Dataviz logo
Communities and Local Government logo Oxford Consultants for Social Inclusion logo

Case study: What is good visualisation?

What is good visualisation?

Good visualisation can help users explore and understand data, and also communicate that understanding to others:

  • Exploring and analysing data: Visualisation is a central tool in carrying out analysis, enabling researchers and other users to explore datasets to identify patterns, associations, trends and so on;
  • Presenting and communicating data: Good data visualisations can help users make robust decisions based on the data being presented. They should provide an effective representation of the underlying data, to help answer a particular question at hand. Communicating data in this way can support senior decision-makers engaged in strategic planning, service managers needing to understand where delivery could be improved, and managers wanting to monitor performance.

In the excellent "The Visual Display of Quantitative Information", Edward Tufte is quite clear on the value that good data graphics have[1]. They should "above all else, show the data", and should:

  • Help the audience think about the the important message(s) from the data, rather than about methodology (graphic design, the technology of graphic production etc), or something else
  • Avoid distorting what the data have to say
  • Present many numbers in a small space - but also emphasise the important numbers
  • Make large data sets coherent, and encourage the audience to compare different pieces of data
  • Reveal the data at several levels of detail, from a broad overview to the fine structure.

Why "visualisation"? Why not "data graphic"?

This project has looked at many different ways of presenting data. Some are standard data graphics (or charts) intended for print (for example, see the bar-chart examples). We have also looked at interactive ways of presenting data that are more appropriate to show on websites (for example, see the data animation examples and virtual world examples), and that are not really covered by the term "data graphic" . We therefore use the term "visualisation" to cover all these different ways of presenting data.

Principles of a good visualisation

Good data visualisation is simply another way to communicate with your audience, and the same things apply as with other ways of communicating. Identify the point(s) that you want to make, identify your key audience, and create the clearest visualisation that conveys that message to that audience.

The table below highlights key principles of visualisation, with some of the issues to think about. In the next guide on practical steps for good visualisation, we show how following these principles can greatly improve the way that data is presented.

Design for your audience
  • Think about the message that you are trying to convey to this audience with this particular visualisation - and focus on emphasising this message. Try not to cram too many key points into the visualisation, unless you want the audience to spend some time looking at it (for example a web-tool for interactive exploration of key indicators might need to show users a great deal of data to help them identify patterns and trends).
  • Different audiences may need different visualisations. A good design for a visualisation to be used by researchers exploring how economic indicators vary over time and between places may not be appropriate when showing the same data to senior officers and members to emphasise key economic trends
  • Visualisations must serve a clear purpose: communication, exploration, tabulation or decoration. Don't try to do everything with one visualisation.
  • Organise the information in order to emphasise what you are trying to say to the audience, and don't bury the key messages in a mass of detail.
  • Write out key points on the graphic itself, and label the important data events - the graphic should speak for itself

Accurately represent the data

  • The visualisation should show the underlying data without distortion.
  • There are many common pitfalls to avoid, for example axes should always show zero to avoid exaggerating the importance of differences between data values (See the guide on practical steps for good visualisation for more examples of pitfalls to avoid). "The representation of numbers as physically measured on the surface of the graphic itself should be directly proportional to the numerical quantities represented" (Tufte).
  • Clear, detailed and thorough labelling should be used. Write out explanations of the data on the graphic itself, and label important events in the data.
Keep it clear
  • "Data graphics should draw viewers attention to the scene and substance of the data" (Tufte), "The purpose of visualization is insight, not pictures" (Shneiderman).
  • In other words, the graphic should focus on the message(s) for the audience, and all visual clutter kept to a minimum. But don't cut out all visual elements unless this is the way your audience likes it - things that emphasise the key message are useful if they help get your points across to the audience
  • Remove unnecessary "non-data ink" (keep asking yourself the question "will this visualisation suffer any loss of meaning or impact for the audience if this element is taken out?")

Different purposes for visualisation

Above we highlighted different uses of visualisation, for exploring/ understanding data and communicating findings to others. There are three different aspects of the way that visualisations can be used (we've adapted this from work by MacEachren et al[2] on how people use maps):

  1. Communication or understanding: Is the visualisation presenting/ communicating knowns to an audience, or revealing unknowns?
  2. Interaction: How is the user able to interact with the visualisation?
  3. Audience: Is the visualisation intended for public dissemination (eg, to a general audience), or private use (eg, by more technical audience)?

Image for this case study
OCSI (2009), after MacEachren et al (1994)

The cube above shows these different aspects plotted across three-dimensions, which can be used to classify different types of visualisation by the way that they are used. For example, visualisations that lie in the furthest-top-right corner are those that are primarily intended to communicate results to an more general audience, for example presenting performance information to citizens using printed (or PDF) reports. The table below highlights different examples of how visualisation can be used in the public sector, based on the cube above:

   
Revealing unknowns
Presenting knowns
High interaction Private Exploring data for patterns, using flexible visualisation tools such as Excel, GIS applications, Tableau, intranet Local Information Systems Interactive performance management tools, providing a series of data reports on service delivery areas such as the economy, health, crime, and so on. Interactive features allow service and performance managers to drill deeper into performance data
Public

Interactive online systems, for example:

  • Gapminder, presenting socio-economic trend data;
  • Many Eyes allowing users to upload data and visualise in different ways
Communicating performance or service information to citizens online using interactive tools, eg location and quality of health services overlaid on Google Maps
Low interaction Private Communicating interim results of research to internal audience Internal research briefings to senior managers
Public Research reports presenting multiple views on data, eg Joint Strategic Needs Assessment (JSNA) Communicating performance information to citizens using printed reports

Where can I find out more?

The case studies and examples on this site provide a wide range of material on good visualisation. We also suggest that a very good starting point (for all those engaged in visualising data) is to look at work by Edward Tufte and Stephen Few, including:

  • Tufte (2001). The Visual Display of Quantitative Information
  • Few (2004). Show me the numbers: Designing tables and graphs to enlighten

We have also provided links to a wide range of useful resources on the web.

Footnotes:

[1] Tufte (2001) "The Visual Display of Quantitative Information"

[2] MacEachren, A. M. and Taylor, D. R. F., Ed. 1994. Visualization in Modern Cartography. Pergamon Press. Also see http://www.geovista.psu.edu/sites/icavis/icavis/poland1.html