Literature Review and Theoretical Expectation

Public concern about GWCC is an important force shaping social and policy responses to climate change. Widespread and strong concern typically corresponds to people's decisions and their willingness to support climate change mitigation and adaptation policies.Footnote ¹⁰ GWCC have had a huge impact and inflicted much damage to China,Footnote ¹¹ especially in coastal areas vulnerable to sea-level rises and storm surges.Footnote ¹² In recent years, the Chinese government has passed various climate mitigation and adaptation policies to reduce GHG emissions and encourage a low-carbon economy and lifestyle. Successful policy implementation not only requires consensus from the public in perceiving the dangerousness of climate change but also depends on people's voluntary choice. Thus, studying GWCC concern is important in the China case. While there are extensive studies on the general environmental concerns of the Chinese public,Footnote ¹³ and a few insightful but non-random sampling-based studies on Chinese climate change attitudes,Footnote ¹⁴ research using a national representative survey to systematically probe Chinese citizens’ concern about GWCC has been lacking.

In contrast to the scant research on this topic in China, the literature on public GWCC concern in Western countries has grown rapidly in recent years. For example, there are numerous studies on public GWCC concern in countries such as the United States,Footnote ¹⁵ United Kingdom,Footnote ¹⁶ GermanyFootnote ¹⁷ and Australia.Footnote ¹⁸ There are also studies that compare climate change and related views cross-nationally.Footnote ¹⁹ Generally, the extant literature suggests that variations in public GWCC concern are associated with two groups of variables: individual-level factors, such as socio-economic status and personal beliefs or values;Footnote ²⁰ and regional-level place-based characteristics, such as local physical vulnerability to climate change.Footnote ²¹

Research on the relationship between individual-level factors and GWCC concern usually focuses on standard socio-economic characteristics (such as gender, age, education, income, place of residence, race, etc.). While some personal-level variables such as household income and place of residence (for example, urban versus rural) are found to be not regularly associated with the variation in citizens’ GWCC concern,Footnote ²² several other individual-level social characteristics, discussed below, have a consistent effect on individuals’ GWCC concern in existing studies.

Among the individual-level social variables, past research reveals that the most consistent predictor for citizens’ GWCC concern is gender: women tend to have greater GWCC concern than do men.Footnote ²³ Based on a review of more than 100 studies, Aaron McCright and his colleagues report that “in no reviewed study did we find that men report stronger pro-climate views than do women.”Footnote ²⁴ Theoretical reasons for this gender gap in GWCC concern (and other environmental concerns) can be found in the literature that explains how individuals in any society learn about gender-based expectations, division of labour and value-formation processes.Footnote ²⁵

The next most consistent social predictor for GWCC concern seems to be the age of the respondent. Much past research finds that younger adults usually report higher GWCC concern and tend to support GWCC policies more than do older adults.Footnote ²⁶ A theoretical explanation for this age effect is that as countries make the transition to postmodern nations, a growing cohort of citizens will experience shifts in personal values, have more educational opportunities, and then develop greater concern for environmental and ecological issues such as climate change.Footnote ²⁷ Robert Jones and Riley Dunlap analysed 11 sociodemographic and sociopolitical factors and found that age was the best predictor of environmental concern, with younger people expressing greater concern than older people about the environment.Footnote ²⁸ Nevertheless, there are a few studies that find that age has mixed effects on GWCC concern.Footnote ²⁹

Past studies also find that education often exhibits a positive effect on GWCC concern, since a higher level of education facilitates the understanding of the phenomenon and enhances the likelihood of believing that global warming has already begun.Footnote ³⁰ Although a few studies find an inverse association between education and GWCC concern in the United States,Footnote ³¹ evidence from many other studies seems to suggest a positive relationship between education and GWCC concern.Footnote ³² For example, using an extensive dataset, gathered from 26 countries, on topics surrounding environmental concern, Emilia Tjernström and Thomas Tietenberg find that a higher level of education is associated with a higher likelihood of expressing GWCC concern.Footnote ³³ Similarly, in another study, which was based on data gathered from 119 countries, Tien Ming Lee and colleagues find that educational attainment is the single strongest predictor of climate change awareness worldwide.Footnote ³⁴

