Professional Documents
Culture Documents
in Political Science:
How to make beautiful data displays
David Sparks
Duke University
Visualization: Three arbitrary categories
• Infographics: Not what I’m talking about
• Primarily factoids
• Statistical Graphics: Make an argument
• Distilled representation of the results of analysis
• Information Visualization: Show the data
• Often high-dimensional, large-N
Visualization: Three arbitrary categories
• Infographics: Not what I’m talking about
• Primarily factoids
• Statistical Graphics: Make an argument
• Distilled representation of the results of analysis
• Information Visualization: Show the data
• Often high-dimensional, large-N
Factoids, not data
http://www.boston.com/bostonglobe/ideas/naps/
Explanations, not data
http://www.dnapolicy.org/resources/Cloning_infographic_final.jpg
Stereotypes, not data
http://www.informationisbeautiful.net/leftvright_world.html
Stereotypes, not data
http://bibliodyssey.blogspot.com
PowerPoint as the enemy
http://docs.house.gov/gopleader/House-Democrats-Health-Plan.pdf
Visualization: Three arbitrary categories
• Infographics: Not what I’m talking about
• Factoids
• Statistical Graphics: Make an argument
• Distilled representation of the results of analysis
• Information Visualization: Show the data
• Often high-dimensional, large-N
Statistical Graphics
• Gelman:
• Make use of familiar forms
• Discovery of the unexpected
• Graphics as alternative to tables and models
• Distilled, reduced, minimized
• Make an argument
• Don’t mislead
Very unnecessary graph, use a table
http://www.neuroscience.com/graphic.html
One number, in text
http://www.nevron.com/
Superfluity
http://mediamatters.org/blog/201007010030
Bad math
http://eagereyes.org/blog/2009/bring-out-your-dreadful-charts
Good color
http://hdr.undp.org/en/reports/global/hdr2009/
Good graphs: Model reporting
Good graphs: Model reporting
Good graphs: Model reporting
Good graphs: Model reporting
Good graphs: Model reporting
Visualization: Three arbitrary categories
• Infographics: Not what I’m talking about
• Factoids
• Statistical Graphics: Make an argument
• Distilled representation of the results of analysis
• Information Visualization: Show the data
• Often high-dimensional, large-N
Information Visualization
• Gelman:
• Creative, cutting-edge
• Visual puzzles
• Alternatives to photos and cartoons
• High density often necessary for multiple
dimensions
• Macro- and micro-level information
• Showing data, not presenting results
Network ties
• Inherently multidimensional
arbitrarian.wordpress.com
Macro impressions
miller-mccune.com
Micro detail
A story in multiple dimensions
Challenge: To illustrate an institutional history
through the movement of its agents
http://www.duke.edu/~dbs9/senatehistory.html
Closer view
Specific detail
url citation
Same data, multiple representations
url citation
Movement over time
url citation
Different audience, different approach
Different audience, different approach
A history in multiple dimensions
Challenge: To illustrate political history and
dynamics through electoral results
Don’t be misleading
Don’t be misleading
Don’t be misleading
Don’t be misleading
Don’t be misleading
Visit www.duke.edu/~cdd11/tex.html
Thanks
d.sparks@duke.edu
sites.google.com/site/dsparks