Data VISUALIZATION is an art form that can increase the impact of your presentation exponentially. But what makes a given visualization memorable or effective?
This is precisely the question that Michelle Borking of Harvard's School of Engineering and Applied Sciences set off to answer in “the largest scale visualization study to date.”1
Along with collaborators from Harvard and MIT, Borkin studied more than 2,000 visualizations from a variety of sources, including infographic sources, government and world organizations, news media, and scientific publications.
The researchers categorized the visualizations according to structural type (e.g., bar, diagram, table, grid, etc.) and further encoded the sample according to a range of visual properties (dimension {2D, 3D}, pictorial properties, multiplicity, etc.) and attributes (black & white, number of colors, visual density, etc.).
Then they selected a subset of 410 “target” visualization images, 145 of which were extreme “minimalist” (i.e., “data-ink” = good), 103 were extreme “chart junk” (i.e., data-ink ration = “bad”) and 162 were somewhere in between on the continuum.
To see which images were the most intrinsically memorable, the researchers conducted an experiment set up as a game on Amazon Mechanical Turk, “where workers were presented with a sequence of images, and had to press a key if they saw an image for the second time in a sequence” (p. 5).
The repeating images came from the “target” visualizations subset and were interspersed among multiple “filler” images.
The study found that the most memorable images have the following characteristics:
1.
2.
3.
4.
Your Presentation &
the IMAGE
10