This article is more than 5 years old.

On March 21st, Chris and I got up really early (can’t say “bright and early” because it was before sunrise) and went to Chapel Hill for the 25th North Carolina Serials Conference. While there were a number of interesting sessions (including one on using the free tool OpenRefine to manipulate metadata), I’m going to focus on just one session, which I think might be of the most general interest to everyone in the library, because it doesn’t just focus on serials or metadata. The session, presented by Megan Kilb and Matt Jansen of UNC-Chapel Hill, was called “Visualizing Collections Data: Why Pie Charts Aren’t Always the Answer,” and it offered tips and advice on how to present data.

The presentation grew out of their need to evaluate the TRLN consortial deal on Springer e-resources. They found that pie charts aren’t always (actually are almost never) the best way to present data, which matches research that has shown pie charts to be sub-optimal for human comprehension. Pie charts get confusing if they have more than 4 or 5 categories, they treat everything as a proportion, they make readers have to compare areas/angles, and the values are only available via labels.

Research into the accuracy of human interpretation of graphical data is on a continuum. The most intelligible graphical data from most accurate to least accurate is:

  1. Position
  2. Length
  3. Angle/Slope
  4. Area
  5. Volume
  6. Color/Density

With this info in mind, alternatives to the pie chart might be the bar chart (because length is easier to perceive), or a simple table (if you have only a few values to consider). Regarding other graphical representations of data, if you have a graph, be aware that backgrounds, particularly lines, can be distracting. Lines and other detail can make it hard to read values of dots on the lines. Stacked charts (charts with multiple jagged lines, each representing different values) can also be problematic, because there may be confusion over what the overall height of columns mean. They require the user to do visual math, which is difficult. Alternatives to this might be to make lines next to each other (rather than on top of each other), or represent each line as a slope, which emphasizes different rates of change.

In addition to attending the conference, I also represented NASIG (as Past President) at our sponsor table, giving out literature and talking up the organization.