The more actions, the higher the performance evaluation? Evidence from the crisis management of COVID-19 in China

Abstract

Governments around the world have taken measures in fighting against COVID-19, but how the government's response affects the public evaluation of government performance in crisis remains to be examined. The study investigated how government actions, the public's trust in government, risk perception, and negative emotions influenced the public evaluation of government performance based on a primary survey. The findings showed that: (1) government actions and public evaluation of government performance in crisis are significantly and positively associated; (2) trust in government can positively lead to the public evaluation of government performance in crisis; (3) risk perception and negative emotions can negatively moderate and attenuate the previous positive relationship. The research revealed that the public evaluation of government performance is formed by the government actions through mediation effect of trust in government and also moderation effect of risk perception and emotional factors. Thus, this study innovatively developed a moderated mediation model of public perceived government performance which may serve as a basis for improving public evaluation while dealing with the crisis in other countries and regions.

1. Introduction

In December 2019, a pandemic outbroke and became a global pandemic in March 2020. On February 12, 2020, the World Health Organization (WHO) officially named the disease, caused by the novel coronavirus, as coronavirus disease 2019 (COVID-19). The COVID-19 pandemic is the most serious public health emergency since the SARS outbreak in China and has become a global health threat [1]. WHO has reported that as of October 4, 2020, the number of confirmed COVID-19 cases has increased to 34.8 million with 1,030,738 deaths worldwide [2]. The COVID-19 remains grim at the time of this writing. The WHO Emergency Committee on COVID-19 anticipates a lengthy duration of the COVID-19 pandemic, noting the importance of sustained community, national, regional, and global response efforts [3]. Indeed, the COVID-19 pandemic has evolved from a public health event into a transboundary crisis [4] and has spread to economic, political, and social dimensions.

Considered to be “dynamic forces in ongoing processes of legitimization, delegitimization, and relegitimization” [5], the crisis was defined as “a situation in which there is a perceived threat against the core values or life-sustaining functions of a social system that requires urgent remedial action in uncertain circumstances” [6]. Major crises, especially transboundary crises, impact and constrain public administration [4]. Thus, crisis management is part of a government's responsibility, and the public expects the government to take effective measures in an emergency [[7], [8], [9]] [[7], [8], [9]] [[7], [8], [9]]. Since the outbreak, governments and health systems around the globe have struggled with COVID-19, and have taken different measures in responding to the disease [10]. In China, where the pandemic was first reported, governments took a range of forceful measures. During the initial phase of the outbreak, the local government in Wuhan announced the suspension of public transportation on January 23, 2020, shutting down airports, railway stations, and other public places in the city to prevent further disease transmission. In hindsight, these measures were both strong and timely.

Do government actions necessarily lead to higher public evaluation of government performance in crisis? Performance evaluations of government mostly use objective evaluation methods, including detailed analyses and scientific measurements [[11], [12], [13]] [[11], [12], [13]] [[11], [12], [13]]. However, in the context of crisis, the link between government actions and the public evaluation of the government performance is both a question of objective reality and a matter of subjective factors [[14], [15], [16]] [[14], [15], [16]] [[14], [15], [16]]. This study aims to better understand the relationship between government actions and public evaluation of government performance. Therefore, a social psychology paradigm is followed, which is suitable for establishing the connection between objective social reality and subjective social psychology, and primary survey data collected at the peak of the COVID-19 pandemic in China were employed. This research developed and tested a moderated mediation model of public perception and emotion towards the public evaluation of government performance in crises, which contributed suggestions for the governments to improve public government evaluation while dealing with the crisis. Besides, this study emphasizes that except for the prevention and control measures the government also should shape public risk perception and sentiment to improve the public performance evaluation.

The remainder of this paper is organized as follows. The following section reviews the literature and formulates our research hypotheses. Section 3 describes the data sampling and measures. Section 4 reports the main result. Section 5 discusses the theoretical value and policy implications. Section 7 concludes the paper.

