Pro-environmental behaviors of farmers

The term "pro-environmental behavior" includes a wide range of behaviors, which can be interchanged with other words, such as ecologically friendly behavior, responsible environmental behavior, environmentally friendly behavior and green behavior. This study adopts Schwartz’s definition of pro-environmental behavior, which it discusses any behavior that protects the environment in the course of daily work, i.e., either a natural environment or an outdoor environment, to minimize the negative impact of human activities on the environment [11].

In the early study of pro-environmental behavior, social theoretical models based on the awareness of, attitudes towards and morality of environmental protection were widely used by researchers. This led to most environmental protection researchers attempting to increase the public awareness of rural environmental issues. Unfortunately, scholars later found that this approach is ineffective. The study of ecologically friendly behavior is mainly carried out in two aspects: altruism and self-interest. Tourism researchers who think that responsible environmental behavior has prosocial motivation usually adopt a theory based on ethics, such as Schwartz’s norm activation model (NAM) [12] or Stern’s recently introduced value-belief-norm theory (VBN) [13]. On the other hand, some researchers think that pro-environmental behaviors are driven by the interests of individuals. Thus, it is based on their high expectations of future benefits that farmers are motivated to protect the environment in specific areas [14]. The theory of self-interest is based on the premise that behavior is driven by desirable outcomes or expected rewards. Zhang Yuling (2014) further noted that the best approach is to mix the two aspects [15]. Keshavarz & Karami found an “attitude-behavior” gap among people of different professional identities. Nonfarmers, especially senior intellectuals, have a higher awareness of environmental protection. However, nonfarmers are less inclined to take positive actions than farmers [16]. Generally, farmers, and more precisely those who grow up in rural areas and rely on the rural ecological environment, will show more positive actions towards environmental protection.

As mentioned earlier, farmers empirically show a higher positive environmental attitude towards agricultural production practices. The awareness levels and norms of famers are found to be important predictors of pro-environmental intention because they serve as important emotional bonds between humans and ecological environment, and different degrees of participations may vary with various environmental protection intentions. Yu et al. found that if people feel more loving emotions towards a place, they are inclined to show a more positive responsible environmental behavior towards that place [17]. Lin et al. found that practitioners who feel a positive emotion of belonging to tourism are more inclined to show environmentally friendly behaviors at tourist destinations [7]. Agliardi & Agliardi pointed out that some respondents link environmental protection behavior with some constraints of time costs and higher spending; however, most of them show better environmental awareness and more positive environmental attitudes [18]. Olya and Akhshik also considered the barriers of pro-environmental behavior, including money cost, emotional input and sense of responsibility [9]. Given the background of this study, money cost, norms, and responsibility awareness may explain the environmental protection intention of farmers.

Norm activation model

Schwartz originally developed the norm activation model, which involves altruistic behavior. "Personal norms" are the core element of this model. Schwartz states that "personal norms" are described "as feelings of moral obligation, not as intentions’’ [12]. The framework of the model aims to put forward a scientific understanding of individual supportive attitudes in the context of environmentalism. The support of the public is now regarded as one of the key powers to solving these social issues. The problem of public environmental protection is a large-scale problem that needs to be solved by social movements. Public environmentalism is progress that government leaders, tourism organization managers and farmers must realize the benefits of diminishing the adverse impact of human beings on the environment. These personal norms are used in the NAM to predict individual behavior, which is influenced by two factors: the feeling of responsibility for a specific behavior and the awareness that performing a particular behavior has a specific consequence [12]. Some studies have interpreted the norm activation model as a mediator model, while other studies have interpreted it as a moderation model. Researchers who consider the NAM as a mediator model have suggested that the awareness of consequences influences norms through ascribed responsibility. Researchers who consider the NAM as a moderation model have suggested that the influence of norms on behavior is moderated by the awareness of consequences and ascribed responsibility. This study deems the NAM as a mediator model; thus, we interpret that farmers should be aware of the consequences of environmental protection behavior before they feel responsible for it. Then, feelings of environmental responsibility will activate personal norms, and these norms will induce one’s intention of pro-environmental behavior [19]. Although effective pro-environmental policies need most of the related groups to be involved, Michel and Moser considered that the most critical group consists of grassroots farmers [20]. Hopper agreed with the assertion that to realize some movement goals, changes in individual-level behaviors among ordinary people are indispensable [21]. Most Chinese farmers have land to cultivate, such that the environmental protection behavior of farmers on their land can be regarded as behaviors related to the private field [22]. Such pro-environmental behavior in private-sphere environmentalism includes the adoption, opposition, and negation of the private sphere; organizational behavior; and products, services, and policies related to their lands that may have a negative impact on the environment. According to the NAM theory, pro-environmental behavior can be divided into four components: awareness of consequences, ascription of responsibility, personal norms, and intention of behavior [12]. Based on the limited scope of the current study, only activities related to collectively owned fields or daily agricultural work are analyzed and discussed. Schwartz considered that these four basic components are relevant to an ideal cross-situational goal with different importance as that of the environmental protection intentions required in the daily work of individuals or other social entities [12].

