1. Introduction
According to the World Health Organization, almost everyone (99%) breathes air that exceeds WHO guideline limits and contains high levels of pollutants, with low- and middle-income countries (LMIC) experiencing the highest amount of exposure. Ambient (outdoor) air pollution in urban and rural areas causes fine particles that lead to stroke, heart disease, lung cancer, and acute and chronic respiratory diseases. Additionally, about 2.6 billion people are exposed to dangerous levels of household air pollution as a result of cooking with polluting open fires or simple stoves fueled by kerosene, biomass (wood, animal waste, and crop waste), and coal [
1]. In China, with rapid urbanization and industrialization, air pollution has become a significant social problem and an important public health issue [
2]. To address air pollution, the Chinese government has implemented several top-down national measures, including three top-level design documents for air pollution control issued between 2013 and 2021. Alongside these efforts, various social entities are increasingly concerned about the impact of citizens’ individual environmental protection behaviors on sustainable environmental development.
In the 1860s, there was already research on pro-environmental behavior; in the 1870s, environmental psychology and public concern about the environment sparked academic discussion, expanding the study of pro-environmental behavior into a wide range of academic fields. Researchers from the fields of psychology, geography, environmental planning and design, and sociology have all contributed scientific findings to the study of pro-environmental behavior. Until now, the influencing factors and generating mechanisms of pro-environmental behaviors have been of particular concern to academic researchers [
3].
At present, the discussion on the influencing factors of pro-environmental behavior mainly focuses on the macro and individual levels. The relevant achievements regarding the influence of macro factors on pro-environmental behavior mainly focus on the environmental pollution driving hypothesis [
4], affluence hypothesis [
5], and environmental Kuznets curve [
6]. From the individual perspective, socio-demographic factors and individual psychological factors are emphasized. The influence of environmental education on pro-environmental behavior has also received further attention. In one study, individuals who had received environmental education had a higher intention of engaging in pro-environmental behavior than those who had not [
7]. In the process of receiving environmental education, different educational methods may have an impact on the training effect of environmental awareness [
8].
Air pollution perception, as a form of individual environmental risk perception, is considered an important factor affecting pro-environmental behavior. The stronger the environmental risk that an individual perceives, the more likely they are to make the behavioral decision to adopt pro-environmental behavior to avoid such risks. Sociology emphasizes the influence of social macro structures on individual behavior [
9]. Environmental knowledge is not only the intellectual basis for individuals to make pro-environmental behavior decisions, but also represents the influence of structural factors such as social norms on individual behavior decisions, to a certain extent. Previous studies on environmental knowledge mainly focused on individuals’ understanding of environmental science knowledge [
10]. In the 2021 China General Social Survey, the measurement of environmental knowledge included the measurement of individuals’ understanding of environmental policies, environmental governance and other knowledge points, so that it could better reflect the influence of social norms and other factors on pro-environmental behaviors. The content of environmental knowledge in this paper is more focused on environmental policy and environmental governance, which are also characteristics of this research.
The main research questions of this paper are as follows: First, this paper discusses whether individuals’ subjective air pollution perception affects their choice of pro-environment behavior. Second, the paper examines whether environmental knowledge affects an individual’s pro-environmental behavior. Finally, this study hopes to discuss whether people with more environmental knowledge will have more pro-environment behaviors when they perceive air pollution. In other words, this study hopes to explore whether environmental knowledge has a regulating effect on air pollution perception and pro-environment behavior.
The contributions of this research are mainly related to the following aspects: First, this paper uses CGSS2021 data and multiple linear regression to test the impact of air pollution perception and environmental knowledge on pro-environmental behavior, enriching the relevant research on pro-environmental behavior. Second, We investigated the moderating effect of environmental knowledge on the relationship between air pollution perception and pro-environmental behavior, which provides a certain basis for the government to promote the public’s pro-environment behavior in the future.
