Data visualization is a very useful tool for summarizing, and sometimes interacting with, complex datasets. It shifts a dataset from being a detail oriented process into a more comfortable visual process. Throughout evolution humans have had a great use for visual information. Being able to see the world and recognize its patterns has enabled us to spot the approaching danger, and choose healthier mates. It’s essential for the preservation and production of life. Granted, comparing the difference between data visualization and textual data is not the difference between life and death.

Despite the benefits data visualization is a methodology that presents itself as a panacea. It is presented as a technique that is “better than the alternative.” The issue with visualization is that it is rarely ever shown in context of the application that is requesting its services. It is the person that shows up to a Christmas party in their Halloween costume. The treatment of a visualization component typically attempts to dominate other modes of understanding on the screen. For example, when a visualization is displayed on a web page in the main content area it typically takes the eye away from any other descriptions within the content area or navigation options of the page. Descriptions of the visualization of the context of the visualization are more important than the visualization itself. The descriptions of the visualizations instruct the reader on what the visualization is summarizing and how to read the graph.

Since most visualizations are not contextually relevant the visualization tends to stand out from its surroundings. The goal of any visualization, unless it itself is the application, is to complement surrounding data give a visual overview. If the visualization is taking the eye away from the surrounding content the visualization moves from information to only data. Additionally, due to the lack of context, the user may become confuse or may underestimate the data behind it. This problem is amplified when the user is seeing the visualization for the first time. By creating more obstacles with the visualization the overall meaning becomes lost or ineffectively communicated. Additionally, differences in cultures may add another layer of complication in effectively communicating the data through visualization.

For the academic studying visualization, the issue of context of the application surrounding the visualization is not a main focus. The context disregarded as “window dressing.” After all, the academic has already had training on what the visualization is, what types of data that it’s good for, and possibly how to interpret the graph. However, they are not taught how to interpret the data from the graph. To effectively interpret the data from the graph, one has to have context.

News papers typically do not have the issues that I have outlined above. They do not have these issues because the focus of the article is not the visualization. Additionally, if a new article has a visualization, the visualization is non-interactive and takes up a very small percentage of the article it accompanies.

Some thoughts on how to improve existing visualization:

  • Involve Human Computer Interface researcher/specialist to fix issues with interactivity

  • Involve editors to fit the visualization within the context of the page and audience.

  • Get graphic designers involved in the design of the visualization, and to have it fit the application.

  • Maintain Constancy throughout the application/website

  • Make the visualization very simple to use

  • If the visualization is not common (most aren’t, and it is typically assumed they are)

  • Give the user a preview of the visualization before they use it. This gives the user time to process the visualization as “I’m looking at something new here, time to switch to right brain.”

  • If the visualization will not fit in the context of the application, then allow the user to open it in a new window.