Awareness of environmental consequences (AEC) is the first construct in the NAM framework; it refers to the awareness that the problem related to environmental protection can be enhanced or diminished by the biosphere, people, and other species [23]. Environmental awareness is often influenced or promoted by different types of collective views and conclusions [12]. Awareness of environmental consequences has been widely considered by various disciplines, and its relationship as a predictor of environmental attitude is often supported by empirical evidence. Schwartz and Boehnke explained that the awareness of consequence construct is complex and can be composed of many variables [24].

Ascription of environmental responsibility (AER) is the responsibility that an individual’s behavior can either reduce or exacerbate the potential negative influence of consequences [25]. Montada and Kals found that the ascription of responsibility is the greatest contributor to farmers’ willingness to support an environmental protective policy [26]. Although the relationships between all the variables of the NAM often produce good statistical results, compared with other theories, that of predictive ability still has problems.

As the next construct, environmental personal norms (EPNs) are considered a social rule that guides how individuals should comply and are restricted by the ascription of responsibility [11]. Kaiser and Bogner found that normative behavior theory has a higher predictive power than many other theories regarding the environmental protection behavior of specific interviewers [27]. Gnoth subdivided tourism motivation according to personal norms. Their findings highlighted the potential for the segmentation of culture and that personal norms are sensitive to the use of cultural interpretation [28].

At the end of this cause-effect chain, the last construct examines pro-environmental intentions (PEI). Although we only measure individual’s intentions, some researchers have considered intention it is an important indicator of behavior. These individual intentions are assumed to be motivated by a personal norm, which is identified as some unavoidable obligation to carry out environmentally friendly behaviors, and the name of the component is used interchangeably [29]. On the whole, awareness of consequences, ascription of responsibility and personal norms are the three influence factor constructs of the norm activation model proposed by Schwartz; they are designed to effectively predict the personal intention of behavior [12]. This framework has been previously empirically verified by many researchers [25, 30].

The causal chain of the NAM can be viewed as follows: AEC→AER→EPN→PEI. In view of this analysis and evidence, the following hypotheses are proposed (H1-H3):

1. H1. Awareness of environmental consequences has a positive effect on the ascription of environmental responsibility.
2. H2. Ascription of environmental responsibility has a positive effect on environmental personal norms.
3. H3. Environmental personal norms have a positive effect on farmers’ pro-environmental intentions.

Expectancy theory

Dunlap et al. explained that human beings need to protect the environment because they have the ability to disrupt the balance of unique nature; in addition, human society has growth constraints, such that humans have the right to adapt and rule the recognized parts of the natural world [31]. Bear argued that compared to environmentalists, the expectation of environmental protection reflects an individual’s view of the relationship between humans and the biosphere [32]. The intention of environmental protection can be predicted according to the types of values, which are an expression of individual expectations for environmental governance [25]. In the context of rural tourist destinations in which the ecological environment is polluted, Parijat and Bagga used expectancy theory to explore the formation of behavior choices about green tourism development [33]. Therefore, it is necessary to provide farmers with good expectations and active guidance to promote positive environmental protection intentions [34].

