Chart Junk
Effective Visualizations Of ethics and junk
March 20, 2012 – Michael Porath Wednesday, March 21, 12
Wednesday, March 21, 12
63 + 60 + 70 = 100?
Wednesday, March 21, 12
By Quarter?
This and following slides adapted from http://mediamatters.org/blog/201006280019 Wednesday, March 21, 12
Actual Job Loss by quarter
Wednesday, March 21, 12
What else? Une m p l o y me n t
This and following slides adapted from http://mediamatters.org/blog/201006280019 Wednesday, March 21, 12
Spacing between data points
15
9 6 Number of months between data points Wednesday, March 21, 12
Actual Unemployment Seasonally adjusted
Wednesday, March 21, 12
Wednesday, March 21, 12
One more
Wednesday, March 21, 12
An honest mistake?
Wednesday, March 21, 12
An honest mistake?
Wednesday, March 21, 12
Actual unemployment rate 9.2 9 8.8 8.6 8.4 8.2 8
Wednesday, March 21, 12
Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Jan Feb
Actual unemployment rate
9.2 9 8.8 8.6 8.4 8.2 8 Wednesday, March 21, 12
Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Jan Feb
Actual unemployment rate Doesn t that look a lot different?
Wednesday, March 21, 12
Edward Tufte A bit outdated... but still true
Graphical Integrity “Tell the truth”
Aesthetics “Do it well”
Wednesday, March 21, 12
Graphical Integrity
Wednesday, March 21, 12
Graphical Integrity According to Tufte
Well designed presentation of interesting data
Complex ideas communicated with clarity, precision and efficiency Wednesday, March 21, 12
Gives the viewer greatest number of ideas in the shortest amount of time, with the least ink and smallest space
Nearly always multivariate
What’s wrong with this picture?
New York Times, August 1978
Wednesday, March 21, 12
What’s wrong with this picture? Lie Factor
Size of effect in graphic Lie Factor = Size of effect in data
5.3in – 0.6in = 7.83 0.6in 27.5 – 18.0 18.0
= 0.53
Lie factor = 14.8
Wednesday, March 21, 12
Cleaned Up Version Better?
Wednesday, March 21, 12
Cleaned Up Version Better?
Wednesday, March 21, 12
Scale Reminds you of something?
Wednesday, March 21, 12
Scale Reminds you of something?
27%
1964 Wednesday, March 21, 12
16%
12%
1975
1990
Scale Reminds you of something?
27%
16%
+33%
+69%
1964 Wednesday, March 21, 12
12%
1975
1990
Principle 1 Don t distort. Use the appropriate number of dimensions (1 dimensional data = 1 dimensional graphic)
Wednesday, March 21, 12
Context is key Connecticut Traffic Deaths
325
before stricter enforcement
300 275 ’55 ’56
Wednesday, March 21, 12
after stricter enforcement
Context is key Connecticut Traffic Deaths, 1951-1959
325 300 275 250 225
Wednesday, March 21, 12
1951
1953
1955
1957
1959
Context is key
Wednesday, March 21, 12
Context is key
Wednesday, March 21, 12
Context is key
Wednesday, March 21, 12
Context is key Comparisons
Incarceration Rate (number of inmates per 100,000) United States 800 600 400 200 0 1992
1995
1998
2001
2004
2007
Data from http://www.prisonstudies.org/info/worldbrief/ Wednesday, March 21, 12
Context is key Comparisons
Incarceration Rate (number of inmates per 100,000) United States
Switzerland
800 600 400 200 0 1992
1995
1998
2001
2004
2007
Data from http://www.prisonstudies.org/info/worldbrief/ Wednesday, March 21, 12
Context is key Comparisons
Incarceration Rate (number of inmates per 100,000) United States
Switzerland
Mexico
800 600 400 200 0 1992
1995
1998
2001
2004
2007
Data from http://www.prisonstudies.org/info/worldbrief/ Wednesday, March 21, 12
Context is key Scale
http://www.flickr.com/photos/philgyford/2932821996/in/photostream/
Wednesday, March 21, 12
Context is key Scale / Baseline
http://www.flickr.com/photos/philgyford/2932821996/in/photostream/
Wednesday, March 21, 12
Context is key Scale / Baseline
Is this more accurate? http://www.flickr.com/photos/philgyford/2932821996/in/photostream/
Wednesday, March 21, 12
Principle 2 Show context and scale. Indicate baseline (not necessarily at 0)
Wednesday, March 21, 12
Aesthetics ...by Tufte s definition
Wednesday, March 21, 12
Data-Ink Ratio Maximizing the use of ink
Data-ink ratio =
Wednesday, March 21, 12
data - ink total ink used
Stripping away “distraction”
–
Wednesday, March 21, 12
=
Stripping away “distraction”
Wednesday, March 21, 12
Stripping away “distraction”
Wednesday, March 21, 12