In addition to the three social base variables (gender, age and education), the extant literature finds that personal post-materialist values are another important individual-level predictor for citizens’ GWCC concern. Starting with the theory of a hierarchy of human needs,Footnote ³⁵ scholars contend that humans begin to emphasize and pursue higher needs once they are satisfied with basic physiological and security needs (such as food, shelter and safety). In a series of influential studies, Ronald Inglehart described this shift as a pursuit of “post-materialist” values.Footnote ³⁶ Emphasizing a greater desire for quality of life, civil rights, autonomy and self-expression, instead of economic gains and material possessions, post-materialist values have been credited with the rise in environmentalism and individuals’ concern about non-material objects.Footnote ³⁷ Recent empirical studies find that post-materialist values exert a positive impact on citizens’ GWCC concern. Thomas Dietz, Amy Dan and Rachael Shwom together report that individuals with stronger post-materialist values (i.e. placing democracy and freedom of speech above maintaining order and fighting rising prices) tend to exhibit a higher GWCC concern.Footnote ³⁸ Similarly, Berit Kvaløy, Henning Finseraas and Ola Listhaug, using data from the 2005–2009 World Values Survey across 47 countries, also demonstrate that perception of the seriousness of the GWCC problem is positively correlated with post-materialism.Footnote ³⁹

While most of the studies on public concern about GWCC discussed above tend to focus on individual-level social base and personal value factors, there are a few exceptions that show that regional context characteristics (for example, environmental pollution) can also affect citizens’ environmental concern.Footnote ⁴⁰ To explain citizens’ GWCC concern, scholars contend that regional factors such as local climate vulnerability and exposure to climate-related physical risks should be taken into account.Footnote ⁴¹ As people exposed to climate change stressors are more likely to experience and understand climate change consequences, or know someone who has personal experience of the consequences of global warming, they are more likely to be concerned about GWCC.Footnote ⁴² Similarly, So Kim and Yael Wolinsky-Nahmias find that a general measure of climate vulnerability correlates with higher commitment to climate policies, including greater willingness to pay and stronger support for pro-climate energy polices.Footnote ⁴³ Patrick Egan and Megan Mullin find that heatwave exposure has a positive effect on GWCC concern because it increases the likelihood of believing that there is solid evidence that the earth is getting warmer.Footnote ⁴⁴ Samuel Brody and his colleagues find that public perceptions of the potential negative impacts of climate change increase when the threat or sense of climate vulnerability is most overt.Footnote ⁴⁵

In addition to climate vulnerability, we extend the current literature to consider another regional-level factor that may potentially condition individual citizens’ GWCC concern – regional economic dependency on carbon-intensive industries. Theoretically, it is reasonable to expect that people from regions where the local economy is more dependent on carbon-intensive and fuel-burning industries would be less likely to care about GWCC owing to their strong carbon-based economic self-interests. Empirically, past studies have presented some evidence that economic conditions shape public environmental attitudesFootnote ⁴⁶ and climate change concern.Footnote ⁴⁷ A recent piece of research on the geographic variation in public GWCC opinions in the United States finds “lower levels of belief that global warming is happening in some Midwestern and Western counties with large greenhouse-gas-producing industries, such as coal-fired power plants.”Footnote ⁴⁸ Studies on Appalachia also find that coal miners tend to support the mining industry while caring less about the impact of coal production on the climate.Footnote ⁴⁹ However, how regional economic factors such as carbon economic dependency affect public GWCC concern has not been systematically examined, and this is one of the contributions of this study.