The classical theory of expectation motivation was proposed by Vroom [14]. His expectancy theory is rooted in the field of industrial and organizational psychology research. In the tourism industry, many researchers use this framework to capture tourism product suppliers’ motivations for destination image improvement, service attitudes, and community harmony. During this study, through a literature search, the author found that there was no previous clear application of expectancy theory in the field of analyzing behavioral pro-environmental intentions combined with NAM theory. However, after reviewing the NAM theory and other relevant theories, it was found that the two might be synergistic.

Expectancy theory is applied to explain the process by which individuals make decisions about various behavioral choices. The framework includes two different constructs: valence and expectancy. These two constructs determine the motivations that guide specific behaviors. The first component is valence, which determines to some extent the driving force or attractiveness of a specific outcome. The second component of expectancy theory is expectancy, which is the perception that a certain amount of performance will lead to the desired result. In other words, the more effort you put in, the more likely you are to get results. Finally, people believe that actions lead to positive expectations. In the study of Keshavarz & Karami, this theory was applied to predict how hard farmers support their environmental protective beliefs. The strength of the farmers’ work was affected by praises or rewards, whereas effectiveness with regards to how strong the effort should be ultimately ensured the desired results [16]. Therefore, the framework is expressed as follows:

In the influence relationship, the two components are examined separately and then determine a motivational force together. However, the study suggests that these two components should be treated within a causal chain, leading to a motivated intention that is conducive to pro-environmental behavior when supporting the environmental protection behavior in the tourist destination. Since expectancy theory was put forward, there have been many explanations. The lack of clarity in explaining the interactions between these constructs of motivational force has prompted researchers to analyze how each construct affects other constructs.

However, testing the construct between causal relationships is often pursued. The above causal chain is supposed resembled the following: Valence→Expectancy→Motivational force.

In this case, valence is at the beginning of this chain, which is indicated by a farmer’s value for a particular outcome. The goal includes the desire for outcomes such as sustainable tourism development that can be achieved by improving the ecological environment. This leads to farmers having the expectation that the more effort they put in to expecting to reduce pollution, the greater their impact on improving the environmental quality will be. Individual contributions may be small, but they are effective overall, thereby emphasizing the role of effort; thus, it is assumed that valence has a positive influence on one’s expectancy. Therefore, it is an opinion that action will lead to the desired result. Expectancy can affect environmental intention. From the discussion of expectancy theory, two assumptions are put forward:

1. H4. Valence has a positive effect on expectancy.
2. H5. Expectancy has a positive effect on farmers’ intention to behave pro-environmentally in tourist destinations in which the ecological environment is polluted.

