Working Thesis Title
The role of network dimensionality in political polarisation.
Research on political polarisation typically focusses on finding a bimodal division between political parties, usually the US-American Republican and Democrat parties. This approach can be difficult to apply to many-party democracies like New Zealand. To address this gap, my thesis explores the idea that polarisation is instead about the political field narrowing down to two possible positions through a process of increasing correlation. As an interdisciplinary scholar, I also draw on work from political science which argues that polarisation is not just correlation in the political field, but is a process of dividing material and social networks as well, for example, through passing apartheid laws. To apply these frameworks to data science, I use Random Dot-Product Graphs to capture social networks in spaces such as Twitter, and use Singular Value Decomposition to assess whether their dimensionality – how correlated the network is – is changing. This method gives easily interpreted results about whether polarisation is occurring in the network.
Supervisors:
Primary Supervisor: Giulio Dalla Riva
