Sent to you by David Andrew via Google Reader:
(Editor's note: this is a guest post by Enrico Bertini from Fell in Love with Data)
The following post describes not only one of my favorite papers from VisWeek this year, but I have the feeling it will be one of those papers which has the potential to lead to a big impact in the way we will use, and see, information visualization in the future.
Edward Segel and Jeffrey Heer decided to analyze the right thing at the right moment: How Do People Tell Stories through Interactive Visualization? (PDF, stanford.edu)
The whole trend behind storytelling with vis is, at least in my opinion, one of the biggest changes we are experiencing in visualization since many years. We can see its effects everywhere: taking inspiration from the NYTimes visualizations and the infographics that appear in several other international journals, many other press outlets and advocacy groups followed the route of data-driven journalism, which is slowly but surely becoming a rather successful visualization subfield of its own. It seems that after several decades of research in visualization being focused on data exploration, academics and practitioners alike are re-discovering what the fathers of Infovis have always been saying: "visualization is a great tool for telling stories".
Method: What the authors did
Edward and Jeff analyzed a large set of interactive visualizations for storytelling and, through an iterative process, identified common patterns and built a classification on top of it. Their paper is very rich and I heartedly suggest you to read it all if you are interested, because it is quite "thick" and I will not be able to convey all the useful details it contains in this single post.
In this post, I rather will focus on the parts of the models I deem useful and personally appreciated the most. I will try to suggest, as usual in these guest posts of mine, how you can use the results in the visualization practice.
The paper contains a classification of the design space organized around the following components:
- Genres - 7 generes of story telling vis exist.
- Visual Narrative Tactics - Visual devices that assist or facilitate the narrative: visual structuring, highlighting, transition guidance
- Narrative Structure Tactics - Non-visual mechanisms to assist and facilitate the narrative: ordering, interactivity, and messaging.
For a encompassing picture of these components, the paper offers a big overview table (click on it for a larger version).
Tacit Tutorial and Stimulating Default Views
These 2 patterns refer to how a new visualization is introduced to the reader and are of enormous importance. The problems of gently introducing a person into reading a visualization has been neglected for a long time and and it is great to see some suggested practical solutions here.
Tacit tutorial refers to the strategy used by some to gently introduce users to the interactive functionalities, for instance by starting with guided examples. Stimulating default views refers to the strategy of providing initial views that stimulate curiosity and encourage to explore further.
The identification of several genres can function as a starting point for any new story telling visualization, and help designers decide which style better suites the need of the story. The image on the top of this post is a visual summary taken directly from the article.
Messaging and Interactivity
Messaging and interactivity refer to the way the author sends messages in the visualization and how interaction is provided into the user's hands. They are somewhat in contrast, as the more messaging you provide, the less freedom is given to the user. Messaging can generate clutter, but interactivity can detract from the intended message. The existing tension between the two is one of the major issues and trade-offs an visualization designer has to take into account when developing visualizations for storytelling. So please take a note: you will probably have to introduce messaging and interactivity in your visualization, but you will be better off if you carefully balance them and understand when they support and when they hinder its intended purpose.
Author-driven, Reader-driven and Hybrid Models
Visualizations for storytelling are author-driven when there is no freedom for the user in the process of exploring the data. That is, there is only one clear and predefined path to follow. In reader-driven visualizations the user has, on the contrary, total freedom to follow any path.
Narrative visualizations are in general in between these 2 extreme cases, and the paper describes a series of hybrid models. Again, keeping these models in mind can help in reflecting what is the best way to convey a given message when designing a novel visualization.
Martini Glass Structure: it starts with an author-driven approach as a way to introduce the reader to the story and the functionalities and at the end it goives total freedom to expolore the data further.
Interactive Slideshow: it follows the traditional slideshow structure but it gives freedom to further explore and interact with the data in the scope of each single slide.
Drill-Down Story: it shows a general theme organized by content, and visually by author, but then it gives total freedom to the reader to choose where to drill down to get more information.
How do you put this into practice?
I gave some suggestions already inline together with the descriptions of the various patterns and parts of the model. But let me summarize.
- The table with the design space can help you during the design phase to understand what components you can use in your solution. Choose a genre, a visual narrative tactic, and a narrative structure. You can also think creatively and see if there are useful combinations of these elements that have never appeared before. Alternatively, you can think out-of-the-box, and come up with new solution altogether.
- Be careful with the level of interactivity and messaging you provide. This is probably a choice that you have to take early in your visualization project: do you want to give lots of freedom and risk that the message is not conveyed clearly or you prefer to guide the reader but remove the thrill of exploration?
- The hybrid models are ready-cooked solutions that you can use right away. Again, these can both serve as a way to find a model that suits your message, or just as a starting point to create new hybrid models.
In any case, I hope you enjoyed this post. As I said, given the complexity of this article the post is necessarily reductive. If you are really interested be sure to give a look to the paper itself.
This post has been written by Enrico Bertini. He is a researcher in the visualization and data analysis group at the University of Konstanz. He regularly posts his ideas, reviews, and experiments in his blog fellinlovewithdata.com, where he tries to bridge the gap between academia and the real world out there. If you have any doubts or question you can contact him on Twitter at @FILWD.