The structure of this paper is as follows: The second part is a literature review and proposed research hypotheses. The literature review includes related concepts of pro-environmental behavior, the impact of air pollution perception on public pro-environmental behavior, and previous studies on environmental knowledge, leading to the proposed research hypotheses based on relevant studies in the literature. In the third part, we introduce the data source, regression model, variables, and the operational process of the variables. The fourth part presents the empirical results of this study. The fifth part discusses the findings of this paper. The last part provides a summary of this paper.
3. Methodology
3.1. Data Source
The Chinese General Social Survey (CGSS), started in 2003, is the earliest national, comprehensive, and continuous academic survey project in China. The CGSS systematically and comprehensively collects data from Chinese society—communities, families and individuals—in order to summarize the trend of social change in China, discuss issues of great scientific and practical significance, promote the openness and sharing of scientific research in China, provide data for international comparative research, and act as a multidisciplinary economic and social data collection platform.
CGSS2021 is a cross-sectional survey carried out by the CGSS project in 2021. The sampling procedure adopts multi-stage stratified sampling; the survey mode is by interview, and the population covers adults over 18 years old. CGSS2021 contains a total of 700 variables, including the core module and topic module of the 2021 survey for all respondents. The following modules were also randomly selected by one-third of the respondents: the additional East Asian Social Survey (EASS) health module; the health module of the International Social Survey (ISSP); and the environment module of the International Social Survey (ISSP). The total number of samples in CGSS2021 is 8148, and the final selection of effective samples in this study was 1901. Among the samples we screened, 931 were female, accounting for 48.97% of the total sample, and 970 were male, accounting for 51.03% of the total sample. The maximum age was 94 years, the minimum was 18 years, and the average age was 53 years. There were 1028 rural residents, accounting for 54.08% of the total sample, and 873 non-agricultural residents, accounting for 45.92% of the total sample. From the perspective of gender, age, urban and rural areas and other factors, the sample data are representative to a certain extent and can represent the broader Chinese population.
3.2. Variable Selection and Assignment
Dependent Variable: The dependent variable of this study was residents’ pro-environmental behavior. This involved questions such as: “Do you often deliberately separate glass, aluminum cans, plastic, or newspapers for recycling?”; “Do you often choose not to buy certain products for the sake of protecting the environment?”; “Do you belong to any organizations that protect the environment?”; “In the past five years, have you signed a petition on an environmental issue?”; “In the past five years, have you donated money to environmental groups?”; and “In the past five years, have you participated in a protest or demonstration on an environmental issue?”. Each of these six questions was assigned a value of “1” for participation and “0” for absence, and the total score was calculated to obtain a pro-environmental behavior score for each sample.
Core Independent Variables: The air pollution perception used in this study is an individual subjective variable. The question is: “How serious do you think the air pollution is in your area?”. Responses were assigned values as follows: “very serious” = 5; “serious” = 4; “general” = 3; “less serious” = 2; and “not a problem” = 1 for the air pollution perception variable.
Environmental knowledge scores were measured by 14 questions related to the question “How well do you know the following items or topics?”. The content included, as follows: “ecological civilization”; “ecological compensation”; “ecological protection red line”; “ecological civilization system reform”; “National ecological civilization pilot zone”; “ecological civilization construction target evaluation and assessment”; “land main functional area”; “circular economy”; “environmental protection inspector inspection”; “air pollution prevention action plan”; “water pollution prevention action plan”; “soil pollution prevention action plan”; “measures to protect public participation in the environment”; and “Party and government leaders’ responsibility for ecological environmental damage”. Responses were scored as follows: “do not know” = 0; “know some” = 1; “know more” = 2; and “know very well” = 3. These scores were then summed to obtain personal environmental knowledge scores.