2. Literature and hypotheses

2.1. Government actions and perceived performance

The relationship between government actions and the public evaluation of the government performance represents traditional concerns about measuring public sector performance, namely, managing for results [17,18]. This instrumentalism orientation, in place since New Public Management (NPM) emerged, emphasizes the correspondence between government actions and performance [19], which reveals specific actions will lead to specific results [20]. However, the prior study reveals that laypeople have insufficient expertise to objectively assess government performance [14]. Therefore, objective evaluations, composed of detailed indicators and scientific measurements based on expertise, have been a primary approach in government performance management [11,21]. While it is impossible to complete a detailed analysis in the context of an urgent crisis. Even if scientific and accurate performance evaluations were available, it would be impossible to implement and spread them effectively in a crisis context [22]. Indeed, a crisis changes the main context of performance evaluation and may change the linear relationship between government action and public performance evaluation.

The public perception of government performance could influence the government's decisions during the crisis response and the crisis learning after the crisis [15,23]. In contemporary low-risk-tolerant societies, the public is increasingly concerned about risks and expects governments to take effective actions in crises and disasters [4]. Government actions expected by the public during a crisis mainly refer to decision making, operations, information sharing, and communications [24]. Requirements for outcome-based performance management increasingly refer to government actions that achieve the provision of public services [25]. In the early stages of a crisis, the more the government acts and the faster it responds means better-coping activities and higher performance [26]. Thus, there appears to be a positive correlation between government actions and the public perception of government performance in the crisis context. Therefore, we hypothesize that:

H1

Government actions positively affect the public evaluation of crisis management performance.

2.2. Mediating effect of trust in government

Complex mechanisms shaped the perception of government performance, and government actions are not the only determining factor [27,28]. Prior research has found that trust in government is positively related to the public perception of government performance and public satisfaction [29,30]. The public perception of trust in government is an ingrained view or outlook, based on past or historical experience, culture, and ideology, rather than a short-term impression of government [27]. Another may provide evidence for the reverse causation: first, members of the public who perceive higher levels of government performance are more inclined to trust the government; second, members of the public who trust the government are more likely to be satisfied with government actions and policies [31].

In the crisis context, public trust in government becomes more important [15,23]. Research has also found that the pandemic itself (and the following lockdown) both relies upon and may change the extent to which the public trust in government or other organizations [32]. Consistent with existing research results, societal trust consistently rises following natural disasters, perhaps due to the shared need for society to work together to respond to disasters [33]. Research from the United States during the H1N1 also revealed that people largely trust public health staff [34,35]. In this study, we hypothesized that a mediating effect may exist among government action, trust in government, and the public evaluation of government performance. Therefore, we hypothesize that:

H2

Trust in government is a significant mediating variable in the relationship between government actions and the public evaluation of government performance during a crisis.

2.3. Moderating effect of negative emotions and risk perception

The public evaluation of government performance is a question of objective reality and a matter of public perceptions and sentiments [16]. Because the crisis results in an immediate risk to public mental and physical health and social relationships [36,37], government actions and their outcomes cannot fully determine the public perceptions of performance. The prior study reveals that the public evaluation of government performance is determined by the risk perceptions and emotions of stakeholders and the publ ic [17]. Individual psychological factors, such as negative emotions, also have an impact on public perception of government performance [38].

Negative emotions, such as blame, frustration, despair, anger, and schadenfreude, seem particularly salient in crisis management [39]. According to Attribution Theory, the public is motivated to search for the causes of unexpected and negative events, and these attributions of responsibility can invoke negative emotions and reactions [40]. This may be why the public generally characterizes public dialogue about historic crisis events using negative emotional references [41]. Research from China indicates the public has experienced significant anxiety and depression during the COVID-19 pandemic [42,43]. The breeding and spreading of negative emotions in public emergencies pose significant challenges for crisis governance [44]. Research in public relations and crisis communication has revealed that emotional reactions to crises can affect how they are perceived and can influence social evaluations of government [[45], [46], [47]] [[45], [46], [47]] [[45], [46], [47]]. In particular, research indicates that the higher the perceived negative emotions are during the crises, the more difficult it is for the government to repair its relationship with the public [38]. Crises do appear to degrade the public's evaluation of government performance, partly due to the effects of negative emotions. Negative emotions impact cognition and perception, which can occur through a mitigation process [48]. Therefore, we hypothesize that:

H3a

Negative emotions are significant negative moderating variables in the relationship between government actions and the public evaluation of government performance.

