The theory of planned behavior (TPB) expands on Ajzen’s theory of rational behavior (TRA) by adding the variable of perceptual behavior control (Ajzen, 1985). TPB is one of the most important theories concerning individual behavior generation in social psychology. The theory considers behavioral intention to be the direct factor that determines behavior. At the same time, behavioral attitude, subjective norms, and perceived behavior control are three variables that can affect behavioral intention. Behavioral intention is a summary of the advantages and disadvantages of an individual’s inner judgment of whether to consider performing a certain behavior. Attitude refers to the evaluation of the approval or disapproval of a behavior. Subjective norms refer to an individual’s perception of whether a particular behavior is a type of social pressure that reflects the influence of important others or groups on individual behavior decisions, such as the views of family, friends, and colleagues. Perceived behavioral control refers to the difficulty of an individual’s perception of specific behavior and reflects their views on factors that promote or hinder the execution of the behavior (Ajzen and Driver, 1991; Ajzen, 2004).

Since TPB was first proposed, it has been extensively applied to explain and predict the occurrence of various behaviors. TPB has also been widely used in human and co-participation behavior research (Chen et al., 2020; Empidi and Emang, 2021; Li et al., 2021). For example, Zeng (2017) used TPB as the research framework to analyze the feasibility and difficulties of public participation in environmental governance and constructed a participation mechanism of “empowerment-recognition-cooperation” to improve the level of public participation. Research has shown that the public’s attitude toward environmental protection positively affects their willingness to participate in environmental protection behaviors (Zheng et al., 2017).

Before the introduction of the norm activation theory, most psychologists believed that social or material rewards were the factors that promoted individuals to engage in altruistic behavior. The norm activation theory proposed by Schwartz (1977) is considered important for the study of altruistic behavior in the field of social psychology. This theory is composed of three variables: awareness of consequences, the ascription of responsibility, and personal norms. The consciousness of the result refers to an individual’s awareness of the result of not engaging in altruistic behavior, as well as the consciousness of causing adverse consequences to others. Under normal circumstances, the stronger the individual’s perception of the results of a particular situation, the stronger the obligation to arrive, and the more likely the individual will activate individual norms to implement the corresponding altruistic behaviors. The attribution of responsibility refers to an individual’s sense of responsibility for adverse consequences. The stronger the individual’s sense of responsibility or awareness of the results, the more conducive they are to the production of personal behavior. That is, the stronger the awareness of the results, the more conducive it is for individuals to produce behaviors that conform to their personal norms. For example, when people think that protecting the environment is good for their health, some will regulate their environmental behavior and take pride in it. This is also a manifestation of social norms and self-morality. Thus, personal norms are internalized by social norms and a sense of self-moral responsibility. Violating personal norms can lead to guilt, self-denial, or a loss of self-esteem as personal norms create a sense of pride and self-esteem.

Since the introduction of norm activation theory, it has been widely used in altruistic behavior research and has been demonstrated to have significant explanatory and predictive power. However, in theory, the relationship between the main variables is constantly being adjusted. There are two relationships between the variables: the moderation model, which influences the behavior of individual norms, and the mediation model, which indicates that awareness of an outcome influences individual norms through responsibility attribution (Schwartz, 1973). Based on the norm activation theory, Onwezen et al. (2013) used an intermediary model to explore the public’s environmental behavior. Lu (2016) researched the factors influencing individual energy-saving behavior based on the extended norm activation theory. Empirical evidence shows that extended norm activation theory can be successfully applied to the study of individual energy-saving behavior. Stern et al. (1999) adopted the mediation model of norm activation theory to combine value theory with new environmental paradigm viewpoints and scales, proposing value–belief–norm theory (VBN). This theory is based on the application and development of norm activation theory in the field of public pro-environmental behavior research, which is considered to be the best theory for investigating various environmental pro-environmental behaviors. VBN theory connects the five variables of personal value orientation, NEP, result awareness, responsibility attribution, and individual norms through causality. In this causal relationship, each variable directly affects the next variable and influences subsequent variables. The value–belief–norm model includes three aspects, namely egoism, altruism, and ecological value, avoiding the unity of norm activation theory based on altruism (Onwezen et al., 2013). Ibtissem (2010) analyzed the public’s energy-saving behavior based on the value-belief-norm theory. Empirical evidence demonstrates that individual value orientation significantly affects awareness of the results. Awareness of the results influences individual norms through responsibility attribution, and individual norms directly affect individual energy-saving behaviors. Steg et al. (2005) analyzed the factors influencing energy policy to lower household carbon dioxide emissions and verified the VBN theory.

