and accurately measure, such as psychology and society, the SEM model provides a method to account for measurement errors—using multiple indicators to reflect potential variables. This study method is more accurate and reasonable than the traditional regression methods and has many applications, such as psychology, management, and other related research. Scholars can design questionnaires according to their own research goals to study people’s behavioral motivations.
This study obtained conclusions consistent with those of Garcia (2009), Pritchard (2012), and Wierucka et al. (2021). That is, that external attractions are important drivers of the spatial reconstruction of rural settlements. Specifically, places with better infrastructure, including good medical conditions and sufficient water and power supply, are more attractive to peasant households and result in a stronger reconstruction intention of peasant households. Additionally, the living environment is an important factor that people consider in choosing a residence. Indeed, people impose higher and higher requirements on their quality of life with improved economic levels in western developed countries. A suitable living environment is a premise and basis for a high quality of life [
74,
84,
85]. Convenient information acquisition and more development opportunities help guarantee a high quality of life [
86,
87]. A different research conclusion of the present study compared to previous studies was that development opportunities (e.g., employment and education) had significantly positive effects on the reconstruction intention of peasant households, but these effects were not very significant. This may be related to implementing the rural revitalization strategic policy in the Panxi area in recent years. Increasing anti-poverty projects have been introduced in the Panxi area, which has provided more employment and education opportunities for local people. This may explain why local peasant households did not have a stronger intention to relocate to places with more development opportunities.
Many countries have researched the driving forces of place attachment in the spatial reconstruction of rural settlements. However, there have been few studies relating to place attachment in China. Place attachment is composed of local dependence (functional attachment) and local identity (emotional attachment) [
88]. Barcus (2010) and Malik and Yoshida (2020) found that place attachment could promote the spatial reconstruction of rural settlements. One of the important factors that affected peasant households’ intention to move to a better place was local attachment [
89,
90]. However, the present study found that although place attachment was one factor that peasant households considered in migration, it was not decisive. All respondents lived in mountainous areas in the Panxi area, where power shortages, geological diseases, and poor living conditions were common. As these places do not provide the ideal residential mode, peasant households had a low emotional and functional attachment to their current residential areas. It was interesting that life cohesion influenced peasant households’ intentions relating to the spatial reconstruction of rural settlements the least, which was different from the research results of Deumert (2005) and Sami (2013) [
91,
92]. This may have been because, with the development of the social economy, local peasant households had increasing contact with people in other places and gradually began to become familiar with and accept the cultures, diets, and living modes of other ethnicities. Therefore, life cohesion was not the primary factor that peasant households considered during the spatial reconstruction of rural settlements.
The unique geological environment in mountainous regions intensifies unfairness and differences in the development among rural settlements. Rural settlements in mountainous regions develop more slowly and with more difficulty than those on plains. Suppose the spatial reconstruction of rural settlements in mountainous areas is unreasonable. In that case, it will inevitably result in or intensify the worsening of the ecological environment and geological disasters, thus posing an increased threat to the life and property of residents. Previous studies have found that the intention of peasant households plays an essential role in the spatial reconstruction of rural settlements [
12,
40,
93]. Government-guided rural planning often fails if it does not respect the intention of peasant households and forces them to move [
28,
94,
95]. Hence, exploring the driving forces for the spatial reconstruction of rural settlements by considering the intentions of peasant households can lead to the implementation of rural planning and construction that are closer to the ideal living mode of peasant households. This would encourage peasant households to take the initiative in reconstruction, thus decreasing conflicts between them and the government.
This study has made the following novel contributions: (1) The Panxi area in Sichuan Province, a typical mountainous region in China, was chosen as the study area to explore the driving forces of the spatial reconstruction of rural settlements in mountainous areas. (2) From the perspective of peasant households, this study explored the five driving forces of external attraction, internal impetus, geological disasters, production culture, and life culture in the spatial reconstruction of rural settlements in mountainous areas. (3) Place attachment was used as an influencing factor in peasant households’ reconstruction intention. It was added to study the driving forces of the spatial reconstruction of rural settlements in mountainous areas. This study provides an important guide for the implementation of rural planning and construction in the Panxi area in the future. It can also provide references for the study of driving forces for spatial reconstruction of mountainous rural settlements in other cities or countries.
This study had certain limitations that can be addressed in future research. The spatial reconstruction of rural settlements is complicated and influenced by many factors. This study did not consider peasant households’ unique features, family features, and policy perceptions. A supplementary questionnaire survey will investigate these factors. Additionally, future studies will explore the reconstruction mode of rural settlement spaces based on the identified driving forces.
5. Conclusions and Implications
In this study, the driving forces for the spatial reconstruction of rural settlements in mountainous areas were explored by using the questionnaire survey data of rural families in the Panxi area of Sichuan Province, China, from 2017, and establishing an SEM. The following major conclusions were drawn:
(1) Among the 16 sampled villages in the Panxi area, most peasant households had a very strong intention towards the spatial reconstruction of rural settlements.
(2) Important factors that influenced the reconstruction intention of peasant households were infrastructure constructions (e.g., electronic facilities and medical conditions), the living environment, the convenience of information accessibility, the farming culture, the pasture culture, dietary habits, and geological disasters.
(3) Place attachment, living mode, language, traffic, water resources, and development opportunities slightly influenced the reconstruction intention of peasant households.
(4) Geological disasters were the main driving force for the spatial reconstruction of rural settlements in mountainous areas, while life cohesion had the least influence.
The spatial reconstruction of rural settlements is an inevitable change in conformance with social and economic development. Although the spatial reconstruction of rural settlements in mountainous areas faces various challenges, the extensive support of peasant households in mountainous areas proves it is the right time for rural space reconstruction. The occurrence of geological disasters poses an overwhelming threat to the safety of rural households in mountainous areas. The Chinese government has introduced a relocation policy for poverty-stricken peasant households living in deep mountains, such as those with many geological disasters, inconvenient transportation, limited information, and poor living conditions. By improving the infrastructure of the resettlement area, supporting education and medical facilities, and guiding the employment of peasant households, the government can attract rural households in mountainous areas to relocate to resettlement areas with better living conditions, such as central villages, central towns, or industrial parks. This model can promote rural revitalization; the furthering of plans for broader, coordinated urban-rural development; and land-use management policies. After peasant households in mountainous areas moved with government assistance, their original homesteads were demolished. The government adjusted the land suitable for farming to reclaim and return to farming through the transformation of villages and towns and the merging of villages and townships, dramatically improving land-use efficiency. The most important thing is that during the spatial reconstruction of rural settlements, governments should consider a combination of the specific local rural conditions, peasant households’ reconstruction intentions, and important influencing factors for these households as the basis for rural planning and construction. Such spatial reconstruction of rural settlements via government guidance combined with peasant household participation would facilitate an efficient reconstruction process by showing respect for their intentions.