H3b

Negative emotions are significant negative moderating variables in the relationship between government actions and trust in government.

H3c

Negative emotions are significant negative moderating variables in the relationship between trust in government and the public evaluation of government performance.

There are two fundamental ways for people to perceive risk: risk as analysis and risk as feeling [49]. Risk perceptions are complicated psychological and cognitive mechanisms and generate profound actions and consequences for crisis management [23]. Recent psychological research has identified that emotions are key ingredients in risk perception [50], with negative emotions leading to higher risk perception [51,52]. Further, the relationship between risk and trust is complex. In general, many empirical studies have shown that trust significantly affects risk perception [53,54], while some studies have also found that the public levels of risk perception and attitudes determine their trust in institutions such as the government [[55], [56], [57]]. Meanwhile, research has found a triangular relationship between trust, emotions, and risk perception [55]. Based on the relationship between negative emotions, risk perception, and trust in government, we hypothesize that:

H4a

Risk perception is a significant negative moderating variable in the relationship between government actions and the public evaluation of government performance.

H4b

Risk perception is a significant negative moderating variable in the relationship between government actions and trust in government.

H4c

Risk perception is a significant negative moderating variable in the relationship between trust in government and the public evaluation of government performance.

Fig. 1 shows the public evaluation of the government performance model in which public evaluation of government performance is a function of government actions, trust in government, negative emotion, and risk perception. Specifically, trust in government plays a mediating effect in the relationship between the independent and dependent variables, while the negative emotion and risk perception have a moderating effect on the path among the independent, dependent, and mediation variables.

Fig. 1.

The conceptual model.

3. Method

3.1. Data and sampling

The data for this research were collected online in China during the 2020 COVID-19 outbreak. The online survey began on January 26, 2020, and lasted for 5 days, which was a critical time for COVID-19 in China. Because the survey was done at the start of the pandemic, the data are very valuable for exploring the research topic. Due to the shutdown caused by COVID-19, we implemented the online questionnaire survey through Wenjuanxing, which is the most reputable professional data collection company and an online survey platform in China. This was the most feasible and effective way to gather data.

A total of 1089 questionnaires were distributed nationwide, and 996 questionnaires were completed, with a response rate of 91.46%. Table 1 shows that the sample was demographically diverse by gender, age, and educational attainment. 41.77% of the respondents were male, and 58.23% were female. Educational attainment was categorized into three levels: 9.24% of the respondents had a lower-level education (primary school or junior middle school educated); 20.28% had a middle-level education (high school or junior college-educated), and 70.48% had a higher-level education (college or above). A comparison of the sample's demographic characteristics to the corresponding census data indicates that the sample over-represented females, people with higher education, and middle-aged residents.

Table 1.

Description of survey sample and control variables (N = 996).

Variable	Item	Frequency	Percent
Gender	Female	580	58.23
	Male	416	41.77
Education	Lower-level (primary school or junior middle school educated)	92	9.24
	Middle-level (high school or junior college educated)	202	20.28
	Higher-level (obtained a college or above).	702	70.48
Age	18–24 years old	169	16.97
	25–30 years old	297	29.82
	31–40 years old	221	22.19
	41–50 years old	173	17.37
	51–60 years old	85	8.53
	Above 60 years old	51	5.12
Total		996	100

3.2. Measures

3.2.1. Dependent variable

The dependent variable in this study was the public evaluation of government performance. Central and local governments in China are organized in a bureaucratic hierarchy. Studies have revealed that trust in central and local government in China have significant disparities [23,58]. The responsibilities of central and local government differ, so the public perception or evaluation of government performance was divided into four dimensions according to hierarchy: the central government, provincial government, municipal government, and county government of the respondent's current location. The dependent variable was calculated based on the mean value of perceived government performance across these four government levels.

The public evaluation of government performance was assessed by the question “To what extent do you think the following organizations or groups have played a positive role in preventing the pandemic for yourself and your family?” [59]. Each item was responded to a ten-point Likert scale (1-extremely low, 10-extremely high). The Cronbach's alpha result was 0.91, indicating very good reliability that the four variables could be treated as one factor.