In the 1980s, the American scholar Bandura (1977) proposed the theory of social cognition, which mainly includes ternary interactive determinism, observational learning, and self-efficacy. Ternary interactive determinism refers to human behavior, environmental factors, and individual factors. Determinism considers the relationship between these three to be independent and interactive. Bandura proposed that the behavior of an individual is affected by the behavior of others. Observational learning indicates that individuals can generate new behaviors in their learning process by observing the behaviors or habits of others, as well as improving and perfecting their original behaviors in this process. Self-efficacy refers to an individual’s self-judgment of their ability to complete a certain activity at a certain level, including their beliefs, feelings, and mastery. Self-efficacy can be understood as an individual’s self-confidence and ability to complete an action. Therefore, self-efficacy can also be called self-confidence. Although there are several related results, with numerous studies being conducted on individual social learning behaviors, few studies have explored pro-environmental behaviors and influencing factors.

The research subject of this article is villagers who show an intense perception of changes in their local living conditions. Based on the formation of environmental factors, Mesch and Manor (1998) classified the perception dimension (PD) into NEP and SEP. Accordingly, personal empathy for the natural environment can trigger two forms of environmental behavior: the personal form, as reported by Lin et al. (2017); Xiu (2021), and Ying (2021), and the social form, which includes participation in social environmental protection organizations (Armitage and Conner, 2001; Berenguer, 2010). This study follows the aforementioned research, subdividing the PD of villager participation in rural micro-landscape construction into NEP and SEP. The following hypotheses are proposed:

According to planned behavior theory, subjective norms refer to social pressure on individuals (Ajzen and Driver, 1991; Berki-Kiss and Menrad, 2022; Dai et al., 2022). Norm activation theory points out that personal norm has the largest effect on personal behavior (Schwartz, 1973; Shi et al., 2017; Song et al., 2019). Based on a literature review of planned behavior theory and norm activation theory, this study divides subjective norms into personal norm (PN) and social norm (SN; Cialdini and Goldstein, 2004). Research indicates that PN would directly affect individuals’ behavioral intentions to enhance or maintain the environment (Hallin, 1995; Wynveen and Sutton, 2015). Thus, in the process of the construction of rural landscapes, the higher the self-regulation of villagers, the stronger their participation intention. In contrast to PNs, SNs refer to the behavioral norms observed by individuals or organizations (Schwartz, 1977; Terrier and Marfaing, 2015). Based on the above analytical results, the following hypotheses are proposed:

Planned behavior theory proposes that individual behavior is primarily affected by attitudes, subjective norms, and cognitive-behavioral control (Boldero, 1995; Kaiser and Gutscher, 2003). Norm activation theory posits that the resulting awareness directly manipulates PN (Nyborg, 2018). The value–belief–norm model argues that personal behavior is affected by PN and the result awareness (Vining and Ebreo, 1992; Bratt, 1999). Fishbein and Ajzen (1977) pointed out that participation attitude (PA) is a continuous and persistent positive (like) or negative (dislike) feeling that an individual has toward a specific thing or problem. Using the planned behavior theory and norm activation theory as research frameworks, Wang and Zhang (2017) verified that individuals’ attitude toward participation positively affected their participation behavior intention. Result awareness (RC) is an individual’s judgment of the consequences of behavior or policy after it has been implemented. Through the perception of environmental problems, individuals can establish an awareness of the results, which can promote their behaviors to participate more actively in landscape construction; in turn, this can help alleviate or reduce environmental problems. Previous studies have demonstrated that the stronger the awareness of certain behaviors, the stronger the sense of morality. In this case, individuals are more likely to engage in personal behavior (WESLEY Schultz and Zelezny, 1999). On this basis, this study adopts result awareness and PA as research variables and proposes the following hypotheses:

