Visualisations

Use of visualisations

Data visualisations can be very useful to fact checkers and other data users. A good plot or graph can clearly show insights – growth in numbers, changes over time – that looking at the raw numbers would make difficult to see. It can save fact checkers hassle by not having to browse through datasets and create their own analyses.

Importantly though, plots, graphs and charts should always be published along with the data on which they’re based.

Fact checkers said they find it frustrating when they have to transcribe values from a graph or image on NSI website. It means having to write down each individual value and create their own data table. Publishing the original data with the image would have saved them a lot of time and hassle.

Visualisation file formats

Visualisations can be created and published in a number of different formats.

Static image

  • Plain image which doesn’t allow for any interaction (can’t click on it or hover over it for more info)
  • Usually a file such as a .jpg or .png
  • Often seen inside reports - graphs or plots of some kind
  • They can easily be copied and used elsewhere (for example a news article).
  • However, they don’t contain any meta data, or way to link to the underlying data.
  • Examples:

Interactive images

  • Found within web pages on NSI websites
  • Can show extra information to the whole visualisation and to individual data points
  • Can be used to hover over a data point in a graph to get more information
  • Also can be used to allow switching between viewing the graph and viewing the data table
  • Can be hard to use this visualisation elsewhere. Although easy to take a simple screengrab to share as a static image.
  • Lots of different software libraries exist to create these interactive images
  • Examples of interactive visualisations:

Dashboards

  • Dashboards are an entire package of interactive images and visualisations
  • Organisations use these to create pages which display lots of interactive visualisations centered around a topic
  • These often can look excellent and provide great insights but should always have the underlying data easily available to download.
  • Statistics Sweden have developed an open source tool, PxWeb , which helps build web applications for publishing statistical tables.

Design of visualisations

Data visualisations raise interface design questions. It involves choices about colours, sizes, layouts and user needs. As such, they can often involve a lot of time to develop.

However, a little can go a long way. Before creating big interactive visualisations or dashboards with elaborate software libraries. It’s worth starting with simple plots or graphs and asking data users if they are sufficient. If they would like to see more or the images adapted, then you can adapt them further.

References

Data visualisation is a big topic. But here are a few guides that may be helpful.