Control Variables: Controlling variables in this study were the individual’s gender, years of education, political status, annual income, and household registration. Among these, gender was a binary variable. Education level was converted into years of education, as follows: “no education” = 0; “private school or literacy class” = 1; “primary school” = 6; “junior high school” = 9; “vocational high school”; “ordinary high school”; “technical secondary school” and “technical school” = 12; “junior college (adult higher education)” and “junior college (formal higher education)” = 15; “undergraduate college (adult higher education)” and “undergraduate college (formal higher education)” = 16; and “graduate and above” = 19. Age was a continuous variable. Political status was the binary variable of CCP members and non-CCP members. Household registration was a binary variable of agricultural household registration and non-agricultural household registration. The measure of an individual’s income was the logarithm of an individual’s annual income.
Second, when considering the influence of environmental concern on individual pro-environmental behavior, we included environmental concern into the control variable. Environmental concern was measured using the New Ecological Paradigm (NEP) scale. There are 15 items in the original scale. Chinese scholar Hong Dayong believes that the fourth and the fourteenth items should be deleted according to the actual situation in China, so as to improve the reliability and validity of the scale [
35]. Therefore, we adopted the remaining 13 items of the new ecological scale as environmental concern variables for this study. Items 1, 3, 5, 7, 9, 11, 13, and 15 are positive questions. The specific measurement topic includes, as follows: “1. The current total population is approaching the limit that the earth can bear”; “3. Human’s destruction of nature often leads to disastrous consequences”; “5. At present, human beings are abusing and destroying the environment”; “7. Animals and plants have the same right to existence as human beings”; “9. Although human beings have special abilities, they are still subject to the laws of nature”; “11. The earth is like a spaceship, with very limited space and resources”; “13. The balance of nature is very fragile and can be easily disturbed”; and “15. If everything continues as it is, we will soon suffer a serious environmental disaster”. The statements were assigned the following values: “totally disagree”, “somewhat disagree”, “neither agree nor disagree”, “somewhat agree”, “fully agree” as 1, 2, 3, 4, and 5, respectively. Items 2, 6, 8, 10, and 12 are negative questions, and the specific measurement questions included, as follows: “2. People are important and can change the natural environment to meet their own needs”; “6. The natural resources of the earth are abundant if we know how to exploit them”; “8. The self-balancing capacity of nature is strong enough to cope with the shocks of modern industrial society”; “10. The claim that humanity is facing an ‘environmental crisis’ is an overstatement”; and “12. Human beings are born to be masters, to rule the rest of nature”. The statements were assigned the following values: “totally disagree”; “more disagree”; “neither agree nor disagree”; “more agree”; “completely agree” as 5, 4, 3, 2, 1 respectively. Cronbach’s alpha coefficient of the scale after deleting items was 0.7442, which is considered as having a high degree of internal consistency and can be regarded as a single-dimension cumulative scale, which can be directly summed to obtain the environmental variables of interest.
Table 1 presents descriptive statistics of the main variables used for regression in CGSS2021.
3.3. Methods
In this study, a correlation analysis of the core variables—air pollution perception, environmental knowledge, and pro-environmental behavior—was conducted to address the multicollinearity problem among these variables. Subsequently, multiple linear regression was used to explore the relationships between air pollution perception, environmental knowledge, and pro-environmental behavior. The dependent variable was pro-environmental behavior, while the control variables included individual gender, income (logarithm), years of education, political status, urban versus rural location, and environmental concern. Model 1 examines the relationship between control variables and pro-environmental behavior; Model 2 explores the relationship between air pollution perception and pro-environmental behavior, including control variables; Model 3 analyzes the relationship between environmental knowledge and pro-environmental behavior, with control variables; Model 4 investigates the interaction between air pollution perception and pro-environmental behavior, including control variables; and Model 5 focuses on the moderating effect of environmental knowledge on pro-environmental behavior.
