The Emotional Dimension of Data Visualisation

It's the final day at Social Media and Society, and today's keynote is by Helen Kennedy. She's beginning with the question of how data make us feel: and it is a question that reflects the power of data, of metrics, in an environment where such data have become ordinary and everyday. As data mining becomes more commonplace, new data relations emerge, and these are increasingly characterised by emotion as much as by rationality. This represents a desire for numbers, and points to some of the contradictions that include a hunger for as well as a criticism of numbers. How we get through the visualisation of numbers also plays an important role.

Social media data form part of a larger ecosystem of connective data in this context. Social media data mining has also become ordinary – and should such ordinary forms of social media data mining concern us in the same way that more exceptional data (such as the Snowden leaks) should concern us? Helen has examined this in close engagement with social media data generators and users, and this has also uncovered the visceral reactions that people have when they encounter a data visualisation. Such responses to 'seeing data' also deserve further research: how do we live with data from the bottom up?

Data are now ubiquitous, especially in an online and social media context. Such data play an increasing role in decision-making and our knowledge of the world; numbers, metrics, and data matter. Numbers appeal because they are familiar and can be understood at a far distance from where the numbers are being generated. 'Big data' allows us to view phenomena from afar; measuring is a way of thinking, therefore, not just of measuring itself. The more we measure, the more data measures also become normal: quantification is a way of generating norms for the world.

But this alone doesn't account for the emotions that surface as data become ubiquitous: little attention has been paid to datafication's emotive dimensions. An experiment such as the Facebook Demetricator, which removes all the numerical data from the standard Facebook interface, shows up how crucial numbers are to the experience of using this social media platform: this constant ordering employs quantification to describe social existence, and this generates a strong desire for further metrics as a validation of the (quantified) self.

This is an emergent pattern, and it can be observed across many aspects of everyday life. There is a strong desire for numbers for instance also amongst civic and public organisations who seek to validate their digital activities by quantifying the social media returns received, and the gathering and reporting of such metrics is now often a requirement for these organisations – but often there is no discernible action taken as a result of reporting such metrics to managers and funders: the desire for metrics reporting is fulfilled, and the numbers are simply filed away.

There is also a growing understanding amongst these organisations, on the other hand, that social media data are not representative of wider demographics; and indeed, some social media analysts are also finding that the accuracy of the data is not very important to their clients: the clients simply want to be able to show that they are tracking their metrics, rather than extracting any meaningful information from them. The clients simply do not have the data literacy to understand the metrics at this stage; they merely have a desire to have these numbers reported to them. There is even a simple desire to reach symbolic, round milestone numbers – which Helen describes as the "fetishism of the 1,000". There is a tautological sense that having social media metrics helps organisations generate better social media metrics, even.

Desiring numbers results from a general trust in 'hard' numbers, combined with a constant desire for more metrics, independent from the actual, practical utility of such numbers. This is further complicated by the fact that the main way for us to access such numbers is through visualisation, which introduces an aesthetic dimension and generates a further opportunity for misrepresentation and misunderstanding. Data are as much felt as they are experienced cognitively and rationally, especially when they are experienced through such visualisations.

Emotional responses to visualisations are driven by the stories intended to be told by the visualisations; by the actual data underlying the visualisations; by the specific visual style selected; the subject matter presented; the source and publishing venue of a visualisation; and the user's self-assessment of their confidence and skills in reading the visualisation. These are also influenced by the prior knowledge and experience that users bring to their reading of a visualisation, of course.

So quantification is not necessarily experienced as distancing here, but can reduce distance and bring readers closer to the matters being addressed by the visualisation: while the data themselves may introduce distance, the visualisation of the data can reverse that distancing. The sociocultural context of engaging with data visualisations is crucial, in academic data visualisation as much as everywhere else.

This has troubling consequences, as it raises issues including: the inability to engage in critical conversations about the limits of data analytics (in the absence of sufficient data literacy amongst the users of metrics); the need to develop new skills for gathering, analysing, and presenting (especially social media) data, often also at short notice and outside of standard working hours, without a great deal of formal training or oversight; the data literacy divides across society that our increasing reliance on data and visualisations opens up, and that can have substantial societal effects (viz. #Brexit); the need to look at data as become ordinary, lived from the bottom up, experienced in everyday; and the need to develop appropriate policy settings to address this, aiming to 'do good with data'.