The distribution of public opinion signals social preferences. People make many decisions in their daily lives based on their perception of the opinion climate. Similarly, policy makers formulate policy proposals based on their understanding of public opinion. With digital and social media becoming an integral part of people’s daily lives for information and communication, the proliferation of digital technologies is changing not only how public opinion can be represented, but also how it can be studied. By adopting a big data-based approach, Dr Chris Shen Fei, a CityU scholar in the Department of Media and Communication specialising in the social and political impact of new media technologies, has proposed ways of understanding public opinion through online textual mining. 

Big data analysis of public opinion

“We are living in a time in which timely and comprehensive understanding of public opinion is greatly needed,” said Dr Shen. He pointed out that traditional polling has many limitations, such as high cost and respondents’ sensitivity to question wording. “But big data provides us with a new direction for public opinion analysis by taking full advantage of people’s openly shared expressions on the internet,” he said. 

Launched by Dr Shen and his team, the Hong Kong Online Public Opinion Data Mining Project (http://www.webopinion.hk/) aims to understand online public opinion using automatic textual analysis. After identifying 12 important online platforms as data sources, including discussion forums, news portal sites, and alternative news media sites, the team performed data crawling, data cleaning, tokenisation, lexicon development and data analysis to transform unstructured data into a visualised pattern of public opinion over time in Hong Kong.

“We hope the datasets and analysis derived from the platform can benefit decision-making by policy makers, the public and the academic community in the long run,” said Dr Shen.

The project found that in the past several years, online political discussions have become more and more sensational, which have led to polarised opinions and a divided society in Hong Kong. This poses a great challenge to political discussion, which is Dr Shen’s other research interest. 

“Hong Kong’s problem does not lie in the ideological differences among citizens; rather, the major issue is that people of different political stripes view each other as enemies. Sensational discussions and hate speech are commonly seen on social media platforms. While it is undesirable and even impossible to eliminate political differences, reducing political affective polarisation is one of the greatest tasks faced by society,” said Dr Shen. “Political affective polarisation means the tendency of people to dislike or distrust others simply because of a different political stance.”

Pilot experiments on deliberation

Much evidence suggests that when communication and discussions are not properly facilitated and conducted, they can easily lead to the proliferation of extreme ideas and negative emotions. To explore the ways of reducing political polarisation in Hong Kong through communication, Dr Shen conducted two experiments to compare the effects of deliberation and casual discussion.
 
In Study 1, people holding opposing views on Article 23 of the Hong Kong Basic Law were invited to participate in a 90-minute discussion session. They were randomly assigned into one of two groups: deliberation or casual discussion. The deliberation group received an information booklet on the issue and had to strictly follow the rules, whereas the casual discussion group had no such stimulus. In Study 2, which tested whether watching other people’s discussion and deliberation would have a similar effect, video recordings from Study 1 were presented to another two groups of participants. One group watched the deliberation video and the other group watched the causal discussion video. Pre-test and post-test surveys were conducted in both experiments. 

The studies revealed the following: 
i)    Both deliberation and casual discussion had mixed effects on reducing political polarisation. While issue attitude and issue polarisation remained largely unchanged, people’s attitude towards those with opposing views became more favourable and affective polarisation was effectively reduced.

ii)    After discussion, people’s knowledge level remained largely unchanged, but their sense of national identity became stronger.

iii)    These effects were more prominent in the deliberation group than in the casual discussion group.

iv)    People who watched others participating in deliberation and causal discussion showed similar effects, but to a much smaller extent. 

Based on these findings, Dr Shen recommends the policymakers provide opportunities and set up platforms for political dialogue among the public. These activities could be organised at the community level by non-governmental organisations. A more realistic approach would be to identify a few communities as field experiment sites and conduct a longitudinal study to follow the long-term impact of a community-based social dialogue programme. “In the long run, Hong Kong society needs institutions and organisations to promote discussions among citizens, with the government acting as an incubator, providing the necessary resources,” said Dr Shen.

This research article originated from CityU RESEARCH.