This paper presented a very interesting topic. Censorship in China has always drawn people’s attention since in a perspective of universal values censorship repress the free will of people to express themselves. This paper, focusing on the censorship in China, is sure to arouse peoples’ hot discussion and exert its impact on society. It occurs to me that western researchers are doing some ‘real‘ research that can have great impact. They can always grasp the most important and interested topics in the world, look into these topics and yield some interesting and deep insights. I can not help but wonder why researchers in Asian countries didn’t find these topics (Well, it seems a little off the point since this topic involves complex problems concerning political and cultural factors).

Let us come back to this paper.

(Source: King et al., 2014)

My focus is on the methodological aspect. This article uses randomized experimentation, which provides a very strong support to the results and conclusion this paper made. I think, this is the most shining point of this paper, which enable it to be published in Science.

How did the authors do the experiment? As we know, an true experiment should have a treatment group, a control group, a independent variable related to treatment, dependent variable that measures the effect, random assignment, and various requirements to guarantee internal validity. The identification of experiment elements in this study is easy. Because the cause precedes the effect, we can tell that each post is the subject which the treatment works on, the treatment is the construction of the content within each post, and the reaction of the censor is the effect. In this sense, the independent variable would be the content of the post, which can be classified into four categories based on two variables: pro-government, anti-government, with collective action potential, and without collective action potential. The dependent variable is the censoring result of the post, in other words, the outcome in terms of censorship. As for random assignment, this paper has made a lot of efforts to ensure its implementation. So far as I can find, the paper did the randomization in a lot of details. For example, when submitting two posts about two different events to each account, the authors randomized (i) the choice between these two pairs, (ii) the order within each pair, and (iii) the specific collective action and policy events we wrote about in each submission. Furthermore, to fulfill the requirements of a true experiment and eliminate possible threats to its internal validity, the author takes many measures, including having the same keyword included in both the treatment and control conditions, controlling for individual writing styles, having all posts in a similar length, ensuring no identical posts to anyone found in social media, and not being obtrusive to change the social media ecosystem.

This study also employed participant observation to obtain the descriptive knowledge of how the censorship process works. in order to supplement the method of experimentation. This method is typically adopted by qualitative researchers. But in this paper, it fits in very well and provides informative materials of censorship to the researchers. Together with experimentation, it gives the authors good way to tell a more interesting story.

I think such an impressive and potentially-illegal-in-China work could not be done without help of a native Chinese. Not surprisingly, I noticed that the second author of this paper, Jennifer Pan ── a talented PhD candidate in Harvard ── is actually a Chinese. She must have got a lot of help inside mainland China, and recruited a myriad of research assistants to help with the extensive post writing and publishing.

Bibliography

King, G., Pan, J., & Roberts, M. E. (2014). Reverse-engineering censorship in China: Randomized experimentation and participant observation. Science, 345(6199). doi: 10.1126/science.1251722

An interesting experiment on China’s censorship