Convergence of the two theoretical frameworks

Through a large number of studies that have focused on pro-environmental behavior research, we can find that discussions of the individual sociopsychological construct are more predictive towards pro-environmental behavior than are discussions focusing on the sociopopulation demographic background [35]. Variables such as awareness, responsibility, norms, and behavioral intention of environmental protection empirically refer to personal feelings and ways of thinking and ultimately affect solutions to environmental problems [36]. The theory has been confirmed, and its effectiveness has been tested [19]. In addition, compared to other theories, its pro-social construct of the NAM model has led researchers to agree that it has superior predictive power. However, the NAM has been studied along with other well-known attitude theories, such as Stern’s value-belief-norm theory [37]. Van and Meertens suggested that the combination of pro-social and rational choice theories can produce higher predictive ability [38]. Park and Ha combined the NAM and planned behavior theory into a single theoretical framework [39]. Their findings support the suggestion of Van and Meertens; the model shows a better precision regarding consumer recycling behavior than the independent NAM framework. The successful application of the merged model in interpreting individual environmental behavior demonstrates the validity of the NAM. In contrast, adopting the expectancy theoretical framework helps improve the performance of separate NAMs. However, while expectancy theory is a rational research framework that has been widely accepted worldwide, it still receives some criticism from researchers [34]. A prominent criticism of this theory is its lack of consideration for social impact [40]. By combining expectancy theory with the NAM model, it is assumed that this limitation will be reduced. The NAM originated from a theory that studies social movements. These measurement items involve individual awareness, responsibility, and personal norms about the behavioral intentions for environmental protection, which affect both human society and the local environment. The important view is that expectations for a better living environment and tourism livelihood income growth may affect one’s eagerness to protect the local environment [41]. Yuan et al. considered that personal norms are usually formed by social interaction, but one’s final decision is not simple [42]. Sometimes, one’s desired expectations may affect the amount of the individual’s actual effort and ultimately affect the environmental norms. Although expectancy theory measures a narrower range of variables than the NAM, the NAM does not need to consider the frequency of actions. Improving farmers’ intentions to support the protection of the ecological environment of tourist destinations requires repeated efforts rather than one-off actions. The impact of norms is essential to promote broad improvements in pro-environmental behaviors. Based on these discussions, the link between the two theories is hypothesized as follows:

1. H6. Valence positively affects awareness of environmental consequences.
2. H7. Expectancy positively affects environmental personal norms.

The proposed theoretical model includes a total of 6 constructs and seven hypotheses. The interactive relationships between the six constructs are illustrated in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. The proposed research model.

Constructs with dashed boxes represent the norm activation model (NAM).

https://doi.org/10.1371/journal.pone.0247407.g001

Study context and case selection

Guanshan village is located in the northeast portion of Ningxiang district, Changsha city, Hunan Province, China. It is a world-renowned Chinese rural tourist destination that is 24 kilometers away from Changsha city in the east. It is named after Guan Yu, who was the general of the Shu dynasty during the Three Kingdoms Period (the Three Kingdoms were composed of the states of Wei, Shu and Wu) and was stationed troops there. Guanshan, as a world-famous rural tourist destination with high-quality ecological environments that are being gradually destroyed, has many types of high-quality tourism resources.

The Guanshan Village is close to the metropolis of Changsha in the centre of China. As a large-scale rural tourism scenic spot, it won the honorary title of "The most charming leisure village in China" in 2012, which is the first famous rural tourism in Hunan Province to win this honorary title. In addition, as a whole scenic area, Guanshan Village has been officially approved by the National Scenic Area Evaluation Committee on September 18, 2013, and successfully established as a national AAAA scenic spot (the highest scenic spot in China is 5A, and Guanshan Village is one of the most famous rural tourism villages in Hunan province at that time).

Due to the rapid development of rural tourism, a large number of farmers in Guanshan Village have joined the rural tourism development industry. However, due to the overloading of local rural tourism services and the limited rural ecological carrying capacity, the contradiction between rural tourism supply and demand is prominent. Moreover, some local villagers even directly destroy the local ecological environment in order to obtain the income from rural tourism. It can be seen that the phenomenon of rural tourism in Guanshan Village is relatively developed, but the local ecological environment is also at risk of being destroyed. It can be seen that the analysis of the intention of local farmers in tourist destinations who participate in tourism for the protection of rural ecological environment is very critical.