Informed by these theoretical explanations and past empirical studies, we investigate how Chinese GWCC concern is affected by both individual-level variables and regional-level characteristics. Specifically, for individual-level characteristics, we expect that in China, based on the best available empirical evidence from past research, women, younger people, the better educated and those with stronger post-materialist values will be more likely to express greater GWCC concern compared to their counterparts. For regional-level factors, we expect that Chinese citizens who are located adjacent to the coastline, where climate risk is relatively higher than for non-coastal areas, will have higher GWCC concern than people living in inland areas. We also expect that Chinese citizens living in higher carbon economic dependency regions will exhibit less GWCC concern than their counterparts, owing to their carbon-vested interests.

Data Sources and Variable Measures

To examine Chinese public GWCC concern and its determinants, we use data from two sources: the China Governance and Public Policy Survey (CGPPS) and the official China Statistical Yearbook.

All individual-level data (such as respondents’ GWCC concern and socioeconomic status information) are drawn from the CGPPS, which is a research project of the Bush School of Government and Public Service at Texas A&M University. The CGPPS consists of multiple surveys conducted during the second half of 2016. In this paper, we use the data from a survey carried out between 28 October and 5 December 2016. All of the CGPPS surveys were conducted by the Survey and Research Center for China Households at the Southwestern University of Finance and Economics in China using its random-dialling computer-aided telephone interviewing system. The surveyors employed a stratified, three-stage probability proportion-to-size random sample design to draw a representative sample of Chinese adults aged 18 and older in mainland China (excluding Tibet, Xinjiang, Hong Kong and Macau).

As of 2016, China administers 33 provincial-level regions (called province, autonomous region, municipality or special administrative region) and 2,854 county-level divisions (called county/autonomous county, district, county-level city, or banner/autonomous banner). The GCPPS survey covered 29 provincial-level regions and 353 county-level divisions. For this survey, a total of 7,298 respondents were sampled and 3,972 interviews were completed, yielding a response rate of 54.43 per cent. After data cleaning, the final survey dataset includes 3,748 valid cases. All interviews were conducted in Chinese, with the median survey completion time of 14.5 minutes.

All regional provincial-level data are drawn from the China Statistical Yearbook, which is published annually by the National Bureau of Statistics of China. From the year books, we collected the past five years’ data (2011–2015) on each province's GDP, population, land size and the share of carbon-intensive industries in the province's GDP. In the statistical yearbooks, China's GDP is contributed by three broad sectors or industries – primary industry (agriculture), secondary industry (construction and manufacturing), and tertiary industry (the service sector). Since the primary and secondary industries depend heavily on fossil fuel burning and produce a great amount of GHG emissions, together they are considered as carbon-intensive industries. The data from the five years prior to the 2016 CGPPS survey were averaged and merged with the CGPPS opinion survey data.

The dependent variable in this study is GWCC concern, which is measured by a survey participant's perception of the seriousness of this issue in China. In the GCPPS survey, respondents were asked, “on a 0–10 scale, with 0 being not serious at all and 10 being very serious, how serious do you think the problem of global warming and climate change is in China?” Participants’ responses to this question are coded accordingly, so the higher score indicates a higher level of GWCC concern.

Through the GCPPS survey, we collected respondents’ sociodemographic background and economic status information, including the three individual-level variables that we are mainly interested in when explaining Chinese citizens’ GWCC concern: gender, age and education. Gender_female is measured as 1 for female and 0 for male. Age is the respondent’s actual age in 2016. Education is coded with 9 levels (1 = no schooling; 2 = elementary school; 3 = middle school; 4 = high school; 5 = vocational school; 6 = advanced vocational school; 7 = college; 8 = masters; 9 = doctoral).

As discussed in the literature review, another important individual-level predictor for GWCC concern is expected to be the extent to which people hold personal post-materialist values. A standard way of measuring post-materialist values in the extant literature is to construct a post-materialism index based on respondents’ answers to a relatively large number of questions regarding the issues they are most concerned about or what they perceive to be the top priorities facing the country.Footnote ⁵⁰ Some studies suggest a simplified, but still valid, approach to measure post-materialism: those respondents who put democracy (for example, giving people more say in government decisions) and civil liberty (for example, protecting freedom of speech) above maintaining order in the nation and fighting rising prices are considered to have stronger post-materialist values.Footnote ⁵¹ In this study, we follow this approach and use participants’ responses to a question that measures the extent to which people hold post-materialist values with respect to political democracy and civil liberty in China. The question asks respondents to indicate how serious they think the issues of political democracy and civil liberty are in China, ranging from 0 to 10, with 0 being not serious at all and 10 being extremely serious. The variable of post-materialist values is thus measured on a 0–10 scale, with a higher score indicating a stronger need for political democracy and civil liberty and thereby stronger post-materialist values.

