News and Notes from PolicyViz - Issue #24

Nov 09, 2022

Hi all,

As I finish this newsletter on Wednesday morning, November 9, 2022, I'm reminded that the political divisions in the United States are as stark as ever. But I'm relieved just a bit that democracy seems to be holding. At least for now.

As for things a little bit closer to the data visualization world, the thing keeping my attention is what is Elon Musk going to do with Twitter? There is a lot of competing information out there and lots of hot takes about what he might do with the platform and what that might mean for regular users.

At the moment, my take--as you can see in the Tweet thread I published earlier this week--is that I'm taking a wait and see attitude. If the racism, misogyny, anti-semitism, and other hate is unleashed, I'll be on my way out. But if it's just some paid features (no, I'm not paying for a blue checkmark) and more advertisements, I'll probably hang around. And while others are starting to move more into Mastodon, I just don't see myself using yet another social media platform, so in that case, I'll rely more on this newsletter, my website, and my Winno community.

Jon Schwabish @jschwabish

I've spent the last few days quietly watching the conversation in the #dataviz community about how Musk's purchase of Twitter may impact our community. I've been thinking and reflecting on my own use of the platform and how I might use it going forward. 1/n

The other thing I'm working on is taking some time off in December. It's been a hell of a year and I need a bit of a break. I'm not totally disconnecting until the last two weeks of the year, so I'm hoping to spend the first two weeks catching up on some blog writing, video making, and exercising. So stay tuned for more of your favorite dataviz content coming soon!

Thanks,

Jon

DRAFT POST: Getting away from the ‘this is how we always do it’ mentality

When I teach data visualization classes and workshops, I present the wide array of graphs, charts, and diagrams that are available to us when we communicate our data. I argue that these “non-standard” graph types can help us be more effective at communicating our data for at least two reasons: First, they are sometimes inherently better at showing the data than the standard chart types like bar, line, and pie charts. Second, they are sometimes more engaging or fun to look at and explore than standard charts, and sometimes engagement can be a goal of our work. If you’ve read my book Better Data Visualizations, this argument won’t surprise you.

One of the ways people push back against learning new graph types is that they say their manager, colleague, reader, or audience member will never understand how to read them. “This is how my boss always looks at the data,” they’ll say, “and they know where to look to find the number they need. They’ll balk if I use a different or new graph that they don’t immediately understand.”

One of the problems with this approach, however, is that what if something important is going on somewhere else in the graph, chart, or table? What happens if the manager immediately goes to the fifth bar chart in the report and ignores everything else? Different, more engaging graphs can sometimes help show your data to more effectively highlight patterns, trends, and values they might otherwise miss.

How do you break through the “this is how I always look at it” mentality? My experience has always been to show your reader or user the difference between the original and your new, proposed graph. If you can help them see the improvement from one visualization type to another, they will then understand the value of changing. And once they understand how to read the new plot type, it becomes part of their data visualization toolbox.

Take this heatmap from Nathan Yau at Flowing Data, which he published back in 2012. He uses a heatmap arranged as a calendar to show auto fatalities in 2012. Here, January is at the top and December at the bottom; Sundays on the left and Saturdays on the right. In my experience, most people can quickly and easily pick out the pattern of more deaths on the weekends. Some, but not all, can also see the seasonal change (more deaths in the summer months) relatively quickly as well.

Calendar heatmap from Nathan Yau at Flowing Data

Now let’s look at the same data as a more standard line chart with orange dots denoting Saturdays. Here, it’s almost impossible to get the ‘more deaths on the weekend’ story, though perhaps a bit easier to see the seasonal pattern.

In this case, the heatmap appears to be a better visualization: You can see the patters more clearly, it is easier to add labels and annotation, and, personally, I find it to be more pleasing to look at.

We are not born knowing how to read a bar chart or line chart or pie chart. We learn how to read those graphs through experience and exposure. The only reason why someone might not immediately like a slope chart, dot plot, connected scatterplot, or so many other graphs not immediately available in your tool’s drop-down menu is because they haven’t learned how to read it.

So, use good annotation in your graph. Use good, active titles to help them understand what they are supposed to learn from the graph. And include additional pointers, highlights, and other elements to help them understand how to read the graph and then how to understand the content delivered in the graph.

I’m a firm believer that we can help our readers and users move beyond the “this is how I always look at it” mentality to create and use a wider array of graphs, charts, and diagrams to enhance understanding and be more effective data communicators. Demonstrate how your new approach yields better results and greater insights to your data.

Please let me know if you have thoughts on this blog post. I post drafts here and then put them on my site about a week later.