How Mondrian minimalism can help us design more actionable dashboards

?? Opinion. We frequently over-engineer our visualizations.

Labels, tooltips, annotations, x-axis ticks, y-axis ticks...

There is of course a valuable place and time for all of these, but there's also a chance that the chart you've created is so busy that it distracts the consumer from the information they really need to quickly make a decision.

Take this example: You're looking at student performance on questions on an assignment, and you want to quickly make a decision about which questions might need reviewing. (Maybe they're too difficult, too easy, or simply worded unclearly.)

To ground yourself, first ask, "What information are consumers primarily wanting to get from this visualization, and how will that drive their interaction(s) with it?".

If consumers are using a visualization to quickly determine whether something belongs to one of two categories (like "ok" vs. "action-needed"), clogging that visualization with labels, ticks, and numbers is counterproductive.

From there, it's form follows function.

Mondrian minimalism in action

Here's a fun little visualization, inspired by Mondrian's art.

Let's imagine we're looking at a summary of student performance on each problem in a homework assignment. Each student can have one possible outcome:

1. Correct on the first try (dark blue)
2. Correct after multiple tries (light blue)
3. Incorrect after multiple tries (red)

We could have implemented this color scheme differently, of course, or just focused on 2 distinct outcomes (correct first try vs correct after many tries), but let's go with this scheme.

Now, if we look at performance by item, we'll see a few things that stand out quickly:

1. Questions with more red are likely questions that students are struggling with. Or possible questions that test constructs that students are struggling to grasp.
2. Questions with more light blue may be confusingly-worded and thus could stand to be reviewed for clarity. If
3. Questions with a high proportion of dark blue might be just right or potentially too simple. This depends on the purpose of the assessment, the level of the students taking the assessment, etc.

??????All of these actionable take-aways were derived without need of y-axis ticks, y-axis labels, or any tooltips crowding the visualization.

What perceptual principles are at play here?

There are a couple of nifty perceptual principles at work here, which play together nicely with the Mondrian-inspired layout.

The principle of proximity holds that things which are grouped together are more likely to be perceived as related. To leverage this in the visualization, problems which belong to the same set (A, B, C, ...) are therefore grouped more closely, with larger gaps between sets (1,2,3, ...).

The Von Restorff (aka "isolation") effect is also at work here, to the extent that outlying distributions of colors attract our attention, particularly in the ratio of red to blue. This facilitates the identification of outliers.

Take-aways

The point of this micro case study was drive home a simple contention — that for a fundamentally qualitative objective (distinguishing between categories), you don't need a crowded, over-annotated visualization.

Thank you kindly for reading!

If you'd like to see more Mondrian, more minimalism, and more Mondrian minimalism, please take a peek at my portfolio.