Additional individual-level social background information was also collected in the survey. Income is measured as logged household per capita income in 2016. The type of residence, Urban residence, is measured as 1 for urbanite and 0 for rural resident. As past research indicates that political orientations can shape individual GWCC concern, we also collected the respondent's political party membership information in the survey.Footnote ⁵² Although the left–right or liberal–conservative political ideology division does not apply to China, one previous study found that being a Chinese Communist Party (CCP) member positively affects one's concern about the environment.Footnote ⁵³ In this study, CCP member is measured as 1 for being a CCP member and 0 being otherwise. These variables are used as control variables in our regression analysis later.

At the regional level, we are interested in how different contextual conditions, particularly regional climate vulnerability and carbon economic dependency, affect respondents’ GWCC concern. We measure climate vulnerability based on a province's geographical location. We create a dummy variable, Coastal region, to indicate whether the province is located in a coastal region. Because these areas are low in elevation and more likely to be stressed by rises in sea level and storm surges, they are more vulnerable to GWCC. We use Carbon-intensive industries’ share of GDP in each province to measure provincial economic dependency on fossil fuel-burning, greenhouse gas-producing industries. A higher percentage of the share indicates that the province's economy relies more on carbon-intensive industries. We also use the province's GDP per capita (log) to reflect the variations in regional economic affluence, and Population density (measured by 10,000 people per square kilometre) to capture different demographic concentrations across provinces. These two regional-level variables are used as control variables in later regression models. Summary statistics of all the variables at both individual level and province level are reported in Table 1.

Table 1:

Summary Statistics

Data Analyses and Results

We conduct data analysis in three steps. First, we begin with descriptive statistics and distribution graphs to show the general patterns and variations of Chinese GWCC concern. Second, we compute bivariate correlations between GWCC concern and each of the individual-level and regional-level variables to examine the possible associations between them. Third, we further employ multiple regressions with different specifications to estimate the conditional and overall effects of these variables on Chinese GWCC concern.

Descriptive statistics and distributions

We begin with descriptive statistics and basic distributions. As reported in Table 1, the mean GWCC concern for the Chinese respondents is 6.215. Recall that GWCC concern is measured on a 0–10 scale, with 0 being completely unconcerned and 10 being extremely concerned. This average concern of 6.215 is just moderately above the mid-point 5 (neither unconcerned nor concerned) – a level similar to that which has been found in the United States in a study using the same 0–10 scale to gauge the American public's GWCC concern.Footnote ⁵⁴ We consider this level of average GWCC concern in China to be relatively low because the intensity of public GWCC concern in many other countries around the world appears to be much greater.Footnote ⁵⁵ Recent public opinion polls across multiple nations show that China, along with several other countries including the United States, is among the least GWCC-concerned countries in the world, and our data here provide corroborating evidence supportive of these studies.Footnote ⁵⁶

The relatively low intensity of Chinese GWCC concern is even more evident in Figure 1, which reports the distribution of the survey participants’ GWCC concern levels across the 0–10 scales. As shown in Figure 1, only 14.2 per cent (aggregated percentage from scales 9 and 10) of Chinese respondents report that they are very seriously concerned about GWCC, compared to the global median of 54 per cent who believe that GWCC is a very serious problem.Footnote ⁵⁷ Given the fact that China is the top GHG emitter in the world, this relatively low level of Chinese GWCC concern is alarming, suggesting there is substantial room for China to improve citizens’ awareness and enhance their understanding of the risks associated with GWCC.