The original control dimension in TPB is perceived behavioral control. Bandura (1977) proposed self-efficacy to specifically refine perceived behavioral control. Subsequently, Ajzen (2002) proposed that the essence of perceived behavioral control is an individual’s confidence in completing behaviors, which they denoted as self-efficacy (SE). Based on this, the present study uses SE as a measure of the control dimension, which specifically refers to the size and confidence of residents that they can play a role in participating in landscape construction. Studies have shown that SE can promote the action of recycling at an individual level by influencing environmental behavior (Bandura, 1977, 1982; Ajzen, 2002). When faced with problems, individuals with greater SE tend to demonstrate positive aspects, thus promoting the development of the event toward success (Tsiligianni et al., 2020; Yu et al., 2020). Therefore, in the construction of rural landscapes, villagers with stronger SE are often more willing to participate. Based on the above analysis, this study proposes the following hypothesis:

Participation behavior intention (PBI) refers to personal intention and effort in making an attempt, which is the intrinsic driving force of personal behavior (Wang et al., 2020). TPB believes that the behavioral intention of the behavior subject is the best way to predict the behavior. Based on TPB, Xiao and Wang (2017) verified that farmers’ willingness to participate in behavior positively affects their participation behavior, and some scholars believe that willingness to participate in behavior directly determines participation behavior. Therefore, the stronger the villagers’ willingness to participate in rural landscape construction, the greater the possibility of their participation. Although planned behavior theory has been demonstrated to be effective in explaining and predicting personal behavior, there are still many behaviors that cannot be completely rendered in practical surveys (Kaiser, 2006; Bortoleto et al., 2012; Hu et al., 2018). According to the research model, villager participation behavior can be divided into two stages: motivation and implementation. Does PBI play a mediating role between motivation and behavior? Accordingly, the following hypotheses are proposed:

The city of Jinjiang is located on the southeast coast of Fujian Province and is at the heart of the Golden Triangle of southern Fujian. With China’s “new socialist countryside” construction (16th National Congress), “beautiful countryside” construction (18th National Congress) and the development strategy of “rural revitalization” (19th National Congress), China’s rural construction began from the outside in, characterized by in-depth ecological protection, cultural heritage, and industrial revitalization. As a pilot project for the creation of a beautiful countryside in China, Jinjiang continues to promote the construction of rural landscapes through micro-landscape creation. Thus far, 42 city-level Jinjiang and over 100 rural micro-landscapes have been established. Although the micro-landscape construction activity in the city of Jinjiang only started 2 years ago, it already had a significant effect on rural landscape construction in Jinjiang. Its fundamental role is to fully mobilize the enthusiasm of the local villagers, increasing the stickiness of rural communities and providing a new model for rural beautification and enhancing the heritage of rural culture. With characteristics including reproducibility, promotion, and positive effects, the city of Jinjiang in China represents a site with a high practical research significance.

Based on both national and international research results, this study designed a variable-measurement scale (Table 3). NEP and SEP primarily used the scale of Keshuai (2008), while PN perception used the scale of Gärling et al. (2003) and SN used the scale of Cialdini and Goldstein (2004) and Bissing-Olson et al. (2016). PA used the scale of Lee et al. (2014), while result awareness conformed to the scale of Onwezen et al. (2013). Self-efficiency was measured using the scale of Chen and Hung (2010), while the scale of Matthew (2017) was adopted for participation behavior intention and participation behavior. Responses to research questions were measured using a five-point Likert scale (1 = strongly disagree to 5 = agree).

A field survey was conducted in the Jinjiang countryside from 25 December 2018 to 30 December 2018. A total of 450 questionnaires were issued and 414 valid questionnaires were returned, with an effective rate of 92%. Although the structural equation model is applicable to large-sample research, the survey questionnaire cannot be infinitely large. Therefore, for the sample size of the survey, this study adopted the viewpoint proposed by Thompson (Thompson, 1998); that is, the ratio of respondents to measurement items should be at least 10:1–15:1. Among the valid samples, 47.3% were men and 52.7% were women. Therefore, the proportion of males and females was roughly equivalent. Most research participants were aged 31–40 and 41–50, accounting for 30.2 and 29.9% of the cohort, respectively. The vast majority of participants had a middle school to high school education. SPSS19.0 and AMOS24.0 were used to process the data and verify the conceptual model.

In this study, the Harman single-factor method and partial correlation coefficient method were adopted for questionnaire analysis (Eby and Dobbins, 1997; Livingstone et al., 1997; Podsakoff et al., 2003). A total of seven common factors with eigenvalues greater than one under non-rotated conditions were obtained, among which the variance contribution rate of the first common factor was 34.19% (i.e., less than 40%), indicating that the common method deviation of the questionnaire measurement scale was insignificant and appropriate for subsequent data analysis.

