Polls, the data analysis of social media, and sociology are all modes of knowing the social world. Social media analysis deploys automated software programs to find and extract meaningful patterns and trends by sifting through massive amounts of data (known as “data mining”). The analysis of social media data can point to nationwide and worldwide trends about social issues and events if the analysts are working with big data sets. However, no matter how sophisticated the tools and programs analysts use, large-scale social media data can hardly present substantial sociological insight into the reason why individuals behave in certain ways and how they make sense of social reality.
Social media analysis can be useful to sociologists for a number of reasons. In addition to demographic information, the computational findings of social media analysis can tell us about general movements, preferences, and trends in political participation, social identification, civil society, and consumption. Nonetheless, there is always a high possibility of inauthenticity that could make findings less reliable, and there are ethical problems with using such data without consent of individuals whose data is studied. Importantly, an online environment can be quite different from the offline environment, because people can easily mask their real preferences and even identities on the internet.
Similarly, opinion polls measure the preferences of the majority of individuals on an issue. The poll analysts deploy well-developed quantitative methods and research tools (some of which, such as surveys, were originally designed for sociological research), which have the potential to offer quite precise predictions about consumption behaviors and voting preferences. The counter to this is that many claim that the polls, for example, failed to anticipate the dynamic of the US vote in the last national election won by Trump, because they could not access individual stories and social localities, which led to significant differences in the election results but were lost in the large-scale quantitative data generated by the polls. The famous sociologist Pierre Bourdieu considered the concept of public opinion to be illusory not only because it represents a shallow summation of diverse views, but also because opinion cannot be reduced to a response one gives after having been asked for to pick from a limited list of pre-ordained answers in a survey.
However, sociological research, in particular qualitative research, which involves in-depth interviews and long-term participant observation, is much more capable of discerning nuanced views and motives, which might make a great deal of sense in explaining a particular social phenomenon. It can penetrate reserved beliefs, concerns, and value-systems. It can map out the diversity of thoughts and emotions more effectively and comprehensively, and offer a fuller understanding of how they play themselves out in practice. Overall, I do not think that we can replace sociology either with polls or with social media analysis.