Figure 1:

Distribution of Chinese Citizens’ GWCC Concern

The distribution of the survey participants’ GWCC concern in Figure 1 also indicates a substantial variation in Chinese citizens’ GWCC concern. While over half of the respondents (61.6 per cent, aggregated percentage from scales 6 to 10) expressed that they were concerned about GWCC, a significant percentage of the respondents (20.3 per cent) reported their concern level at the mid-point 5, yet the remaining portion (18.1 per cent, aggregated from scales 0 to 4) showed little or no concern at all about GWCC.

Still, the intensity of Chinese GWCC concern varies greatly across regions. Figure 2 displays the average GWCC concern of citizens in each province. Hubei (6.714), Beijing (6.680), Tianjin (6.613), Shaanxi (6.521) and Hebei (6.494) ranked at the top of the list, while Heilongjiang (5.848), Henan (5.850), Guangxi (5.942), Anhui (5.952) and Jiangxi (6.027) showed the lowest average GWCC concern among all provinces.

Figure 2:

GWCC Concern across Provinces

In addition to the variations in GWCC concern among Chinese citizens and across provinces, there also appears to be a difference in the average GWCC concern between coastal provinces and inland provinces. Figure 3 presents the average GWCC concern scores for coastal provinces (dark grey area) and inland provinces (light grey area).Footnote ⁵⁸ The average GWCC concern score for respondents from coastal provinces (6.255) is slightly higher than that for respondents from inland provinces (6.183). This difference might be attributed to two factors. First, as the coastal provinces have advanced levels of economic development, respondents from these areas tend to have less economic pressure and more environmental concern, including concern about GWCC. Second, this difference might also be attributed to different levels of vulnerability to climate change, as the coastal provinces, owing to their lower elevations, are more vulnerable and prone to GWCC risks (sea-level rises, storm surges, etc.) than inland provinces.

Figure 3:

Average GWCC Concern between Coastal and Inland Provinces

Bivariate correlations

To show the relationship between GWCC concern and each of the explanatory variables, we first report the bivariate correlations between GWCC concern and each of the individual-level factors. As shown in Table 2, four individual characteristics are significantly correlated with GWCC concern. GWCC concern is positively associated with Gender_female (r = 0.070, p < 0.01) and Education (r = 0.035, p < 0.05), indicating that Chinese women and better-educated Chinese tend to express higher GWCC concern than do their counterparts. Age is negatively correlated with GWCC concern (r = 0.090, p < 0.01), suggesting that younger Chinese are more concerned about GWCC than are older Chinese citizens. GWCC concern is also positively correlated with personal Post-materialist values (r = 0.300, p < 0.01). In comparison with the coefficients between GWCC concern and Gender_female, Age and Education, the coefficient between GWCC concern and Post-materialist values is much higher, suggesting that post-materialism is a relatively stronger predictor for individual citizens’ GWCC concern in China. In addition, GWCC concern appears to be positively related to Income and Urban residence, and negatively associated with CCP Membership, but the effects are not statistically significant.

Table 2:

Correlations of Individuals' GWCC Concern with Individual Characteristics

Notes:

*** p < 0.01, ** p < 0.05, * p < 0.1.

The bivariate correlations between GWCC concern and regional-level variables, including Coastal region, Carbon-intensive industries’ share of GDP, GDP per capita and Population density are reported in Table 3. Respondents located in coastal provinces (Coastal region) with better economic affluence (GDP per capita) and higher population density (Population density) seem to have higher GWCC concern than do their counterparts, but none of the correlations are statistically significant. Nevertheless, the bivariate correlation between GWCC concern and Carbon-intensive industries’ share of GDP shows a significantly negative relationship (r = −0.042, p < 0.05), suggesting that people located in provinces with higher dependency on a carbon-intensive economy are less concerned about GWCC. This significant and negative correlation between regional carbon economic dependency and individual GWCC concern seems to be consistent with our expectation.