According to the standard of normal distribution proposed by Wu (2009), the absolute value of the skewness coefficient should be less than 3, and the absolute value of the kurtosis coefficient should be less than 10. The closer the coefficient values of the skewness and kurtosis coefficients are to 0, the more they conform to a normal distribution. As shown in Table 4, there was no significant difference in the standard deviation of all observed variables. A standard deviation value in the range of 0.544–0.881 can ensure high identification. The corresponding deviation coefficient was in the range of −1.294–0.297, while the kurtosis coefficient was in the range of −0.975–3.724, proving that the sample data approached the normal distribution. At present, Cronbach’s alpha coefficient is often used to test the internal consistency of variables (Liu et al., 2020). As shown in Table 4, Cronbach’s α coefficient of each latent variable was greater than 0.7, while that of the gross scale was 0.938. All composite reliability (ρ_c) values were higher than 0.7, proving the reliability of internal consistency, according to Esbensen et al. (2009). The average variance extracted (AVE) was used to explain the convergent validity of the measurement index. Except for the latent variable PA (AVE = 0.464), the AVE of the remaining latent variables was above 0.5. In confirmatory factor analysis, standardized regression coefficients, also known as factor loadings, represent the influence of common factors on measured variables. Except for the PA1 standardized coefficient of 0.459, the standardized factors of all observed variables were above 0.5, and significant at the level of 0.01, demonstrating the favorable convergent validity of the scale. By comparing the AVE square root, row, and column correlation coefficients of all latent variables in Table 5, the square root of AVE was the maximum value of its row and column, indicating that the latent variables had a high distinction validity. The critical ratio (C.R.) value is equal to the ratio of the unstandardized estimate to Standard Error (S.E.), which is equivalent to the t-test value. If the absolute value of this value was greater than 1.96, the P-value reached a significance level of 0.05, denoted by the symbol “*”; if the absolute value was greater than 2.58, the P-value reached a significance level of 0.01, denoted by the symbol “^**”; if the absolute value was greater than 3.25, the P-value reached a significance level of 0.001, denoted by the symbol “^***”.

In terms of the goodness-of-fit test of the villager participation behavior measurement model, the AVE quantity of the PA1 standardized coefficient participation attitude fell short of the requirement. In other words, although some indicators in the PA measurement model did not satisfy the requirement, they were a close fit. Therefore, the overall degree of fitting of the measurement model was consistent with the requirements, as shown in Table 6. The chi-square value (CMIN) of the model was 599.83, the degree of freedom (DF) of the model was 295, and its chi-square degree of freedom ratio (CMIN/DF) was 2.033, indicating that the model had a good degree of fit. The SRMR was 0.051. Although it did not meet the standard of <0.05, the difference in value was very small. Thus, it was considered to meet the standard level. The root mean squared error of approximation met the requirement of being less than 0.08. The parsimony normed fit index (NFI), Trucker-Lewis index (TLI), the comparative fit index (CFI), and the incremental fit index (IFI) all satisfy the criterion greater than 0.9. Therefore, the model has a good fit.

By combining the measurement model with the structure relation model test, the standardized structure model verified that the standardized load of the observed variables was in the range of 0.322–0.722, which was significant at the level of 0.01. As shown in Figures 2A,B, H1, H1(b), H2, H2(b), H3, H3(a), H3(b), and H5 were significant at a level of 0.001, while H1(a), H2(a), and H4 were significant at a level of 0.01. To determine whether different independent variables played a role in mediating variables, we adopted the bootstrap method to investigate the indirect effects of different independent variables on PBI. According to the results, the 95% confidence interval of the indirect effect estimate value among all independent variables (e.g., NEP, SEP, PN, SN, PA, result awareness, and SE), as well as PBI and participation behavior, excluded 0, indicating that the mediating effect was significant (Zhao et al., 2010), as shown in Table 7. This confirmed the ninth hypothesis of this study.