3.2.2. Independent variable

Government Actions: In this research, summarized variables were used to measure the respondent's judgment of government actions. Specifically, the participants were asked a multiple-choice question: Have you ever witnessed the following pandemic prevention measures adopted by government departments in your village or community? The options were summarized according to the Law of the People's Republic of China on the Prevention and Treatment of Infections Diseases and response measures occurred during COVID-19 in China, as follows: (1) Put up posters related to COVID-19. (2) Disseminate information on pandemic prevention. (3) Check and quarantine people returning from pandemic areas. (4) Distribute pandemic prevention materials. (5) Provide disinfection for public places. (6) Lockdown public places. (7) Take your temperature in public places. (8) Other targeted measures.

Each option was recorded as a binary variable. If the respondent selected yes, the question was marked as “1”; otherwise, the option was marked as “0”. The value of the perceived government action variable was the total score for all eight items, ranging from 0 to 8. Higher scores indicate a higher level of government actions among citizens perceived and vice versa.

3.2.3. Mediation variable

The mediation variable was trust in the government. A scoring method was broadly used as the measurement in assessing government trust or political trust [60]. We adopted the single question method in this research. The original question was, “On the whole, I trust the government.” A five-point Likert scale was adopted for answers, ranging from strongly agree, somewhat agree, neither agree nor disagree, somewhat disagree to strongly disagree, which were scored as 5, 4, 3, 2, and 1, respectively. The higher the score, the more trust respondents invested in government.

3.2.4. Moderation variable

Negative emotions: Negative emotions were operationalized as three variables: the degree of anxiety, fear, and worry [59]. Three original questions were asked to measure each respondent's negative emotions: To what extent are you anxious/fearful/worried about COVID-19? The answers were fitted onto a five-point Likert scale, ranging from 1 (not at all) to 5 (very high), representing an increased degree of negative emotions. The Cronbach's alpha result was 0.88, indicating good reliability that the three variables could be treated as one factor.

Risk perception: Participants were asked to indicate the possibility of being infected with COVID-19, using a scale ranging from 1 for “out of the question” to 4 for “very likely.” The original question was: “How likely do you think you are to be infected with COVID-19?”

3.2.5. Control variables

The control variables included gender, age, and education. Gender was a dummy variable with value 1 as “male” and the value 0 as “female”. Education was a categorical variable coded as 1, 2, or 3, representing “primary school or middle school educated”, “high school educated or equivalent”, or “college or above”, respectively. As a categorical variable, age was coded from 1 to 6 for “18–24 years old,” “25–30 years old,” “31–40 years old,” “41–50 years old,” “41–50 years old,” and “above 60 years old,” respectively.

3.3. Data analysis

In this study, we applied the conditional process analysis advocated by Hayes to test the mediation and moderation model, and this method is very suitable for revealing the complex relationship between two variables due to the influence of many other factors [61]. In particular, based on multiple regression methods, a specialized form of moderated mediation, known as conditional process analysis modeling, was conducted in SPSS using the Hayes’ ordinary least squares regression-based plug-in component PROCESS. PROCESS provides a convenient method to implement conditional process analyses and to generate results that are similar to models that would be estimated using SEM programs such as MPlus and LISREL [61,62].

Specifically, we used the PROCESS macro for SPSS (V.20.0.0) Model 16 to estimate the conditional indirect effect of the independent variable “government actions,” through the mediator “trust in government,” on the outcome variable “public evaluation of government performance,” with “risk perception” and “negative emotions” included as moderators [61]. This enabled the moderating effect of risk perception and negative emotions to be tested on all three paths simultaneously: ‘high’ risk perception or negative emotions (mean plus one standard deviation), ‘mean’ (mean), and ‘low’ risk perception or negative emotions (mean minus one standard deviation) (as illustrated in Table 4, Table 5). In this analysis, age, education, and gender were included as controls. The conditional process model generated (bias-corrected) 95% confidence intervals for the estimated indirect effects at different values of the moderator variable. Conditional process analysis allowed the results to be probed at various point estimates by generating by bootstrapping with 5000 re-samples [63].

Table 4.

Conditional process analysis showing direct effects (government actions -> the public evaluation of government performance).