There are two characteristics of the rural tourism found in Guanshan; one is the farmhouse tourism area, which was built in 2004. The farmhouse tourism model adopts a variety of mixed modes, such as "theme picking + agricultural experience + science education + green catering", and it takes the Fangyuan Mountain villa as the core area. Relying on five modern leisure agricultural bases, Guanshan can provide local agricultural products to foreign tourists. Another characteristic of rural tourism in Guanshan is its high-quality natural landscape and local historical buildings (see Fig 2). However, due to the increase in the number of foreign tourists and the excessive service pressure of these tourists, a large amount of sewage and domestic garbage is generated and directly flows into the local water body, causing damage to the local water body landscape (see Fig 3). At the same time, local farmers’ domestic wastewater is discharged arbitrarily, there is an excessive application of chemical fertilizers, garbage is incinerated in the core area of the tourist destination, and a massive amount of land is occupied for tourism reception. These behaviors have seriously damaged the quality of the natural ecological environment of the local tourist destinations, resulting in a decline in the tourist attractions in the recent period and a recent reduction in the number of tourists. Moreover, the tourist satisfaction score of Guanshan found on the internet has also decreased significantly. To improve the local rural tourism, it is important to study how to stimulate local farmers’ pro-environmental intentions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Landscape of Guanshan village.

Source: Photographed during field research.

https://doi.org/10.1371/journal.pone.0247407.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Appearance of Guanshan village.

Source: Photographed during field research.

https://doi.org/10.1371/journal.pone.0247407.g003

Data collection

Data were collected by the author and five undergraduate students majoring in human geography and urban planning from Hunan Agricultural University in 2020. The six investigators conducted field surveys in Guanshan village from May 1 to May 3, 2020. The author provided training and guidance to the investigators before their departure. Before issuing the questionnaire, respondents were asked if they were farmers in Guanshan village, whether they were affected by the harm of environmental pollution, whether they were currently engaged in tourism, and whether they agreed to fill the questionnaire anonymously for an academic study. Questionnaires were issued only after verbal agreement by the farmers. As the overall identity of the sample is relatively consistent, thereby providing a fairly homogeneous sampling profile for research and analysis, the collected data have high internal validity. A total of 305 responses were gathered for the survey. To maintain the credibility of the responses, if more than 10% of the data in a questionnaire’s responses were missing, the questionnaire was removed. Therefore, 9 surveys were deleted at the beginning of the process. The standard deviation of each response was then calculated to remove the unengaged responses if a case with a value less than 0.5 was noticed when reviewing the data for patterns exhibiting the nonengagement of the respondents. Thus, 21 responses are deleted due to unengaged performance. As a result, the actual number of valid responses was 275, and the effective rate of questionnaires was 90.16%. For a few missing data points, this study used the series mean to replace and then further analyze these valid responses.

Measurement instruments

The scale and items used in the measurement study were obtained from previous studies validated in various research contexts. Vroom’s expectancy theory is widely used in the study of the behavioral will of community farmers around tourist destinations. The use of a validated scale provides a solid theoretical basis; it also allows for accurate comparisons with other research results. Thus, the related items were modified to make them fit the setting of the current study. The measurement items of pro-environmental intentions, simulated by the final construct, are adopted from previously widely accepted research achievements, including I would give priority to accepting the guidance of agricultural experts on how to reasonably carry out agricultural activities around the tourist destination; I would sacrifice my rest time to promote the environmental protection policy; I would persuade others to protect the ecological environment of tourist destinations; I would spend more money to improve the tourism service facilities in order to reduce environmental pollution, I would learn relevant environmental protective skills and knowledge, etc. Additionally, the tone and language of the questionnaire used in the field survey was discussed to adapt an understanding of the habits of the target sampling population, i.e., farmers with low education levels. The purpose of this approach was to make it easier to understand the measurement items to minimize the number of invalid responses and reduce the potential errors. In this study, a five-point Likert scale was used for all measured variable items, and there was no coding problem. The scale of the questionnaire’s scores ranged from "extremely disagree" (1) to "extremely agree" (5). Before the final version was confirmed, the team invited some experienced academics, environmental experts, and undergraduates to conduct pretests and assessments. Based on their feedback, minor adjustments were made to the formatting, grammar, wording, experience, and visual construct of the respondents. Since the sample consisted of farmers engaged in tourism in Guanshan village, the final version of the questionnaire was presented in Chinese to help the farmers, many of whom do not have enough education to understand or read English. To avoid bias and personal input, the translators knew nothing about the research beforehand. In addition, to ensure successful expression, the results were in good agreement with the Chinese language scale version. A cover paragraph was placed at the beginning of the questionnaire, which briefly summarized the purpose of the study and briefly described the discipline of this study. Finally, the questionnaire included demographic questions; however, sensitive personal issues and personal contact information were not involved.

