The mass public’s science literacy and co-production during the COVID-19 pandemic: empirical evidence from 140 … – Nature.com

Baseline regression

Table 2 reports the baseline regression results of the influence of the publics science literacy level on co-production in the fight against COVID-19. With the test logic of econometrics starting from the general to the specific cases, a series of control variables were gradually included in the regression model.

Model (1) only included the core explanatory variables without the addition of any control variables, and the science literacy was significantly positive at 1%. In model (2), six control variables at the objective level of the city were added based on (1), and the science literacy was still significantly positive at 1%. On the basis of (2), model (3) further included three subjective control variablessuch as the publics government trust in the modeland the science literacy was still significantly positive at 1%, consistent with the results of the previous two regression steps. The determination coefficient increased from 0.38 to 0.59, and the fitting degree was thus improved. This indicated that the higher the level of the citys science literacy, the more the co-production against the pandemic. The coefficient of science literacy gradually increased from 0.117 to 0.142, indicating that science literacys influence was increasingly apparent. Model (3) demonstrates that every 1% increase in the publics science literacy can increase the per capita search volume of COVID-19-related keywords by the public by 14.2%, that is, public co-production against the pandemic increased by 14.2%, which verifies H1.

To further address the potential endogeneity problem in the model, a 2SLS model was used to accurately estimate the impact of public science literacy on the public co-production, with the ratio of urban R&D personnel to the annual average population in 2017 as the instrumental variable.

As demonstrated in Table 3, (1) and (2) reported the results of the two-stage regression with the instrumental variable. The regression results of the first stage (1) indicated that the regression coefficient of the proportion of R&D personnel in the city was significantly positive at the 1% level, which denoted that the higher the ratio of urban R&D personnel in the city, the higher the publics science literacy in the city. The correlation hypothesis of the instrumental variable is valid. Meanwhile, the partial R2 is 0.32, and the F-statistic of the significance test is 32.68. The instrumental variable has strong explanatory power. The results of the second stage (2) regression demonstrated that after addressing the endogeneity problem, the positive influence of the publics science literacy on co-production was still significantly positive at 1%. Specifically, with the increase of science literacy by 1%, public co-production increased by 42.5%, which was about three times that of the baseline regression result, which indicated that the promoting effect of the publics science literacy on co-production during COVID-19 may be underestimated due to the endogeneity problem. It was verified that science literacy contributes to promoting co-production against the pandemic. Thus, H1 is supported.

To test the moderating effect of regional educational levels, the proportion of urban secondary school students (Edu_c) was used to measure educational level. The intersection term of regional education level and science literacy was added into the regression model, along with a series of control variables. As illustrated in Table 4, no control variables were added to model (1); only objective control variables were added to model (2), and subjective control variables were further added to model (3). The results demonstrated that the coefficient of the intersection term gradually increased from 0.471 to 1.049, and the significance level gradually rose. The intersection term of model (3) was significantly positive at the 5% level, and the determination coefficient was 0.69, which was better than that of model (3). The coefficient of the interaction term gradually increased from 0.038 to 0.046, and the significance level gradually increased; the determination coefficient increased from 0.51 to 0.62, and the degree of fitting was improved. Model (3) illustrated that the intersection term was significantly positive at the 5% level.

Similar to the approach for testing the moderating effect of regional education level, the number of discredited people (Capacity_c) of the city was used to measure the local government capacity. The greater the number, the worse the local government capacity. As demonstrated in models (4), (5), and (6), the coefficient of the interaction term gradually decreased from 0.009 to 0.020, and the significance level gradually became higher. The interaction term of the model (6) was significantly negative at 1% level.

Figure 3 shows the separate plotting of the moderating effects of regional education level and local government capacity on the publics scientific literacy and co-production in the fight against COVID-19. In the left graph, when Edu_c is greater than 0, the marginal effect is significantly positive within the 95% confidence interval. This means that the marginal effect of the publics science literacy on co-production of fight against COVID-19 gradually increases with the increase of the proportion of the number of students in the city. In the figure on the right, when Capacity_c is <11.77 (The natural logarithmic value of 129,000 is about 11.7, so the number of discredited people at the provincial level in the city is 129,000), the marginal effect is significantly negative within the 95% confidence interval. This means that when the number of provincial-level discredited people in the city is less than 129,000, the marginal effect of the publics science literacy on co-production in the fight against the pandemic gradually increases as the number of discredited people decreases.

Controls were applied for GRP, income level, science and technology level, number of foundations, government network transparency, digital government development level, publics government trust, social trust, and social justice. The dashed line was at a marginal effect of zero. Full regression estimates are provided in Table 4 models with 95% confidence interval.

Both the regression results and the moderating effect graphs indicate that the level of urban education and the local government capacity have a positive impact on the effectiveness of the publics science literacy in promoting co-production fight against COVID-19, supporting hypothesis H2 and H3.

Whether the baseline regression results are affected by sample selection needs to be further tested. As demonstrated in Table 5, owing to a small number of outliers in the explained variables, the explained variables in models (1) and (2) were, respectively, treated with bilateral tail reduction and bilateral censoring at the 5% quantile to avoid the deviation of coefficient estimation, and all control variables were added to perform regression estimation. Additionally, this study further replaced the data and the publics science literacy at the city level with those at the provincial level and added all control variables, as demonstrated in the model (3). Clearly, the coefficients of science literacy are all significantly positive at 1% level, and the baseline regression results are still robust.

The results of the baseline regression are the embodiment of the total effect, and the unique properties of different stages and regions may affect the manifestation of science literacy. This section analyzes the heterogeneity of the stage of the pandemic, geographical location, and city size in terms of the two dimensions of time and space.

