Wednesday, 23 September 2015

Social media research: Issues of performativity and validity

Keeva Rooney is a researcher at NatCen focusing on health surveys using bio-medical data. She has a BA (hons) in Sociology with a specialism in Social Policy from University of Warwick. During her studies, Keeva often used social media in her research, with one of her project’s focusing on online facebook groups for disabled people affected by the bedroom tax.

Social media research attracts many researchers due to its offer of practicality, creativity and accuracy which comes from its often covert nature. The idea that you can produce valid research by analysing people’s online activity is attractive to researchers looking for more innovative and contemporary ways to achieve a true insight into society. However, is the assumption that people act online as they would offline a valid one? And if not, does this raise concerns over the reliability, representativeness and overall accuracy of using social media for research purposes?

Feminist theorist Judith Butler studied how people ‘perform’ gender, be it within or outside their assigned gender norms. She believed that it was this expression of gender, not the biological sex itself, which determined gender [a]. Whilst Butler used the concept of performativity to deconstruct issues around gender, the same concept could be applied to how we use social media: do we use social media to construct a distinct identity, or is social media purely an extension of how we already perform within wider society, and does this matter for Social Media researchers?

I will be analysing this argument using two research ideologies; positivism and interpretivism [b].

On the one hand, a positivist approach suggests that social research can uncover an empirical truth about people and society. Positivists aim to find this truth in all aspects of society, using mainly quantitative data such as surveys or content analysis. For example, when analysing social media, positivists may often use sentiment analysis to analyse online attitudes and opinions of the user. For this data to be valid for understanding ‘real world’ attitudes, the opinions expressed online by social media users must be an accurate reflection of their offline beliefs. Therefore it could be argued that if someone is not being their ‘true self’ online, any conclusions drawn from this research may also not be valid.

However, it could be argued that the online persona cannot be disconnected from the offline reality. That is to say that how people act online is often a valid reflection of how they act offline. For example, a recent psychological study into internet trolling showed that those who enjoyed trolling often displayed characteristics such as sadism, psychopathy and Machiavellianism offline [c].

On the other hand, an interpretivist approach suggests that in order to analyse actions and attitudes accurately, the researcher must take the user’s interpretation of themselves as truth; when it comes to social media, how users portray themselves online is what researchers use as fact. From this perspective, any online ‘performance’ of the participant doesn’t make the research any less valid because the participant’s interpretation of their reality is always accurate.

However, what happens if people reject their online persona once it has been researched, as is often the case with people who display offensive or criminal online behaviour? As Stephen Webster’s study showed, some people will try to disassociate their online behaviour from their ‘real’ offline life, claiming that how they act online is not how they ‘really act’ [d] and some may even deny that it was them, instead saying that their account was hacked [e]. Whilst this raises several questions (such as how can you know who you are researching online), it also raises ethical concerns if someone believes that a researcher has misrepresented them by basing research solely on their online behaviour.

As the ‘troll’ studies show, online personas can often be a true reflection of offline actions and opinions. However, the extent of this may be unclear, and participants may disassociate themselves from those personas. Researchers should always consider whether social media can be used to accurately portray and analyse personal and public opinion, and if more traditional research methods should be paired with this to create a more triangulated and accurate set of data.


[a] Felluga, D: ‘Modules on Butler: On Performativity’ Introductory Guide to Critical Theory 
[b] While I acknowledge that not all researchers will fit into either ideology, I believe that analysing validity in online research can be done by discussing these two fundamental research perspectives
[c] Buckels, E.E; Trapnell, P.D; Paulhusc, D.L: ‘Trolls just want to have fun’ 
[d] Webster, S: ‘What is trolling, and why do we behave so differently online?’  
[e] Wainwright, M: ‘Man who racially abused Stan Collymore on Twitter spared prison’