Tools for analysis

PASW Statistics 26 and AMOS version 21 were used for data analysis in this paper. In recent research, the combined use of SPSS and AMOS has been widely popular, mainly due to the combination of being easy-to-visualize and user-friendly when dealing with conceptual models. In the first stage, the adequacy of the proposed measurement model was analyzed, and the author used confirmatory factor analysis to assess the validity and reliability of the proposed research model. After achieving satisfactory fitting results, the second stage examined the proposed hypotheses in this study by providing path coefficients for the different hypothetical relationships. A model comparison between the existing model and the model proposed in this study was used to verify the constructed model, that is, to establish an adaptive and robust intention model to explain the environmental protective intentions of farmers in tourist destinations.

Sample profile

Farmers of GuanShan village comprised the research sample, and they regard tourism as the main source of their livelihood income. A summary of the study sample is shown in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic characteristics of valid responses.

https://doi.org/10.1371/journal.pone.0247407.t001

From Table 1, it can be seen that there are 275 valid samples. A total of 58.91% of the valid survey respondents were male (162), and 41.09% of the responses were female. The ratio of males to females was close to 3:2, which is in line with the local village officials’ judgment of the population. The average age of the 275 valid respondents was 50.47 years old. Thus, the respondents were mainly from the middle-aged and elderly populations; a total of 137 persons were aged between 36 and 50, accounting for 49.82% of the sample. Seventy-two persons aged between 51 and 60 made up 26.18% of the sample. The majority of the respondents had a mid- or low-level education; a total of 92 farmers had only a primary school education or below, accounting for 33.45% of the sample, while 106 farmers had a junior high school degree, comprising 38.55% of the sample, and 53 farmers had senior high school degree, accounting for 19.27% of the sample. Approximately 77.09% of the respondents reported that they were engaged in tourism full-time, while only 22.91% of the respondents reported that they were engaged in tourism part-time. From the perspective of annual income (after taxes) obtained from local tourism, most of the farmers’ reported incomes were not high compared to the per capita disposable income of Chinese residents (the per capita disposable income of Chinese residents was $4,433 in 2019). The specific differentiations were as follows: 24.36% of the farmers made between 1,732 dollars and 3,462 dollars (12001 to 24000 yuan); 13.45% of the farmers made between 3,463 dollars and 5,193 dollars (24,001 to 36,000 yuan); 31.27% of the farmers made between 5,194 dollars and 8,655 dollars (36,001 to 60,000 yuan); and 18.55% of the farmers made between 8,656 dollars and 14,425 dollars (60,100 to 100,000 yuan) (the average exchange rate of RMB against USD was 6.9326 yuan to 1 US dollar in January 2020).

It can be inferred that these responses about after-tax income distribution from those engaged in local tourism are relatively optimistic compared with the income level of other local farmers in Guanshan village. Under the promotion of environmental protection in rural areas, the relatively good income related to rural tourism will enhance farmers’ expectations for their future life, which will also prompt farmers to make positive pro-environmental intentions or behaviors. It can be inferred that the sustainable development of rural tourism closely refers to farmers’ intention of pro-environmental behavior. The ratio of the effective sample responses (275 valid samples) to the 24 observational items is much greater than 5:1, and the number of valid responses is more than 200, which can significantly reduce the risk of the nonnormal distribution of samples [43]. Therefore, the 275 valid responses cover the main attribute characteristic information of the sample, and the data distribution is in line with the actual situation.