With the continuous development of the pandemic situation, the external conditions and the publics willingness in co-production may change. To further explore the dynamic changes of the publics anti-pandemic efforts, this section studies separate regression testing across all stages - including Stage I: Swift Response to the Public Health Emergency (27 December 201919 January 2020), Stage II: Initial Progress in Containing the Virus (20 January20 February 2020), Stage III: Newly Confirmed Domestic Cases on the Chinese Mainland Drop to Single Digits (21 February17 March 2020), and Stage IV: Wuhan and Hubei-An Initial Victory in a Critical Battle (18 March28 April 2020); various models that joined all control variables.

As demonstrated in Table 6, the science literacy coefficients of stages I, II, III, and IV were 0.160, 0.132, 0.152, and 0.160, respectively; they were all significantly positive at 1% level, which indicated that the publics science literacy to the pandemic was effective in all stages. The science literacy coefficient first decreased and then increased. The science literacy coefficient was the same in stages I and IV, and the promoting effect of science literacy on the co-production was relatively obvious, which indicated that the promoting effect of the publics science literacy in different stages was different. In general, the promoting effect of science literacy was statistically significant in the whole process of the fight against the pandemic in each stage, which further verified the correctness of H1.

China is a country with vast territory, and heterogeneity in different regions will affect the publics willingness and cost of co-production in response to the pandemic. The sample cities, based on different characteristics in different geographical locations, were classified into three subregions: eastern, central, and western regions. Specifically, the eastern region includes cities in Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan; the central region includes cities in Shanxi, Jilin, Heilongjiang, Henan, Hubei, Hunan, and Anhui; the western region includes cities in Inner Mongolia, Chongqing, Sichuan, Guangxi, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang. Thereafter, subsample regression was performed. All control variables were added to each model.

As demonstrated in Table 7, models (1), (2), and (3) reflect the differences in the publics anti-pandemic effort based on scientific knowledge in the cities across regions. The science literacy coefficients of cities in eastern, central, and western regions were 0.141, 0.250, and 0.193, respectively, and all of them were significantly positive at 1%, which further supported H1. The estimated coefficient of 0.141 in the eastern region was smaller than that in the central and western regions. It is reasonable to suspect that the promoting effect of the publics science literacy is smaller in the eastern region and relatively larger in the central and western regions.

To answer the aforementioned questions, we further explore the differences between regions. By adding the intersection terms of regional dummy variables and science literacy (E_MS_L, E_WS_L, and E_MWS_L), the difference test of regression coefficient between eastern and central, eastern and western, and eastern and central and western regions, was conducted. The results demonstrated that the coefficients of the interaction terms were all significantly positive, which indicated that a statistically significant difference existed in the coefficient of science literacy between the eastern region and other regions, and the science literacy in the central and western regions played a more evident role in the promotion of the public co-production against the pandemic.

The size of a city affects the difficulty of urban governance and challenges the level of governmental governance. Unlike small and medium-sized cities, large cities are more difficult to govern due to their large population, complicated public affairs, and diverse service demands, and various problems will become more prominent. On the contrary, larger cities have stronger incentives to innovate management models with refined management, improved institutional norms, diversified technical means, and higher enthusiasm of the public to participate in urban governance (Zou and Zhao, 2022). Therefore, the scale of the city may also affect the effectiveness of the role of the publics science literacy in co-production, and still the role of science literacy in the promotion of co-production in those cities is not known. In China, the urban hierarchy is relatively complex, mainly consisting of municipalities directly under the central government, provincial capital cities, sub-provincial cities, prefectural level cities and county-level cities. In comparison to most prefectural level cities, municipalities directly under the central government, provincial capital cities, and sub-provincial cities possess unique advantages in terms of economy, politics, culture, and population. In this paper, they are referred to as large cities, including Beijing, Tianjin, Shanghai, Chongqing, Dalian, Qingdao, Ningbo, Xiamen, Shenzhen, Shijiazhuang, Shenyang, Nanjing, Hangzhou, Fuzhou, Jinan, Guangzhou, Haikou, Taiyuan, Changchun, Harbin, Zhengzhou, Wuhan, Changsha, Hefei, Nanchang, Hohhot, Chengdu, Nanning, Guizhou, Kunming, Xian, Lanzhou, Xining, Yinchuan, Urumqi, Lhasa, a total of 36 cities, and all the rest cities are considered non-large cities. Therefore, this paper divided the sample into two subsamples according to aforementioned categorization; large city, and non-large city; and conducted the subsample regression, after adding all the control variables.

As demonstrated in Table 8, models (1) and (2) reflect the differences in the publics co-production based on science literacy in cities of different scales. The science literacy coefficient of large cities was 0.082, and the significance level was 5%. The science literacy coefficient of non-large cities was 0.260, with a significance level of 1%, which also supported H1. The estimated coefficient of 0.082 for large cities was about one-third of that for non-large cities.

To further verify the statistical significance of the difference in the promotion effect of science literacy, the intersection term (L_CityS_L) was added to the model to test the difference. The results demonstrated that the coefficient of the interaction term was significantly positive, indicating that the promoting effect of science literacy on co-production against COVID-19 was stronger in non-large cities but weaker in large cities.

The above results provide us with many interesting findings. The publics science literacy plays an important role in promoting co-production in the fight against pandemic, and there are significant differences in the performance of the effect of this role in different temporal and spatial dimensions. We also found that regional education level and local government capacity can positively moderate the relationship between the two, verifying the previous hypotheses H1, H2, and H3.

Continued here:

The mass public's science literacy and co-production during the COVID-19 pandemic: empirical evidence from 140 ... - Nature.com