Dr Luke Sloan is a lecturer in Quantitative Methods in the
School of Social Science at . You can contact him at SloanLS@cardiff.ac.uk Cardiff University
On its own social media data is messy, unfocused and lacking in context - however it is potentially a rich and plentiful source of information on the attitudes, beliefs and experiences of individuals if we can work out how to harness it. In my presentation I will discuss how we can geolocate Tweets within ONS geographies and thus create a new augmented secondary dataset which will enable a much wider range of social issues to be explored than through either dataset in isolation. Through linkage with rigorous survey data we can start to test the legitimacy of social media as a data source and develop methods and models for its use. I will also explore the importance of bridging the gap between the computing and social sciences so that we can respond to what Savage and Burrows (2007) have notoriously dubbed 'the coming crisis of empirical sociology'.
Yet even with the development of augmenting methods we still have to face the question of what social media data is (or can) actually tell us about the social world. Is it possible to reconcile elicited and naturally occurring data? Is the fact that it occurs naturally an advantage compared to traditional survey research which can be criticised over measurement bias and artificiality? Can we conduct legitimate social science through data mining or do we have to test theories? Are online representations of individuals simply 'avatars' that are not representative of their real attitudes and beliefs - and is this not the case with survey respondents anyway? Considering that around 500,000 tweets have been made since I started writing this post, we'd be fools not to pursue these questions further.
We have access to more data now than ever before, but like the castaway stuck on an island we are surrounded by water that we cannot drink.
I do not intend to go thirsty - do you?