Information Visualization

facilitated by John Emerson

Introduction

What is information visualization? information visualization tells a specific story with graphics, whereas data visualization is more exploratory, and about finding patterns in the data.

"Picture superiority effect" - images with words have a greeted impact than words alone.

We see difference in particular ways. For example, we notice a line that's twice as long more than we notice a color that's twice as dark.

Examples of information visualization

Annual reports: heavy use of graphics to engage and make a point.
- Lets funders see that their money is working.
Comic books (e.g. "I got arrested. Now what?")
Brochures with schematics a la IKEA, that bypass language barriers (e.g. Vendor Power!).
Maps with walking tours to tell history.
Food wrappers with info about the food and the stories it tells about culture (e.g. Conflict Kitchen).
Maps.
- Mapping donors - Mapped zip codes tell a story that's not visible in numbers
- Mapping layers - e.g. overlay lead paint with demographic mapping.
- CrashStat 2.0 (mapping ped/cyclist accidents in NYC)
- Green Maps in NYC
Logging cronyism in Cambodia with royal family (networking viz)
Valdis Krebs
Collusion extension for Chrome
Little Sis
Journal of Aesthetics and Protest
Information is Beautiful blog

Print vs online

What are the demographics of the people you're trying to reach, and how can you reach them? Is your target audience online?
Can you hand someone a website? (hint: no).
Printed material has a certain permanence.
Budget is always a factor.

Tools

Online D3 for building vector graphics (keep design simple, focus on key focus point)

Things to Think About

Do usability testing!

Do usability testing! Do usability testing! Make sure the information is as accessible and obvious as you think it is.
Edward Tufte's rule of thumb: Eliminate "chart junk" (drop shadows, phony 3d-ness) as much as possible. By the way, Tufte teaches classes that are pretty reasonable.

Lies/misleading documents

Representing values with area (inaccurate areas make differences seem greater).
Phony 3d-ness distorts area (e.g. pie charts that appear to be tilted)
Pyramid charts
Zooming in that distorts relation between data sources (e.g. starting the Y-axis of a bar chart at 7 makes the difference between 8 and 9 seem greater than it is).

Subjective intensities of sensations tend to fall on log scales.

Other issues

How to design research questions to get neutral/non-leading questions (e.g. asking about gov't's intentions vs asking about what's actually happening).
Data is the artifact of its collection (not a natural fact).
It is easy to be unethical, and to tell slightly dishonest stories by including or not including information.
Question about making a specific point vs. just giving info andletting the user decide.
Data to frame the problem vs to propose a solution
Paradox of the more data leads to less scrutiny