	Direct Effect	SE	t	P	LLCI	ULCI
Low negative emotions	.194	.045	4.326	.000	.106	.283
Mean negative emotions	.140	.031	4.507	.000	.079	.201
High negative emotions	.086	.042	2.060	.040	.004	.168

Note: Low (mean minus one standard deviation); High (mean plus one standard deviation).

CI=95% confidence interval for the indirect effect: if CI does not include zero, the indirect effect is considered statistically significant and is displayed in bold.

Table 5.

Conditional process analysis showing indirect effects.

Negative Emotions	Risk Perception	Indirect Effect	Boot SE	Boot LLCI	Boot ULCI
Low negative emotions	Low risk Perception	.191	.034	.131	.264
Low negative emotions	Mean risk Perception	.157	.028	.107	.215
Low negative emotions	High-risk Perception	.122	.025	.076	.174
Mean negative emotions	Low risk Perception	.137	.021	.101	.182
Mean negative emotions	Mean risk Perception	.102	.014	.077	.132
Mean negative emotions	High-risk Perception	.066	.015	.039	.097
High negative emotions	Low risk Perception	.081	.023	.042	.131
High negative emotions	Mean risk Perception	.045	.017	.015	.081
High negative emotions	High-risk Perception	.010	.018	-.026	.045

Note: Low (mean minus one standard deviation); High (mean plus one standard deviation).

CI=95% confidence interval for the indirect effect: if CI does not include zero, the indirect effect is considered statistically significant and is displayed in bold.

4. Results

4.1. Descriptive statistics

Descriptive statistics and correlations for all variables are shown in Table 2 . The mean score of the public evaluation of government performance (M = 8.25 of 10) was higher than the midpoint (5.0), suggesting that respondents reported a relatively high level of perceived government performance. The respondents also reported a perceived high trust in government (M = 4.22 of 5), risk perception (M = 2.36 of 4), negative emotions (M = 3.89 of 5) as compared with perceived government actions (M = 2.83 of 8).

Table 2.

Descriptive statistics of the variables.

Variable	Means	S.D.	Min	Max	1	2	3	4	5
Public evaluation of government performance	8.25	2.01	1	10	1
Government actions	2.83	1.82	1	8	0.235***	1
Trust in government	4.22	1.00	1	5	0.362***	0.308***	1
Risk perception	2.36	0.82	1	4	−0.138***	−0.147***	−0.150***	1
Negative emotions	3.89	0.97	1	5	0.285***	0.021	0.072*	0.111***	1

Note: N = 996. *p < 0.05, **p < 0.01, ***p < 0.001.

Correlations between the variables are also shown in Table 2. These correlations show significant relationships among public government performance evaluation, trust in government, risk perception, negative emotions, and government actions. In contrast, perceived government actions were not significantly correlated with negative emotions. The correlation matrix indicates significant associations in the hypothesized direction between the model variables.

4.2. Conditional process analysis

SPSS version 20.0 and PROCESS version 3.5 were used to test the hypotheses. All variance inflation factors (the diagnostics for multicollinearity) were within the acceptable range. Table 3 lists the results from the conditional process analysis, generated using the PROCESS macro. In the first multiple regression model, we tested whether negative emotions moderated the path from government actions to the public evaluation of government performance (depicted as path i in Fig. 1). The results indicated that government actions had a significant positive association with the public evaluation of government performance (β=0.254; p<0.001). The interaction term (Interaction1, computed as the product of government actions and negative emotions) appeared to have a significant negative relationship on the public evaluation of government performance (β=−0.113; p<0.001), controlling for gender, age, and education. However, the interaction term (Interaction1, computed as the product of government actions and negative emotions) appeared to be non-significant in the final model (Model 5). This indicated that negative emotions did not have a significant moderating effect on the impact of government actions on perceived government performance, which differed from the hypothesis (H3a).

Table 3.

Conditional indirect effect of risk perception and negative emotions on trust in government to public evaluation of government performance.