4. Empirical Results
4.1. Correlation Analysis of Core Variables
The dependent variable in this study is public pro-environmental behavior, while the core explanatory variables are individual air pollution perception and individual environmental knowledge. To preliminarily assess the relationships between public pro-environmental behavior, individual air pollution perception, and individual environmental knowledge, Pearson correlation analysis was conducted on these three variables. According to the correlation coefficients shown in
Table 2, individual air pollution perception and environmental knowledge both have significant correlations with individual pro-environmental behavior at the 0.001 level, with correlation coefficients of 0.148 and 0.274, respectively. The correlation coefficient between air pollution perception and environmental knowledge is 0.062, indicating a low level of correlation between these core variables. A multicollinearity test was performed on the three variables, yielding a variance inflation factor of 1.00, which suggests that there is no multicollinearity among the independent variables, allowing for further analysis.
4.2. Pro-Environment Behavior, Air Pollution Perception, and Environmental Knowledge
Models 1–5 are the outcomes of multiple linear regression analyses. Model 1 primarily examines the influence of control variables on pro-environmental behavior. Building on Model 1, Model 2 explores the correlation between perceptions of air pollution and pro-environmental behavior. Further to Model 1, Model 3 investigates the impact of knowledge about environmental policies on pro-environmental behavior. Model 4 integrates both air pollution perception and environmental knowledge, extending Model 1. Model 5 introduces the moderating effect of environmental knowledge, building upon the variables included in Model 4.
Table 3 shows the results of multiple linearity.
Figure 2a shows the fitting line between air pollution perception and pro-environment behavior, and
Figure 2b shows the fitting line between environmental knowledge and pro-environment behavior.
Model 1 primarily assesses the impact of control variables on pro-environmental behavior. Holding other variables constant, the regression analysis showed that gender significantly affected pro-environmental behavior, with males generally scoring 0.088 units lower than females (β = −0.088, p < 0.05). A significant positive correlation was observed between the logarithm of income and pro-environmental behavior, with an increase of 0.056 units for every unit increase in the logarithm of income (β = 0.056, p < 0.01). Years of education positively influenced individual pro-environmental behavior, which increased by 0.023 units for each additional year of education (β = 0.239, p < 0.001). Political affiliation was also significantly correlated with pro-environmental behavior, with Communist Party members scoring 0.123 units higher than non-party individuals (β = 0.123, p < 0.05). There was a positive correlation between environmental concern and pro-environmental behavior, with an increase of 0.019 units for every unit increase in environmental concern (β = 0.019, p < 0.001).
In Model 2, air pollution perception was added to the model, and the regression analysis showed a significant positive effect on pro-environmental behavior (β = 0.055,
p < 0.001), supporting Hypothesis 1. For each unit increase in air pollution perception, pro-environmental behavior increased by an average of 0.055 units. This is consistent with the results of previous studies [
36].
In Model 3, environmental knowledge was added to the model to measure its correlation with pro-environmental behavior. The results showed a significant positive correlation (β = 0.033,
p < 0.001), with pro-environmental behavior increasing by 0.033 units for each unit increase in environmental knowledge, thus supporting Hypothesis 2. In this study, the positive effect of environmental knowledge on pro-environmental behavior was confirmed, but the effect of environmental knowledge has not been verified in other studies [
37]. This may be because the measured content of environmental knowledge is related to the underlying variables implicit in the content.
In Model 4, both air pollution perception and environmental knowledge were added to the model simultaneously while controlling for other variables. The results indicated a significant positive effect of air pollution perception (β = 0.058, p < 0.001) and a significant positive correlation with environmental knowledge (β = 0.034, p < 0.001).
In Model 5, the moderating effect of environmental knowledge was added; the interaction between environmental knowledge and air pollution perception was found to be significantly positive (β = 0.005, p < 0.05). This indicates that environmental knowledge strengthens the influence of air pollution perception on pro-environmental behavior, with more knowledgeable residents experiencing a stronger impact of air pollution perception on their behaviors, supporting Hypothesis 3.
4.3. Robustness Test
We used the method of increasing control variables to test robustness. According to Wan et al. [
38,
39], it is believed that individuals’ social trust will affect residents’ pro-environmental behavior; therefore, this study incorporated social trust into the benchmark model.
