The Gestalt of Data Visualization

Perception Lab

We see groups, paths and shapes before we read individual values. Those habits of perception help charts communicate—and sometimes suggest relationships the data does not contain. This interactive version of a 2015 essay series shows five principles at work. Change each view to see what your eye groups together.

How Marks Become Groups

Similarity, proximity, and enclosure all signal category. Once you name the intended signal — even something as plain as shared color — you also start noticing the groups a graphic creates by accident.

Fig. 1

Color, distance, and boundary establish a group

All forty marks share the same neutral gray, so the field has no grouping signal yet.

Revelation. Watch the transition between steps. The order in which marks move and recolor is itself a signal — the original essay called this "revelation." Animated charts carry memory of prior position and color, so sequencing has to be deliberate.

The Gestalt Gaze

Check the relationships your design suggests.

Nearby marks look related. A strong color draws attention. A transition can make two events seem connected. Look for those signals when reviewing a chart, then adjust the grouping, emphasis and movement to support the comparison you want readers to make.

A transition of Elijah Meeks's 2015 essaysGestalt Principles for Data Visualization (§1,§2,§3,§4) — bespoke D3 then, Semiotic charts now. Approval data from the American Presidency Project at UCSB and Gallup.