Most major news organizations are using RIAs in their election guides this year. The common factors are: one page design (multiple summary views with drill-down options), animated transitions from one piece of information to the other, and exceptional graphic quality. Anyone working on using RIAs to visualize large data sets should have a look at these various sites and how they approached the problem. What follows is a quick review!
RIA technology is superbly suited to rendering this kind of large data set. What you typically want to do with a data set is explore it, drilling down into different sections of the data or turning it around to see different summary views of the data. This interaction pattern doesn’t work if there’s system latency: during a pause of more than a second or so, users lose attentional focus, and forget the information from the data view they just saw. This is why MacroMedia always mentions “executive dashboards” as a primary use case for RIA technology.
Check out the animated transitions of the US map in the new york times election guide (click the “view map according to” buttons to see transition). The animation helps the user from becoming disoriented during the transition between the views. This is animation used for good, rather than for evil.
The Washington Post uses a nice tabbed view to show the results for President, Senate, and Governor races for the whole country. Tabs are a good way of showing multiple views of data, (as long as the shape of the visualization itself doesn’t change).
The LA Times has an amazing visualization that also lets you run scenarios. Click here, (then click on the links “Presidential Poll Map”, “Senate Poll Map”). It starts out with a map rendering the latest polling data. Users can drill down on data by mousing over a given state (the data appears in a table below the map). Most intriguingly, you can run scenarios by “calling” states for various candidates. What’s the electoral math like if Bush wins Florida? Ohio? Surprisingly addictive. This visualization moves beyond simply rendering data, allowing the user to interact with and overwrite the data.
CNN’s voter guide doesn’t support exploration of the data very well. Drillng down on state data requires a page refresh, which disorients the user and removes the context of the visualization. I gave up playing with it after looking at only a few states. On the other hand, it’s static view has more data than the other pages: this guide is better used as a static web site than as a way of exploring the data in real time.
One important design issue is raised by these visualizations: it’s really hard to tell how old the data being rendered is. RIAs can go back to the server go get fresh data behind the scenes: but I have no idea whether or not this is happening. Users may not expect dynamically updating content on a web site, given their experience with html-based web pages. Only the CNN visualization explicitly labels the age of the data. RIAs that visualize a dynamic data stream should always make clear how old the data they are rendering is, and where it came from.
What did you think of the various electoral guides? Are there any other RIAs rendering this data that designers should check out? As always, feel free to post below?
UPDATE: Read Flash only winner in US Election and markme for more Flash map goodness.