Social trust influences the public’s environmental awareness and pro-environmental behavior intention [
40], which is an important reason why we added social trust as an important variable in the robustness test. First, trust is an important reason for common efforts for the benefit of social goals [
41]. Trust is, to some extent, an important driver of commitment, reciprocity, and collaboration. Second, social trust is conducive to people’s awareness of social interests beyond personal interests, which is an important reason why individuals are willing to engage in pro-environmental behaviors [
42]. Third, a good trust environment can avoid free-riding behavior in pro-environmental behavior to some extent [
43]. Therefore, we believe that social trust should be an important factor influencing pro-environmental behavior.
The measurement of social trust in this study was obtained from the topic of CGSS2021—“Generally speaking, you and others agree that the vast majority of people in this society can be trusted”; the values of “strongly disagree”, “somewhat disagree”, “cannot agree to disagree”, “relatively agree”, and “very agree”, were, respectively, assigned as “1”, “2”, “3”, “4” and “5” in order to measure the social trust of each sample.
Table 4 reports the model results between air pollution perception, environmental knowledge, and pro-environmental behavior after social trust was added. Model 1 represents the relationship between the control variable and the pro-environment behavior. On the basis of Model 1, Model 2 included the variable of individual subjective air pollution perception; the results showed that after adding the variable of social trust, air pollution perception was still significantly positively correlated with pro-environmental behaviors. Model 3 showed that environmental knowledge was also significantly positively correlated with pro-environmental behavior. Model 5 showed that the higher the level of environmental knowledge, the more likely people are to engage in more pro-environmental behaviors when they consider air pollution. The results of Model 1–Model 5 are basically consistent with the previous conclusions, which indicates that the conclusions of this paper have a certain robustness.
5. Discussion
In this paper, we examined the relationship between air pollution perception, environmental knowledge, and pro-environmental behavior using data from the CGSS2021 survey. A significant positive correlation exists between air pollution perception and pro-environmental behavior, which may be due to residents’ motivation to participate in pro-environmental behavior as a means of risk avoidance. To encourage public participation in pro-environmental behavior, it is essential to enhance the dissemination of environmental information and to strengthen environmental education. The disclosure of environmental information enables residents to intuitively perceive air pollution and other pollution conditions around them, thereby enabling them to form an objective risk perception and enhancing their ability to engage in pro-environmental behavior.
Second, environmental knowledge is also significantly positively correlated with pro-environmental behavior; environmental knowledge plays a positive regulating role in the relationship between air pollution perception and pro-environmental behavior, which is in line with our initial research expectations. From the objective basis of realizing pro-environment behavior, individuals with higher environmental knowledge levels are more likely to understand the causes of air pollution and how to engage in pro-environment behavior. From the perspective of subjective motivation, individuals with higher environmental knowledge levels are more likely to recognize the environmental pollution caused by human activities and the individual’s responsibilities towards air pollution control.
On the policy-level discussion, at the policy-application level, we offer a possible train of thought for policymakers based on our research. First, based on the results of this paper, both air pollution perception and environmental knowledge play a positive role in pro-environmental behavior, and environmental knowledge plays a positive moderating role. On the basis of not considering the increase in pro-environmental behavior caused by the increase in external pollution, it is feasible for policymakers to further promote the disclosure of environmental information and the disclosure of pollutant information [
44]. It is not only the government that can participate in the disclosure of environmental information, but also relevant companies, which should be subject to extensive supervision. It is necessary to further build an environmental governance system with government as the lead, enterprises as the main body, and public participation [
45]. Therefore, it is necessary to issue relevant information disclosure policies and pollutant supervision policies at the macro level. On the other hand, environmental knowledge plays a very important role in promoting pro-environmental behavior. Daily dissemination of environmental knowledge is an important source from which individuals can receive environmental knowledge; it is also a feasible policy to incorporate environmental education into the formal education system. From the topic of environmental knowledge measurement, environmental education should include not only environmental science knowledge, but also environmental policy knowledge.
