The contents of a Facebook page. The middle of the graph represent all of the posts in time order, the left side is a network of users who interact with them and the right side is a network of words that appear in them.
As people in different industries frantically amass digital data such as patient records, social media posts, images, feedback, online transactions, or motion tracking data there is an urgent need to rethink how data is processed and analysed. Much of this data is made sense of using automated computational processes like algorithms, machine learning and statistical procedures in which human experts only supervise or ‘train’ the scripts to find patterns in the data. This results in very opaque systems which squeeze people into categories, or give them uneven access to services (health care, loans, insurance policies or jobs) while the inner workings of these systems remain unaccountable and often incomprehensible to their users. Popular commentator Cathy O’Neil has even dubbed these “weapons of math destruction”.
Qualitative researchers, anthropologists and philosophers have regularly complained that these methods of data analysis are reductive, and fail to capture the complexities of lived experience. They have shown how people frequently fall between the cracks of algorithmically assigned categories or try to game their metrics. They have argued that these types of analyses contain all sorts of assumptions and judgements which are concealed underneath a language of objectivity, yet have wide reaching political and ethical consequences. However, these researchers have rarely proposed alternative ways of analysing data at scale.
Visualizations as Exploratory Maps
New techniques for visualizing data offer the possibility to represent complex information without unnecessary reduction or aggregation, which means that digital data could be analysed in more interpretive and exploratory ways. These visualisations could be read like maps charting an uncertain terrain – like a doctor looking at a MRI scan of a brain – instead of being used to convince us of correlations and quantities or explain the relationship between narrowly defined variables. Such techniques could allow practitioners to ask broader questions of their data, problematise their data sources and measurement systems and allow more fruitful collaborations with researchers from other disciplines. However, these possibilities for research are often complicated by the fact that they fall between different disciplines and traditions of so called “quantitative” and “qualitative research”.