Table 3:

Correlations of Respondents’ GWCC Concern with Regional Characteristics

Notes:

*** p < 0.01, ** p < 0.05, * p < 0.1.

To visually illustrate the negative correlation between regional carbon economic dependency and respondents’ GWCC concern, we use scatter plots in Figure 4 in which each spot represents a province (or a province-level administrative unit). The vertical axis is the average provincial GWCC concern, which is the mean of all respondents’ GWCC concern scores in a province. The horizontal axis is the carbon-intensive industries’ share of GDP in a province. Despite the two outliers in the data (Beijing and Shanghai), the fitted value (dotted line) in Figure 4 clearly shows a downward trend, suggesting that when the degree of a province's carbon-intensive industries’ share of GDP increases, the average GWCC concern decreases. After the two outliers (Beijing and Shanghai) are removed from the data, the scatterplot with the linear trend line, as shown Figure 5, still displays a negative relationship between regional economic dependency on carbon-intensive industries and a province's average GWCC concern.

Figure 4:

Provincial Average GWCC Concern and Carbon-intensive Industries’ Share of GDP

Figure 5:

Provincial Average GWCC Concern and Carbon-intensive Industries’ Share of GDP (excluding Beijing and Shanghai)

Multivariate regressions and results

The bivariate analyses above suggest that GWCC concern is significantly correlated with individual respondents’ gender, age, education level and post-materialist values, as well as a province's carbon economic dependency. However, these relationships need to be further assessed with rigorous methods. In this section, we conduct multivariate regression analysis to model the conditional and overall effects of the individual-level factors and provincial-level variables on individual Chinese GWCC concern.

Based on the discussions in the literature review and theoretical expectation section, we first propose the following equation to assess the explaining power of individual-level variables and provincial-level characteristics for respondents’ GWCC concern:

(1)$${\rm GWC}{\rm C}_{ijk} = {\rm} {\rm \beta} _0 + {\rm \beta} _1{\rm IC}^{\prime}_{ijk} + {\rm \beta} _2{\rm PC}^{\prime}_k + {\rm \varepsilon} _{ijk}$$

In Equation 1, GWCC_ijk denotes the GWCC concern of respondent i in county j of province k. The vector IC’_ijk stands for individual-level characteristics, which include the social base variables (i.e. gender, age, education, household income, urban residence status and CCP membership) and the social values variable (i.e. post-materialism score). The vector PC’_k represents the provincial-level characteristics, which contain the four regional variables discussed earlier (province's physical vulnerability, carbon economic dependency, economic affluence and population density).

Since estimations of individual-level variables may be biased by omitted local variables (for example, local policies and regulations) that could affect both individual characteristics and GWCC concern, we further consider location-fixed effects and perform county-fixed effects estimates to assess the impacts of individual characteristics on respondents’ GWCC concern with the following equation:

(2)$${\rm GWC}{\rm C}_{ijk} = {\rm \beta} _0 + {\rm \beta} _1{\rm IC}^{\prime}_{ijk} + {\rm \beta} _2{\rm Count}{\rm y}_{\,jk} + {\rm e}_{ijk}$$

As discussed above, the CGPPS survey interviewed representative respondents from 353 counties across China. By including location-fixed effects, defined by the county dummies County_jk, Equation 2 directly controls for the average differences across these counties, reduces the threat of omitted variable bias, and therefore can obtain better estimations of the impacts of individual characteristics on GWCC concern.

Based on Equations 1 and 2, we perform four separate regressions with different model specifications. In Model 1, GWCC concern is regressed only on respondents’ individual-level social base variables – i.e. Gender_female, Age, Education, Income, Urban residence and CCP membership. In Model 2, we add the Post-materialist values variable to Model 1 to examine how personal post-materialism conditions respondents’ concern about GWCC. In Model 3, we introduce the regional-level variables to Model 2. The four provincial-level variables include physical vulnerability (Coastal region), carbon economic dependency (Carbon-intensive industries’ share of GDP), economic affluence (GDP per capita) and population concentration (Population density). In Model 4, we perform county-fixed effects estimates.