	Model 1		Model 2		Model 3		Model 4		Model 5
Dependent Variable	Public evaluation of government performance		Public evaluation of government performance		Trust in government		Trust in government		Public evaluation of government performance
Independent Variable	Coefficient	LLCI/ULCI	Coefficient	LLCI/ULCI	Coefficient	LLCI/ULCI	Coefficient	LLCI/ULCI	Coefficient	LLCI/ULCI
constant	8.377***	7.842/8.895	8.479***	7.964/8.995	0.398***	0.148/0.648	0.447***	0.196/0.699	8.160***	7.703/8.616
Gender	−0.107	−0.340/0.132	−0.058	−0.301/0.185	−0.020	−0.139/0.099	−0.027	−0.146/0.092	−0.094	−0.308/0.120
Education	−0.005	−0.100/0.093	−0.027	−0.117/0.062	−0.135***	−0.179/-0.091	−0.137***	−0.181/-0.093	0.064	−0.004/0.157
Age	0.009	−0.007/0.023	0.003	−0.011/0.016	0.024***	0.017/0.031	0.022***	0.015/0.029	−0.009	−0.022/0.003
Government actions	0.254***	0.191/0.318	0.235***	0.169/0.301	0.162***	0.130/0.194	0.157***	0.124/0.189	0.140***	0.079/0.201
Negative emotions	0.601***	0.481/0.720			0.070*	0.009/0.131			0.613***	0.079/0.201
Trust in government									0.629***	0.514/0.744
Risk perception			−0.303***	−0.454/-0.152			−0.112**	−0.186/-0.038	−0.243***	−0.378/-0.108
Interaction1	−0.113***	−0.184/-0.040							−0.056	−0.118/0.005
Interaction2			0.031	−0.047/0.108
Interaction3					0.002	−0.030/0.035			−0.365***	−0.485/-0.245
Interaction4							0.020	−0.018/0.057	−0.267***	−0.395/-0.139
N	996
R	0.150		0.071		0.143		0.147		0.286
F	28.653***		12.526***		27.094***		28.007***		38.908***

Note: *p < 0.05, **p < 0.01, ***p < 0.001.

Interaction1 = Government actions* Negative emotions, Interaction2 = Government actions* Risk perception, Interaction3 = Trust in government * Negative emotions, Interaction4 = Trust in government * Risk perception.

In the second regression analysis model, we tested whether risk perception moderated the path from government actions to the public evaluation of government performance (depicted as path e in Fig. 1). Panel 2 of Table 3 (Model 2) shows that government actions had a significant positive effect on the public evaluation of government performance (β=0.235, p<0.001). However, the interaction between government actions and risk perception appeared to be non-significant, which differed from the hypothesis (H4a).

Similarly, in Panel 3 and 4 in Table 3 (Model 3 and Model 4), we tested whether negative emotions and risk perception moderated the path from government actions to trust in government (depicted as path d and h in Fig. 1). The results indicated that the interaction terms (Interaction 3 and Interaction 4, computed as the product of government actions and negative emotions or risk perception) did not have significant relationships with trust in government, controlling for gender, age, and education. This differed from the hypothesis (H3b and H4b).

Finally, we tested whether risk perception and negative emotions moderated the path from trust in government to the public evaluation of government performance (depicted as path f and j in Fig. 1). The result in Model 5 shows a statistically significant association between government actions and the public evaluation of government performance (β=0.140, p<0.001), consistent with the hypothesis (H1). The outcomes also indicated that trust in government was positively related to the public evaluation of government performance (β=0.629, p<0.001), consistent with the hypothesis (H2). In Model 5, the interaction terms (Interaction 3 and Interaction 4) were statistically significant (β=−0.365 and −0.267, p<0.001), indicating that risk perception and negative emotions had a significant moderating effect on trust in government to the public evaluation of government performance. This result was consistent with the hypotheses (H3c and H4c).

Moreover, we also drew the different pathways operating among the public evaluation of government performance, government actions, and trust in government, using different levels of the respondents’ negative emotions and risk perception. Indirect and direct effects in the conditional process analysis are shown in Table 4, Table 5 , respectively. First, based on the three response levels for negative emotions, Table 4 shows a significant direct effect between government actions and the public evaluation of government performance at any level of negative emotions, after controlling for gender, age, and education. The direct effect of government actions on the public evaluation of government performance through negative emotions is generally positive but decreased with a high degree of negative emotions (low negative emotions, β=0.194; mean negative emotions, β=0.140; and high negative emotions, β=0.086).

