On Thursday 22 August, the third meeting of the Policy Visualisation Network took place. The event involved some great presentations and was followed by workshop activity. A big thanks to Evan Hill from the Department of the Prime Minister and Cabinet, Merry Branson of the Australian Bureau of Statistics and Mark Kobal of the Australian Taxation Office for putting together and running a great event. The following is my write-up of the event (unfortunately it’s non-visual!).
Opening speaker – Brian Pink
The event was opened by Brian Pink, the Australian Statistician (head of the Australian Bureau of Statistics). Brian spoke of his pleasure at seeing the growing interest in data visualisation across the Australian Public Service and the move to applying it in policy contexts.
Brian noted that the use of data visualisation is something that is being pursued across many countries, and that it was one way of bringing sense to complex data sets. It is about helping make informed decisions based on the underlying data, which is important when not everyone is comfortable or literate with complex data.
He noted that while it is a valuable approach, there are also risks associated with data visualisation (not least of which he warned of the perils of pie charts, which he suggested were one of the worst ways of presenting underlying data).
Brian finished by noting the goals of the Policy Visualisation Network and expressing his hope that the information and skills shared through the Network would help the continuing development of data visualisation.
Second speaker – Alan Smith
Alan Smith OBE is Principal Methodologist of the Data Visualisation Centre at the UK Office for National Statistics (ONS) (Alan very kindly spoke with us live via video link despite the time difference with the UK).
Alan noted that the Data Visualisation team at ONS was set up a number of years ago because of two reasons – the way people are accessing information on the Internet is changing and becoming more visual; and due to a wider need for data literacy.
Alan used a number of examples to show the power of data visualisation in framing a debate – including the scales used for depicting data, the assumptions embedded in how the data is presented, and how contextual data is used. As much as possible, Alan advocated that the aim is to use data visualisation to provide illumination – to reveal things that people are not expecting to see. (One of the examples he used was of a visualisation tool of commuter journeys inside Greater London).
He also spoke of the importance of symbology – choosing how you represent the data. Simpler forms of symbology are more effective for communicating.
Alan noted the many different forms of depicting statistical relationships – including magnitude, change over time, distribution, part-to-whole and correlation. He took us through a number of examples of ways of depicting change over time.
Next Alan spoke of the importance of using colour effectively and advised it was best to design your visualisation in grey scale, and then if wanted/needed add colour to it – this makes it accessible to people who might be colour blind, and ensures that it is still meaningful if someone prints the visualisation in black and white.
Alan finished by talking about the option of adding visual flair to charts and graphs, and noted that this can help make them memorable, but also risk making them seem lest trustworthy or credible.
In response to a question about whether some visualisation tools might mean you lose control of the message, Alan noted it depended on whether the visualisation was part of telling a story (in which case static visualisations are more appropriate) or whether it is being used as an exploratory approach (in which case providing the user freedom to explore is important).
Third speaker – Mitchell Whitelaw
Mitchell Whitelaw is from the Centre for Creative and Cultural Research at the University of Canberra. Mitchell spoke to us on ‘Data Visualisation as a Cultural Practice’.
Mitchell began by talking about the role of context for data visualisation. Data visualisation is a medium for carrying a message, and it is part of a social and political context – the data cannot be understood unless the context is also known.
Mitchell showed us the data visualisation example from Periscopic of gun deaths in the USA. He noted it was strongly authorial, evocative and emotional – but also strong and effective in conveying its message. A visualisation cannot be understood separately from the context.
Visualisation can also be a rhetorical device in and of itself – the data may not be meaningful, but the visualisation might be used to convey some other meaning (e.g. complexity).
Mitchell noted that there is a spectrum of prosaic to poetic, and that this is often linked to functional (prosaic) to non-functional (poetic). However, he suggested that there can be different functions for many different contexts and the poetic end can also serve a purpose. Mitchell gave his own example of his ‘Weather Bracelet’ – which is poetic but also representative and challenges people to think about the data in a different way.
In terms of creating visualisations, Mitchell noted that in order to create a visualisation, you first need to create a visualisation – i.e. generally with large data sets there is no way to get a hold on the data until you have put it in some (rough) visual form. Data visualisation is an iterative, creative process.
He noted that there is a risk that the visualisation can become an end in itself, so visualisation should only be done with the data that you are trying to explore (e.g. don’t create a data visualisation from some dummy data that you will then insert the real data into – always create according to the data that you have).
Mitchell noted the difference between explanatory and exploratory visualisations. Exploratory visualisations are about finding questions, rather than answers. They can reveal unknown, unexpected attributes.
Mitchell finished by emphasising that how you represent the data that is missing or not there is very important, and that you should never underestimate the political weight or significance of data.
Many of the attendees then joined in for some structured workshopping to create visualisations around particular hypothetical policy issues. It was a great way to meet other people interested in visualisation from across the APS and to get insight into different ways that visualisation can be approached, and the varied methods that can be applied.
The next event for the Policy Visualisation Network will likely be towards the end of 2013. Notice will be on the Policy Visualisation Network site and through the Public Sector Innovation Network.