Confirmatory factor analysis (CFA)

Testing the intention model goodness of fit is the first stage of the two-step approach proposed by Cheung. Mike and Chan. Wai [44]. Before the CFA, the research data were screened to check for possible violations of the proposed seven hypotheses. The test showed no apparent violation of the assumptions. The value of skewness ranges between -1 and 1, and the measurement items did not exceed this value range. Therefore, the items were not considered to violate the assumptions. The kurtosis values of the measurement items were all below the recommended threshold of 3. Further examination revealed no violations related to homoscedasticity, linearity, or multivariate normality.

The CFA factor analysis method is a maximum likelihood method, and the test results show that the proposed model has a goodness of fit (χ² = 566.559, df = 218, p<0.001, χ²/df = 2.599, IFI = 0.913, TLI = 0.888, NFI = 0.865, PGFI = 0.618, RMSEA = 0.076, CFI = 0.911). All the indices showed that the measurement model has good acceptability. According to Gholami, Asli et al., the CFI scores of the 275 valid samples were all above 0.90 for the 24 measurement items [45]. Each latent variable involves multiple measurement items; thus, to verify the internal consistency, the composite reliability test was carried out between different latent variables. The test scores showed that the values of each indicator were larger than most the minimum threshold of 0.70 accepted by most scholars, and their scores ranged between 0.899 and 0.912. In Table 2, the validity tests of the constructs are explained; the convergence validity of the different latent variables displays an average variance extraction (AVE) score between 0.511 and 0.688, which is larger than the generally accepted minimum requirement of 0.5 [45]. Finally, the AVE scores are larger than the correlation between the study constructs, providing evidence for discriminant validity [46].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. AVE, ASV, correlation, reliability, mean and standard deviation values.

https://doi.org/10.1371/journal.pone.0247407.t002

Structural equation modeling

The second stage of the two-stage approach proposed by Cheung, Mike W-L and Chan, Wai is to examine the structural model in the study [44]. Structural equation modeling (SEM) analysis was used to test the goodness-of-fit of the proposed model, and the results indicate satisfactory statistics (χ² = 583.536, df = 231, p<0.001, χ²/df = 2.526, IFI = 0.911, TLI = 0.893, NFI = 0.861, PGFI = 0.650, RMSEA = 0.075, CFI = 0.910). Based on the indices recommended by Hair (2014) [46], the fit of the hypothesized model was satisfactory. However, some suggestions concerning the modification indices from the AMOS output showed significant improvements in the proposed model. After adding one path from the ascription of environmental responsibility to expectancy, the goodness-of-fit statistics improved significantly. The originality of the proposed model of farmers was minimally influenced (χ² = 528.887, df = 217, p<0.001, χ²/df = 2.437, IFI = 0.922, TLI = 0.900, NFI = 0.904, PGFI = 0.620, RMSEA = 0.072, CFI = 0.921). For this purpose, the structural model after the path adjustment was made was taken as the final model for in-depth analysis and discussion in the current study. The purpose of the study is to develop a theoretical framework to help researchers to explain the pro-environmental intentions of farmers in tourist destinations where the ecological environment is polluted. Therefore, the adjusted final model was compared with the NAM’s original framework (χ² = 298.173, df = 102, p<0.001, χ²/df = 2.923, IFI = 0.936, TLI = 0.913, NFI = 0.905, PGFI = 0.701, RMSEA = 0.084, CFI = 0.935). The final model showed a higher goodness-of-fit than the proposed model mentioned above (χ²/df = 2.437) compared to (χ²/df = 2.526), and it also showed a better goodness-of-fit than the original NAM (χ²/df = 2.923). The results from a chi-square test of the final model and the proposed model revealed that there were significant differences between them (Δχ² = 37.672, Δdf = 1, p < 0.01). The goodness-of-fit of the final model was also relatively better than that of the original NAM. Moreover, these two models were distinctly different (Δχ² = 230.714, Δdf = 115, p < 0.01). The final model has a relatively excellent ability to predict farmers’ environmental protective intention (R² = 0.727) compared with the original NAM (R² = 0.684) and the proposed model (R² = 0.702). From the contrastive results, the final model has a 6.3% stronger predictive power of intention than the original NAM alone. A comparison summary of the structural models is displayed in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Results of the structural model comparisons.

https://doi.org/10.1371/journal.pone.0247407.t003

The relationships among the different constructs were examined as hypothesized. Then, an additional significant path was discovered from the analysis of the structural equation modeling.

