Understanding How Ordinary Users Comprehend Data Visualisations

The final speaker in this session at AoIR 2015 is Helen Kennedy, whose interest is in how people interact with data visualisations. This is very important in the context of the current datafication trend. But existing literature in this field lacks a user-centred knowledge base – much is driven by designers' instincts of what constitutes a good data visualisation. It mobilises narrow definitions and measures of effectiveness and provides little information about participants, while ignoring social and cultural factors.

This work, then, could learn from media and cultural studies work that investigates how audiences engage with media texts. Helen's work operates in this field, and undertook focus groups and interviews with visualisation 'users' who had engaged in shorter- and longer-term diary keeping.

The sample topic for this work was migration, and so the work was undertaken in urban as well as rural areas with low as well as high migration. Several social and cultural factors emerged from this. Visualisations do not exist in isolation from the context of the subject matter they represent, and this affects user engagement; attitudes towards media outlets and data sources affected trust in specific visualisations; visualisations confirming as well as challenging pre-existing beliefs and opinions were especially effective; the time required to engage with specific visualisations was an important factor (which means that focus groups with their dedicated time to examine these graphs present an artificial environment for engaging with visualisations); perceptions of personal confidence and skills in critically 'reading' data – or graphicacy – affected the understanding of visualisations; and emotional responses towards visualisations as such, the data, the subject matter, and the data source or media outlet also played an important role.

Broader understandings about what makes a visualisation effective are required, therefore. This needs a broad range of definitions about effectiveness, well beyond information retention alone. As daily life is further quantified, for example through the omnipresence of Facebook metrics, we must better understand the emotional response towards data – the feeling of numbers. This requires a rethinking of the cognitive-rational standard model of statistical education, in pursuit of greater visualisation literacy.