Don’t Hide Good Data Analyses in Difficult Graphs Gary McClelland & Julie Schiro Analyze Boulder 4 June 2014
Florence Nightingale “The Passionate Statistician” 4 June 2014 & 2
John Snow 4 June 2014 & 3
Charles Minard 4 June 2014 & 4
4 June 2014 & 5
3D Column Chart 4 June 2014 & 6
“Height of Green Columns” 4 June 2014 & 7
Blue > Green 4 June 2014 & 8
Both Answers Correct? 4 June 2014 & 9
Green Columns = 5 4 June 2014 & 10
Same data in 2D 4 June 2014 & 11
But even here… High contrast where not needed Low contrast where discrimination needed 4 June 2014 & 12
Scatterplots from R 4 June 2014 & 13
However DOTS 4 June 2014 & 14
What is the Relationship? 4 June 2014 & 15
4 June 2014 & 16 Not at all worth it Very worth it Neither helpful nor hurtful
Identical graphs, except the red cluster appears in the lower left or upper right 4 June 2014 & 17
Without red clusters - identical 4 June 2014 & 18
With “red” clusters 4 June 2014 & 19
Regression Line Helps 4 June 2014 & 20
Punchlines “It’s not you, it’s me”: If the decision maker can’t understand your graph, don’t blame the decision maker. Make a better graph. Unless there really is a 3 rd dimension that you really need to display, avoid 3D graphs. Enhance scatterplots and other graphics from stat programs to help the viewer focus on all of the data, not just some of it. 4 June 2014 & 21
4 June 2014 & 22 May you make great graphs! Thank you! Gary & Julie
Useful Links & Addresses Gary’s ground-breaking but now aging online interactive statistics textbook using Java applets: Gary’s data analysis textbook: Gary’s data visualization consulting firm: Java applets for teaching statistical concepts: [the magic words are ‘model’ and ‘error’] Steven Johnson’s TED talk about John Snow’s “ghost map”: Michael Friendly’s data visualization website, the “Milestones Project” describes and illustrates the ‘Golden Age of Data Graphics’: Edward Tufte is usually considered the guru of modern data graphics. His website has LOTs of resources: This presentation: 4 June 2014 & 23