Table 4 shows a summary of these findings; awareness of environmental consequences was found to be significantly related to the ascription of environmental responsibility. Ascription of environmental responsibility was found to be significantly related to environmental personal norms and expectancy. Valence was found to be significantly positively correlated with expectancy. Expectancy was found to be positively correlated with environmental personal norms. Expectancy and environmental personal norms were found to be significantly positively correlated with pro-environmental intention. Therefore, hypotheses 1, 2 and 3 were supported. As expected, the relationships among the NAM constructs were significantly positively correlated. Hypotheses 4 and 5 were also supported, as expected. Hence, the two constructs involved in expectancy theory were significantly positively correlated. Finally, we need to focus on the two hypothesized relationships that link the NAM framework to expectancy theory (H6 and H7). While hypothesis H7 displayed positive relationships and was thus supported, hypothesis H6 was found to have an insignificant positive correlation. Therefore, hypothesis H6 was rejected.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Standardized parameter estimates of the different influence paths.

https://doi.org/10.1371/journal.pone.0247407.t004

As mentioned above, the modification index analysis of the different influence paths referred to structural equation modeling and indicated the discovery of an additional relationship not found in the initial hypotheses. The study attempted to introduce this additional relationship into the final model, and it will be retained for further discussion with other models in future studies. The additional discovered relationship (D1) based on the proposed model was between the ascription of environmental responsibility and expectancy. Due to the inclusion of the newly discovered path, compared with the proposed model and the norm activation model, the final model was demonstrated to be improved in regard to model fit statistics. Hence, this study has opted to retain the discovered path. Fig 4 shows the results of the final model and the structural equation modeling analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Results of the final model of the structural equation modeling.

*p<0.01, constructs with dashed boxes represent the NAM theory.

https://doi.org/10.1371/journal.pone.0247407.g004

From the total variance explained by the final model, the results indicated that 25.5% of the ascription of environmental responsibility was explained by awareness of environmental consequences. Valence then explained 28.6% of the expectancy. As we expected, the incremental increases referred to the R² values in the ascription of environmental personal norms (44.3%), indicating that the total variance was explained by the antecedent variables. Finally, 72.7% of pro-environmental intention was explained by environmental personal norms and expectancy.

The indirect impacts among those constructs were assessed, and their results were as follows. Except for EPN, all the other factors had a positive and significant indirect impact on farmers’ intentions of pro-environmental behavior, namely, awareness of environmental consequences (β = 0.207, p<0.01), ascription of environmental responsibility (β = 0.412, p<0.01), valence (β = 0.214, p<0.01), and expectancy (β = 0.350, p<0.01). Environmental personal norms did not show a significant indirect impact on farmers’ intention of pro-environmental behavior. Additional evidence of mediation effects also shows that expectancy was indirectly affected by valence (β = 0.005, p<0.05) and awareness of environmental consequences (β = 0.163, p<0.01). Environmental personal norms were indirectly affected by valence (β = 0.195, p<0.01), awareness of environmental consequences (β = 0.242, p<0.01), and ascription of environmental responsibility (β = 0.154, p<0.01). Pro-environmental intention was also indirectly affected by valence (β = 0.214, p<0.01), expectancy (β = 0.350, p < 0.01), awareness of environmental consequences (β = 0.207, p<0.01), and ascription of environmental responsibility (β = 0.412, p<0.01). Details of the indirect impact assessment are shown in Table 5.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. The assessment of indirect impact.

https://doi.org/10.1371/journal.pone.0247407.t005