6. Conclusions
Using CGSS2021 data, we investigated the relationships between air pollution perception, environmental knowledge, and pro-environmental behavior. The study measured pro-environmental behavior from two dimensions—private and public—assessed using six questions. Within the public dimension, behaviors were further differentiated into radical and non-radical actions, allowing for a more nuanced assessment of individual pro-environmental engagement. Air pollution perception is primarily concerned with individuals’ perceptions of pollution levels in their local environment. The measurement of environmental knowledge encompasses recent Chinese policies and knowledge related to environmental governance and pollution control.
The findings indicate that both air pollution perception and environmental knowledge are significantly and positively associated with public pro-environmental behavior. Furthermore, environmental knowledge exerts a positive moderating effect on the relationship between air pollution perception and public pro-environmental behavior. However, the paper has some limitations. First, although the measurement of pro-environment behavior was derived from two dimensions, we combined the questions from the two dimensions into the overall pro-environment behavior variable and did not separately explore the impact of air pollution and environmental knowledge on pro-environment behavior occurring in the public sphere or the impact of air pollution and environmental knowledge on pro-environment behavior occurring in the private sphere. Nor did we measure whether the regulating effect of environmental knowledge in these two situations is significant. Second, integrating objective air pollution data could further refine the exploration of the interplay among air pollution levels, perceptions, and pro-environmental behaviors. Therefore, further research could be carried out from the following aspects. The first aspect is to further classify and discuss pro-environment behavior. Based on the classification of pro-environment behaviors, the paper discusses whether air pollution perception has the same effect on different types of pro-environment behaviors, and whether the moderating effect of environmental knowledge still plays a role in this process. The other aspect is to incorporate objective data on air pollution into the model. Some excellent studies in this section provide a good model for how we can calculate objective data on air pollution. In one study, researchers proposed a Multi-AP learning network that estimates pixel-level (grid-level) concentrations of multiple air pollutant species based on fixed-site measurements and multi-source urban characteristics, including land use information, traffic data, and meteorological conditions [
46]. Although our current research mainly focuses on the use of subjective pollution perception data, the inclusion of objective data will provide us with a broader thinking space with which to further explore the relationship between air pollution and pro-environmental behavior. The inclusion of objective data on pollutants will increase the logical chain of our study; some theories, such as stimulus–organism response (SOR) theory, could then be applied. Objective pollution data are external stimuli, while pollution perception is a subjective feeling. How the stimuli and subjective feelings affect residents’ pro-environment behaviors, whether the effects are comparable, and whether there is some conduction relationship between them and pro-environment behaviors, are research directions that require further exploration. In the process, further excellent research on pollutant calculation should be conducted. A special emphasis should also be placed on the choice of model for this study. The use of multiple linear regression is quite reasonable in this study, but it also has its limitations. For example, it is very possible that some important variables have been missed that are not included in the model, and it is also easy to overlook the possibility of mutual influence between variables. Finding a more suitable empirical model is also an important direction for future research.
This study explores the relationship between air pollution perception, environmental knowledge, and pro-environment behavior. The stakeholders directly involved include, as follows: the government, which has the responsibility to environmental information disclosure, supervision of the environmental pollutants, and transmission of the necessary environmental knowledge to residents; enterprises, which are important pollution discharge subjects and environmental information disclosure subjects; and residents who are engaged in pro-environment behavior. Further applications of this research mainly include the following aspects. At the theoretical level, this study verified the correlation between air pollution perception and residents’ pro-environmental behavior, as well as the moderating effect of environmental knowledge, which has enriched the relevant research. At the policy level, this study provides a decision-making basis for the government to further promote environmental information disclosure and environmental knowledge education. It is conducive to the government itself and requires enterprises and other subjects to further strengthen the disclosure of environmental information.