Welcome to episode 36 of Data Viz Today. Have you ever pushed the boundaries of visualization? Did you receive any push-back? Do you want to experiment more with new chart types, but you’re not sure where to start or maybe you’re worried about people’s reactions? In this episode, we’ll hear how Richie Lionell created his thought-provoking data viz, how he handled criticism gracefully, and how you can get started creating something new in spite of potential negative feedback.
Welcome! I'm Alli Torban.
00:23 - Today’s episode is all about experimenting with visualization and in turn receiving feedback and critiques on it.
01:30 - He first came across senate voting data on the senate.gov website and was super excited to find it in a clean xml format, so he scraped the xml files and converted it to CSV.
01:42 - His very first thought was to find out how each senator’s voting pattern compared to the other senators. So he wrote a simple python script to compute the simple similarity score for every senator with every other senator. He did this by taking a sentor’s vote (yea or nay) for each issue and comparing it to every other sentor’s vote.
02:25 - Initially he tried conveying this story through a network layout, but it turned out to be really cluttered and hard to read. So he had start making some tough choices. He knew he couldn’t show all the information because it was just too complex, but it still needed to be interesting for people to explore.
02:51 - His big breakthrough came when he decided not to show a many-to-many network relationship and ditch the traditional network layout, and instead try a radial representation where one chosen senator would be placed at the center, and the other senators would be placed around him or her based on how similar their voting records were.
03:25 - So in his final visualization, Richie used Gramex to handle data & the user interface, and, D3.js for the visual representations.
04:28 - Like I mentioned, some people seemed to really like this viz and some people really did not like it. The biggest complaint seemed to be that the position of the senators around the circle was random - Richie just randomly positioned them around the circle so they wouldn’t overlap.
05:10 - Some people were a bit more harsh and found it difficult to comprehend the radial layout and thought that it was completely useless since the position around the circle didn’t mean anything.
05:20 - But Richie took it all in stride and was glad that it sparked debate, and actually found it really insightful to hear how different people perceived the viz - some people found it really hard to read and some people found it really intuitive, which gives us insight into how people understand visualizations.
05:33 - In the end, Richie was happy to have tried something new and achieved his goal of visualizing one interesting theme… he wasn’t trying to answer all the questions. But he did wrestle with creating a rich graphic that was still readable. It’s a tough balancing act that all data visualization designers have to contend with.
05:55 - So how can you experiment with new visualization techniques and how can you prepare yourself for the inevitable critique?
06:01 - I really loved watching Maarten Lambrecht’s OpenViz talk about Xenographics - he created an entire website where he compiles new and strange chart types, and in his talk he gave some tips on how you can create your own xenographics. One tip was take your chart and just flip the axes, another tip is to crossbreed two different chart types. Pick two that show the info you want to convey and try to integrate them into one chart.
06:35 - While I was on his xenographics site, I came across the chart type called the solar correlation map, which reminded me a lot of Richie’s viz. The idea is to use the solar system as a metaphor for a chart where you place a variable in the center and then place other variables at varying distances from the center determined by how correlated the variable is to the center variable. In the article introducing the solar correlation map by By Stefan Zapf and Christopher Kraushaar, they offer some tips on how to create a new visualization
Identify a problem in data analysis
Find an analytical tool that solves this problem
Use a visual metaphor to explore and communicate your results
07:25 - I dive more into visual metaphors in episode 14 if you’re interested in hearing more about that.
07:33 - So say you try out a new chart type and the critiques start rolling in… first, be prepared for it. There’s a reason why Maarten calls new charts xenographics - some people are scared of new charts and will automatically dislike it because it’s different. Second, keep in mind that creating something new is hard and most people completely underestimate the creativity and effort involved in it.
08:00 - I think following Richie’s mindset is the best way forward following critiques - know that some people will find it difficult to understand because it’s new, but their critiques are still valuable even if they don’t seem to understand what your goal was, because it’ll give you insight into how people are making sense of your visualization, which will help you in the future.
08:20 - My final takeaway is that we need people to experiment with new visualization techniques and chart types, and it’s tough being a pioneer - it takes creativity and effort, but it’s important to keep the data viz field moving forward. Just be ready to hear feedback on it, and try to take it as insight for your next viz. If you’re giving feedback, remember to critique respectfully.
08:52 - Listen for Richie’s advice to designers just starting out!
09:45 - Did you know? You can sign up for my newsletter that I send out every Sunday with a quick recap of the top tips from the last episode to help commit it all to memory, or to give you the highlights in case you missed the episode. :)