Perception refers to a series of actions carried out with an awareness of internal and external environments. In the PD, the results of the model analysis confirmed H1(a) and H1(b). Specifically, the perception of the natural environment positively affected the willingness to participate in behavior at a significance level of 0.001, and a willingness to participate in behavior had an indirect effect on participation. That is, for each additional unit of NEP, the willingness to participate in a particular behavior increased by 0.427 units. At the 0.001 significance level, SEP positively affected the willingness to participate in a particular behavior and had an indirect effect on participation behavior through the willingness to participate in the behavior. In the PD, SEP had a greater impact on willingness to participate in behavior than NEP. In terms of the mediation effect, the mediating effect of NEP (0.361) was stronger than that of SEP (0.332). The villagers’ perception of their living environment reflected their realistic needs for ecological livability. On the one hand, from the perspective of urban–rural integration, there is room for improvement in the rural environment and landscape, with villagers being particularly worried about their living environment. On the other hand, a good social atmosphere can effectively promote communication among villagers, which is conducive to the establishment of civil environmental landscape organizations, thereby enhancing the enthusiasm of villagers to participate in the construction of rural landscapes.

Norm refers to an individual’s code of conduct toward oneself or others (Terry et al., 1999). In the norm dimension, from the results of the model analysis, H2(a) and H2(b) were confirmed. Specifically, PN produced a positive influence on PBI at a significance level of 0.01 and exerted indirect effects on participation behavior by virtue of PBI. SN perception had a positive effect on PBI at a significance level of 0.001 and exerted indirect effects on participation behavior by virtue of PBI. In the norm dimension, the influence of PN (0.630) on PBI was greater than that of the SN (0.322). In terms of the mediation effect, the mediating role of personal norms (0.53) was greater than that of social norms (0.272). This may be because both personal and social norms are “soft constraint” procedures, making villagers to not only exercise self-discipline but also supervise each other. Once an individual violates rural value recognition and common norm behaviors, they inevitably become subject to rejection and moral condemnation from other members.

In the attitude dimension, the model analysis results confirmed H3(a) and H3(b). Specifically, PA positively influenced PBI at a significance level of 0.001, exerting indirect effects on participation behavior by virtue of PBI. Result awareness positively affected PBI at a significance level of 0.001, exerting indirect effects on participation behavior by virtue of PBI. The influence of result awareness (0.722) on PBI was greater than that of PA (0.491). With regards to the mediation effect, the mediating role of result awareness (0.611) was greater than that of PA (0.415). With the transformation of rural society from traditional to modern, the improvement in villagers’ subjectivity will continuously strengthen their sense of ownership of rural governance. When villagers assume the governance of the rural environment as their responsibility and obligation, their intention to participate in the construction of the rural micro-landscape will be reinforced accordingly.

In terms of the control dimension, H4 was confirmed. Specifically, SE was found to have a positive effect on PBI at a significance level of 0.01, exerting indirect effects on participation behavior based on PBI. In terms of the mediation effect, the mediating effect of SE on participation behavior was 0.344. This indicates that, with the improvement of villagers’ awareness of rural landscape construction, villagers become increasingly convinced of the important impact of the rural landscape environment on their lives and are willing to make an effort to create a healthy rural living environment. In addition, this positive self-control ability naturally encourages villagers to participate in the construction of rural micro-landscapes.

As shown in Table 8, the effect intensity of the four dimensions on villager PBI in the new first-order structure equation model was as follows, from highest to lowest: attitude dimension (0.635) > norm dimension (0.557) > control dimension (0.407) > PD (0.401).

Based on the above conclusions, we propose the following policies to promote villager participation in the construction of rural micro-landscapes:

Although villagers’ participation in landscape planning is an emerging field of research, there is an extensive history of achievements in research on public participation. From the perspective of sustainable development goals, it is clear that rural micro-landscapes with villager participation better promote the sustainable development of rural areas, accelerate the construction of attractive villages, and achieve the overall goal of rural revitalization. However, research on villager participation in rural landscape construction is a comprehensive process. As far as this field of research is concerned, due to the limitations of time and labor, this study has several limitations. The construction of rural micro-landscapes involves complex system engineering that integrates multiple fields and disciplines. This is further complicated by the fact that different regions have different economic cultures. Although the current micro-landscape construction model in Jinjiang has achieved good results, whether this can be popularized and promoted across China more generally remains to be seen. Therefore, future research will need to be conducted in different study areas, and the resulting similarities and differences in the results of villagers’ participation behavior will need to be analyzed.