Tuesday, 15 May 2012

Using the Social Web to Study Something Else

Mike Thelwall is Professor of Information Science and leader of the Statistical Cybermetrics Research Group at the University of Wolverhampton. In this post he explores the potential of the social web to investigate social issues.

The rise of the social web has made many new types of study possible: research into the social web itself (e.g., who uses Facebook and why?) or its use for a specific purpose (e.g., what role does YouTube play in political campaigning?). In addition, the social web can also be a data source to help study something else: in other words a phenomenon for which the social web is not central. For this purpose, new online social research methods may supplement or replace traditional methods like interviews and questionnaires. For example blogs have been used as a replacement for diaries in diary research (Hookway, 2008) and to get retrospective public opinion about issues by investigating historical posts (Thelwall, 2007).  Another example is tracking the public mood through automatic analyses of public tweets (Bollen, Pepe, & Mao, 2011). In theory, it seems that the social web may help to shed light on any social issue, as long as it is not ignored by social web users.

Some advantages of using the social web to help investigate an issue are that accessing the web is convenient and free for internet users, gathering public data is relatively unproblematic from a research ethics perspective (I would argue), and that analysing the social web is unobtrusive and so does not place any burden upon others, in contrast to interviews and questionnaires. Nevertheless, there are also significant drawbacks with using the social web to investigate something else. The sample investigated, social web users that write about a particular topic, is likely to be a biased sample of the population of interest. For instance it may be more wealthy and more educated than average. Social web content may also be ambiguous or misleading, especially if it is intended to be read as part of a long-standing communication. For instance, the text, “I really hope that Dwain Chambers runs for England” could be ironic or not, depending upon the attitude of the poster towards punishments for drug taking in sport. Moreover, internet trolls and others may deliberately misrepresent their opinions for the purpose of generating arguments. These major disadvantages may seem to rule out using the social web to study something else, but alternative methods also have significant drawbacks. Questionnaires may generate respondent biases, respondents may answer inaccurately, or it may be impossible to generate any kind of meaningful sample. Similarly, interviews are typically too few to generate statistically valid conclusions, although they can produce deep insights into an issue.

In summary, at the abstract level discussed here, it seems that the social web has the potential to be used to investigate other social issues but researchers should be cautious with its use due to problems that affect the validity of the results obtained. These problems can be reduced by using multiple methods, or by taking advantage of the potential speed of social web research to use it for pilot studies, with the results informing additional, follow-up research.

What do you think? Can social research using social media overcome the serious questions raised about their validity? What other issues need to be borne in mind? Let us know over at Methodspace.