The indirect effect of government actions on the public evaluation of government performance through risk perception and negative emotions was generally positive but decreased to a high degree (see Table 5). With respect to high negative emotions and high-risk perceptions, the indirect effect did not significantly vary from zero, as the bias-corrected 95% CI included zero (bootstrapped confidence intervals LLCI/ULCI =-0.026, 0.045).

Fig. 2 and Fig. 3 show the outcome of the interaction between negative emotions or risk perception and trust in government on the public evaluation of government performance, respectively. The slope of the graph in Fig. 2 illustrates that the positive effect of trust in government on the public evaluation of government performance was strongest for the respondents reporting a low level of negative emotion, compared to respondents who reported high levels of emotions about COVID-19. Similarly, Fig. 3 shows the moderating effect of risk perception on the direct effect of trust in government on the public evaluation of government performance. The effect was more positive and stronger when the levels of risk perception were low.

Fig. 2.

The outcome of the interaction between negative emotions or risk perception and trust in government on the public's evaluation of government performance.

Fig. 3.

The outcome of the interaction between negative emotions or risk perception and trust in government on the public's evaluation of government performance.

5. Discussion

The research revealed that the public evaluation of government performance is formed by the government actions through the mediation of trust in government and also moderation of risk perception and emotional factors of specific.

This research revealed that government actions and perceived performance in a crisis are significantly and positively associated. This study's findings add to the evidence that the government needs to actively take measures and respond to the crisis to improve the public evaluation of government performance. Crises are pivotal times for government, and addressing crises is a core responsibility of the government and the public sectors [64]. Crises can erode the government-public relationship, however, crises also provide opportunities to strengthen the government's ties to its stakeholders [39].

Analysis of the moderated mediating model showed that trust in government was strongly and positively associated with the public evaluation of government performance, and that trust in government was a significant mediator variable in the relationship between government actions and public evaluation of government performance. In other words, as perceptions about government actions increased, public trust in government also increased, which increases the public evaluation of government performance. When the public faces a shared external threat, there may be a reflexive increase in their trust in institutions, in part because they have few other options [65], which may mediate the influence that government actions affected on the public evaluation of government performance. Prior studies have found that perceived government performance can lead to trust in government [31,66], while other studies suggest that there should be an reverse causation between the two variables [67,68]. Our findings demonstrate that the link between trust in government and perceived government performance is mediated by other relative factors but not a linear relationship [27,69], which makes one of this study's innovations.

The research results also indicate that risk perception was significantly positively correlated with negative emotions, and risk perception or negative emotion appears to moderate the indirect effect of government action on the public evaluation of government performance through trust in government. When the risk perception or negative emotion was high, the positive effect of trust in government on public perceived government performance was attenuated, which highlights the importance of subjective factors in shaping the public perception of the government performance. Prior research has revealed that the spread of psychological distress especially the negative emotions was pervasive during the pandemic, irrespective of the actual severity of the risk [59].

With data collected during the COVID-19 pandemic from China, these findings provided evidence that the public perception of government performance is influenced by both objective and subjective factors. Thus, in addition to prevention and control measures, the government also should take public emotions and risk perception into consideration when conducting information disclosures and crisis communication [70,71]. These conclusions drawn from authoritarian China have implications for other countries as well since the public perception of government performance during the crisis clearly has a direct impact on the election of leaders in some countries with democratic elections.

6. Conclusion

This article examined the mechanisms involved in the public evaluation of government performance, using a conditional process (moderated mediation) model and survey data collected during the COVID-19 pandemic in China. This result demonstrated that the public's perception of the government's performance is influenced by subjective factors, and further concluded a framework model among government actions, public perception of government performance, trust in government, negative emotions, and risk perception.

Since key variables can only be accurately measured by questionnaires at the beginning of the pandemic, the current study only studies the relationship between government actions and public evaluation of government performance at this stage based on questionnaires. Meanwhile, the measurement of some variables was simplified due to cyber-censorship. Future studies will attempt to measure public opinion of the government using methods other than questionnaire surveys, such as text analysis based on online social media statements, which may also help to extend the study to the entire period of the crisis, not just the early stage.

Declaration of competing interest

The authors declare no competing interest.