Since GWCC concern was recorded in 11 scales running from 0 to 10, we consider the dependent variable as a continuous variable and thus opt for OLS regression. The results of the four OLS models are reported in Table 4. We include all the models in a single table to allow comparison across different specifications. As survey data tend to have heteroscedasticity issues, we use a robust estimation of the standard errors in all regressions. Post-estimation diagnostic tests for multicollinearity indicate few multicollinearities among the independent variables (all VIFs < 3Footnote ⁵⁹).

Table 4:

Determinants of GWCC Concern

Notes:

*** p < 0.01, ** p < 0.05, * p < 0.1. Robust standard errors in parentheses.

In Model 1, the results show that two social base variables, Gender_female and Age, are significantly correlated with individuals’ GWCC concern: Gender-female shows a positive effect, indicating that women tend to have higher GWCC concern than men (β = 0.280, p < 0.01). Age shows a negative association with GWCC concern, suggesting that relative to older people, younger Chinese citizens are more concerned about GWCC (β = −0.015, p < 0.01). Both effects are consistent with our expectations. However, Education does not have a significant influence on individuals’ GWCC concern. With respect to other social base variables, Income (household per capita income), Urban residence (urban resident status), and CCP membership do not exert significant effect on Chinese GWCC concern.

In Model 2, we add the variable of Post-materialist values to the first model. Consistent with our expectation, the level of GWCC concern is positively correlated with post-materialism (β = 0.254, p < 0.01), indicating that Chinese citizens with stronger Post-materialist values are more concerned about GWCC. More importantly, adding Post-materialist values substantially increases the R-square from 0.013 in Model 1 to 0.096 in Model 2. This suggests that post-materialist values are an important driver for Chinese people's GWCC concern and far more influential than sociodemographic variables. The effects of Gender_female and Age remain statistically significant in the model, while all other social base variables do not show significant influence on respondents’ GWCC concern.

In Model 3, we add the regional-level characteristics and regress respondents’ GWCC concern with all individual-level and regional-level variables. Consistent with the bivariate correlation analysis earlier, the results show that individual GWCC concern is negatively and significantly shaped by regional Carbon-intensive industries’ share of GDP (β = −0.015, p < 0.1). This finding is supportive of our expectation, suggesting that residents located in provinces with more economic dependency on fossil fuel-burning and greenhouse gas-producing industries tend to have less GWCC concern than residents in provinces with lower carbon economic dependency. While the results show that other regional-level variables, including Coastal region, GDP per capita and Population density, have non-significant impacts on GWCC concern, the three individual-level variables, Gender_female, Age and Post-materialist values, which were found to be significant predictors for GWCC concern in previous models, remain statistically significant and influential in Model 3.

In Model 4, the county-fixed effects estimates of individual characteristics demonstrate very similar results to previous estimations. The R-squared statistic substantially increases from 0.096 in Model 2 to 0.210 in Model 4, suggesting that the location-fixed effects account for more than half (54.29 per cent) of the explained variations in Chinese citizens’ GWCC concern. Consistent with previous estimations without the county-fixed effects, the three individual-level variables – Gender_female, Age and Post-materialist values, maintain statistically significant effects on Chinese GWCC concern.

Taken all together, the regression results with different specifications shown in Table 4 indicate that Chinese GWCC concern is affected by both individual-level and regional-level variables. Across the models, three individual-level variables, Gender_female, Age and Post-materialist values, are significantly and consistently correlated with individual Chinese GWCC concern. Under the four different specifications, the coefficients of these three variables are all statistically significant, and the signs all point in the same direction. This pattern indicates that there is a common social base and a post-materialist values base underlying the Chinese public's GWCC concern: compared to their counterparts, Chinese women and younger people with stronger post-materialism tend to be more concerned about GWCC. In addition, the regression results also suggest that regional economic dependency on carbon-intensive industries exerts an inhibiting effect on local residents’ GWCC concerns – that is, residents in provinces that depend more on carbon-intensive and fuel-burning industries tend to be less concerned about GWCC than residents in provinces where carbon-intensive industries account for a smaller percentage of the province's GDP. We discuss the key implications of our findings in the next section.