Critical Data Visualisation

One chapter in the book deals with visualising data in different ways using Python. However, data visualisations exist in society much more generally, and have done for a long long time already! The field of Science and Technology Studies is particularly attuned to the use of visualisations as a way of construction knowledge about the world – some key readings worth picking up, if you’re interested in this, are:

Lynch, M., & S. Woolgar (eds) (1990) Representation in Scientific Practice. Cambridge, Massachusetts: MIT Press.

Coopmans, C., J. Vertesi,J., Lynch, M. E.,  & S. Woolgar (eds) (2014) Representation in Scientific Practice Revisited. Cambridge, Massachusetts: MIT Press.

Carusi, A., Hoel, A. S., Webmoor, T., & S. Woolgar (eds) (2015) Visualization in the Age of Computerization. London: Routledge.

Though these volumes are exclusively about the use of visualisations in and as scientific work (Science and Technology Studies is where much of the thinking around visualisations as social objects emerges, after all!),  there is much research on the use of visualisations in other domains of relevance to social science too, and many areas in which data visualisations can be seen to be embedded in wider social problems. For instance, a BBC report from 2014 (https://www.bbc.com/news/business-29898083) revealed that in the lead-up to the 2015 UK General Election, the sitting Conservative government had sent out a data visualisation to the 24 million income-tax-payers of the UK to inform them of where their income tax money goes. The visualisation – a pie chart showing percentages of income tax spending across different areas including defence, overseas aid, health, education, environment and transport etc – was widely criticised for presenting a picture that indicated, misleadingly, that nearly 25 per cent of peoples’ income taxes were spent on ‘welfare’. This fed directly into existing stigmatising discourses around unemployment benefits and other social welfare benefits which portrayed recipients as ‘scroungers and  ‘too lazy to work’. However, the government visualisation did not specify exactly what constituted this category of  ‘welfare’, and getting under the hood of how this visualisation was constructed (see the aforementioned BBC report) shows that the sitting Conservative government included such things as pensions paid to public servants (e.g. nurses, doctors, police officers and fire crew), thereby artificially inflating the  ‘welfare’ portion of the pie chart to suggest that welfare spending was particularly problematic. The government’s choice to visualise data in this way – to only vaguely suggest what ‘welfare’ is and categorise things like nurse’s pensions as ‘welfare’ rather than a ‘health’ expenditure – produces a misleading and potentially stigmatising account of the UK society.

Therefore, visualisations can be powerful representations of what we assume to be ‘truths’ about the world, but which might in fact be socially problematic in various ways. However, we can use Python to deconstruct and reconstruct such problematic visualisations for transparency and clarity, and to present alternative accounts. Hence, one additional exercise would ask:

Can you identify a potentially-problematic or biased visualisation (i.e. one produced by powerful institutions with a vested interest in managing public discourse, e.g. government or commercial companies), unpick/deconstruct the information used to build that visualisation, and reconstruct it in a different (and more socially responsible) way?

Find a topic that aligns with your research interests, find an interesting visualisation, and take it from there!