*3.1. Participants and Procedure*

The data for the present study were drawn from a large-scale research project conducted between 11 and 18 February 2020 by the School of Public Administration of Hohai University that investigated the psychosocial impact of COVID-19 on the public in China. The project distributed questionnaires via the Internet and conducted a survey using quota sampling. It collected 8000 questionnaires in 13 prefecture-level cities in Jiangsu province and another 30 provincial capitals in mainland China. Before beginning the survey, the participants were informed that their participation was voluntary and could be discontinued at any time. They were also informed that no personal information would be collected; their survey responses would remain anonymous and have no bearing on their academic standing. The project was approved by the Human Subjects Ethics Sub-Committee at the university with which the corresponding author is affiliated.

Originally, 8138 people completed the survey. After eliminating the survey responses of participants younger than 18 and questionnaires with many missing values, a total of 7092 valid samples were ultimately obtained. Infection rate was calculated based on the numbers of confirmed cases published by local health committees and the official 2020 population data for the cities surveyed.

#### *3.2. Measures*

Infection rate: 'Infection rate' refers to the number of confirmed cases over the past year per 100,000 population. We collected the number of confirmed COVID-19 cases announced by the health commission of each surveyed city between February 2020 and February 2021 and the permanent population data in the statistical yearbooks of each city for 2020. We then calculated each city's infection rate based on these data.

Government response: 'Government response' refers to the actions taken by the government to advise or mandate that the public and private sectors take certain measures to restrict the severity or spread of the pandemic. Based on the 'Level I Response Measures for Pneumonia Outbreak in Response to Novel Coronavirus Infection' issued by each province, the research team compiled a list of 20 common prevention and control measures (see Table 1). The respondents were asked in the questionnaire whether their local governments had adopted these measures. If a measure had been adopted, the response was recorded as '1 and '0 otherwise. The sum was divided by 20 to calculate the government response index.


**Table 1.** Measures of government response.

Risk perception: Public conceptions of risk are complex and influenced by qualitative factors [50], including the extent to which a given risk is viewed as fatal, uncontrollable, and unknown. We adopted the measurement method of Liu et al. [51] and measured these factors using three items rated on a 5-point Likert-type scale. A sample item is 'How seriously do you take the COVID-19 epidemic in mainland China?' We conducted factor analysis of the results to generate a three-item risk perception scale. The Cronbach's alpha coefficient for the three items on this scale was 0.764, indicating acceptable internal consistency. The response distribution was linearly transformed to range from 0 to 100, with 100 indicating the highest level of risk perception.

PAR adoption: Four items from the Guidelines for the Public's Protective Behaviour for COVID-19, produced by the Chinese Centre for Disease Control and Prevention [52], were adopted to measure the protective behaviours undertaken by the respondents [22]. A sample item is, 'Have you taken the recommended protective action of wearing a mask when going out in the past two weeks?' For each of the recommended protective behaviours, the respondents indicated whether they had complied or not complied. If the respondent had adopted all four recommended protective behaviours over the preceding two weeks, he or she was considered to be a good adopter of the recommended protective behaviour and assigned a value of 4. If the respondent had not adopted all four recommended protective behaviours over the preceding two weeks, he or she was assigned a value of 0.

We controlled for the demographic characteristics of gender, age, household registration, years of schooling, health status, urbanisation rate, and region. The descriptive statistics for each variable are presented in Table 2.


**Table 2.** Descriptive statistics for the main variables.
