3.3 Variable Definitions
In this article, we examine three outcome variables that reflect an individual's political participation. The first variable is the frequency with which the individual uses the internet for the purpose of protecting his/her rights or other people's rights in the past year. The frequency takes values from 1 to 5, meaning never, few times, sometimes, often and always. The second variable is a dummy indicating whether the individual would resort to the internet to protect his/her rights when his/her rights were infringed. Specifically, the variable takes value 1 if the individual would resort to the internet and 0 otherwise. The third variable is a dummy showing whether the individual would support others online when others appeal to the internet to protect their rights when their rights were infringed. Specifically, the variable takes value 1 if the individual would support others online and 0 otherwise.
The independent variable we focus on refers to an individual's level of digital literacy. We build an index based on several variables, which represent individual digital skills and competences. We look at three aspects of individual digital skills and competences. The first aspect covers six basic skills, namely, knowing how to open websites on a computer, knowing how to download and install an application on a smart phone, finding it not difficult to search for and find information needed, verifying first the authenticity of a message when reading a message retransmitted by others on social media, knowing how to operate to express opinions on the internet, and verifying first the safety of the environment before making a payment on the internet. Each of these skills is denoted by a dummy variable.
The second aspect includes two dummy variables: actively protecting one's personal privacy online and worrying about that one's personal privacy will be leaked on the internet.
The third aspect contains two dummy variables relating to the use of social networking applications: having used Wechat and having used QQ. Unfortunately, CGSS 2017 do not give information about the frequencies of using Wechat and QQ. Wechat and QQ are the two most popular social networking tools in current China. People can use these tools to communicate, to share and exchange thoughts, to post words, pictures and videos, to retransmit and comment on messages and articles, and so on. However, CGSS 2017 do not provide information on what kinds of activities the individuals use Wechat and QQ for.
Based on the aforementioned ten variables, we construct a single index of digital literacy by averaging the values of the ten variables following the method applied in previous studies [Hargittai
2005]. Larger values of the index indicate higher levels of individual digital literacy.
Apart from outcome variables and the key independent variable, we also consider some explanatory variables in the empirical analysis to control for possible estimation biases. We include individual characteristics in regressions, namely, gender, age, educational attainment, whether being married or not, number of kids, whether having a local registration (so-called Hukou in China), annual income, health status, whether belonging to a minority group, and whether living in the urban. The gender variable indicates whether the individual is a male. Age is measured in years. Educational attainment variable takes values from 1 to 13, indicating the lowest to the highest educational levels. Annual income is measured in ten thousand. Health status variable takes values from 1 to 5, representing the worst to the best health conditions. Besides, we control for fixed effects at the province level.
To check whether the relationship between digital literacy and political participation varies according to people's characteristics, we further carry out empirical analyses by looking at subgroups of people. First, we would like to see if the relationship between digital literacy and political participation is strengthened when focusing on people who care about news. Specifically, we examine people who read news at least one hour per day. Second, we would like to see if the relationship gets larger when focusing on people who believe in the internet for attracting the attention of the government. Specifically, we concentrate on people who believe that government workers pay attention to the public's thoughts online.
3.4 Descriptive Statistics
In the benchmark sample, there are around 22.6% of respondents sometimes, often or always use the internet for the purpose of protecting his/her or other people's rights in the past year. There are approximately 26.4% of respondents would resort to the internet to protect his/her rights when his/her rights were infringed and approximately 23.2% of respondents would support others online when others appeal to the internet to protect their rights. About the key independent variable, i.e., the index of digital literacy, the average value of the index is 3.56.
Based on the benchmark sample, we summarize in Table
1 independent variables by different values of the dependent variables. We distinguish two subsamples in columns 1 and 2 according to the frequency with which the individual uses the internet for the purpose of protecting his/her or other people's rights. We give in column 1 sample means of the independent variables when the frequency variable is less than 4 and in column 2 those when the frequency variable is equal to or larger than 4. Two subsamples are distinguished in columns 3 and 4 according to whether the individual would resort to the internet to protect his/her rights when his/her rights were infringed. Similarly, two subsamples are distinguished in columns 5 and 6 according to whether the individual would support others online when others appeal to the internet to protect their rights. Finally, figures in column 7 is from the entire benchmark sample.
As we can see, people with higher levels of digital literacy are more likely to frequently use the internet for the purpose of protecting his/her or other people's rights in the past year, more likely to resort to the internet to protect his/her rights when his/her rights were infringed, and more likely to support others online when others appeal to the internet to protect their rights.
Concerning other control variables, it seems that the male, the younger people, the better educated, those who are unmarried, those with fewer kids, those without a local registration, those with higher incomes, those who are healthier, and those living in the urban show higher probabilities in participating in online political activities.
Since descriptive statistics can only give us preliminary results about the relationships between the dependent and the independent variables, it is thus necessary to implement econometric analyses to obtain better understandings.
