This dissertation is an exploration of the potentials for utilizing geotagged social media data drawn from Twitter – one of many emerging sources of so-called ‘big data’ – for geographical research. The three papers that make up this dissertation examine how novel combinations of existing conceptual and methodological approaches from geography can be applied to datasets that are popularly understood as being revolutionary and requiring wholly new and previously unforeseen methods of analysis. Specifically, this dissertation seeks to combine the methodological approach of critical GIScience with the conceptual disposition of relational socio-spatial theory in order to explore a wider range of socio-spatial processes embedded in this data than is conventionally done in more popular or technically-oriented social media mapping projects. As such, this dissertation attempts to demonstrate how particular combinations of theory and methods can allow for more substantive insights to be drawn than is commonly thought possible due to the proliferation of relatively simplistic analyses of this data. This dissertation asks three key questions: (1) How have geographers and big data researchers conceptualized the spatiality of big data, and how do these conceptualizations constrain or enable the analysis of geotagged social media data? (2) How can existing frameworks for conceptualizing the multidimensionality of socio- spatial relations be usefully applied to the analysis of big data? (3) How can relational socio-spatial theory and critical GIScience be applied to big data in order to produce alternative ways of imagining urban spaces and places and the inequalities between them?