Data-Ink •
Above all else, show the data
•
Maximize the data-ink ratio
•
Erase non-data-ink
•
Erase redundant data-ink
•
Revise and edit
Wednesday, March 21, 12
Data-Ink •
Above all else, show the data
•
Maximize the data-ink ratio
•
Erase non-data-ink
•
Erase redundant data-ink
•
Revise and edit
h T
? s t h g u o
Wednesday, March 21, 12
Stripping away “distraction”
Wednesday, March 21, 12
Stripping away “distraction”
Wednesday, March 21, 12
Wednesday, March 21, 12
Chart Junk
Wednesday, March 21, 12
Chart Junk An example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk The same example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk The same example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk An example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk Another example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk Another example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk Another example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk Another example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk The Rise in Cigarette Prices
Average retail price per pack of cigarettes, in cents 90 81.9
60 54.3
56.8
1978
1979
60
63
30
0
Wednesday, March 21, 12
1980
1981
1982
Chart Junk The Rise in Cigarette Prices
Average retail price per pack of cigarettes, in cents 90 81.9
60 54.3
56.8
1978
1979
60
63
30
0
Wednesday, March 21, 12
1980
1981
1982
Chart Junk The Rise in Cigarette Prices
Average retail price per pack of cigarettes, in cents 90 81.9
60 54.3
56.8
1978
1979
60
63
30
0
Wednesday, March 21, 12
1980
1981
1982
Chart Junk The Rise in Cigarette Prices
+51%
Average retail price per pack of cigarettes, in cents 90
in 4 years
81.9
60 54.3
56.8
1978
1979
60
63
30
0
Wednesday, March 21, 12
1980
1981
1982
Chart Junk The Rise in Cigarette Prices
+51%
Average retail price per pack of cigarettes, in cents 90
in 4 years
81.9
60 54.3
56.8
1978
1979
60
63
30
0
1980
1981
1982
federal cigarette excise tax doubled in late 1982 Wednesday, March 21, 12
Principle 3 Avoid Chart Junk, but don t be bland
Wednesday, March 21, 12
Causation / Correlation Don t fool yourself.
http://www.businessweek.com/magazine/correlation-or-causation-12012011-gfx.html
Wednesday, March 21, 12
Causation / Correlation Don t fool yourself
http://www.businessweek.com/magazine/correlation-or-causation-12012011-gfx.html
Wednesday, March 21, 12
Principle 4 Don t imply what isn t
Wednesday, March 21, 12
Common Pitfall Population
http://www.npr.org/2011/07/03/137536111/by-the-numbers-todays-military
Wednesday, March 21, 12
Common Pitfall Population
http://www.npr.org/2011/07/03/137536111/by-the-numbers-todays-military
Wednesday, March 21, 12
Common Pitfall Population
http://www.axismaps.com/blog/2008/11/a-new-kind-of-election-map/ Wednesday, March 21, 12
Recap
Wednesday, March 21, 12
Show & Tell
Wednesday, March 21, 12
Labs Assessing effectiveness of Infographics
Wednesday, March 21, 12
http://www.flickr.com/photos/goodmagazine/3576589207/sizes/l/in/photostream/
Wednesday, March 21, 12
http://www.informationisbeautiful.net/visualizations/colours-in-cultures/
Wednesday, March 21, 12
http://www.economist.com/node/21524821 Wednesday, March 21, 12
https://votizen-upvote.s3.amazonaws.com/blog/facebook.jpg
Wednesday, March 21, 12
http://blog.massivehealth.com/post/19358238102/are-bananas-really-as-bad-for-you-as-cookies
Wednesday, March 21, 12
http://www.nytimes.com/interactive/2010/07/02/sports/soccer/facebook-worldcup.html Wednesday, March 21, 12
March 20, 2012 – Michael Porath Wednesday, March 21, 12
Wednesday, March 21, 12
63 + 60 + 70 = 100?
Wednesday, March 21, 12
By Quarter?
This and following slides adapted from http://mediamatters.org/blog/201006280019 Wednesday, March 21, 12
Actual Job Loss by quarter
Wednesday, March 21, 12
What else? Une m p l o y me n t
This and following slides adapted from http://mediamatters.org/blog/201006280019 Wednesday, March 21, 12
Spacing between data points
15
9 6 Number of months between data points Wednesday, March 21, 12
Actual Unemployment Seasonally adjusted
Wednesday, March 21, 12
Wednesday, March 21, 12
One more
Wednesday, March 21, 12
An honest mistake?
Wednesday, March 21, 12
An honest mistake?
Wednesday, March 21, 12
Actual unemployment rate 9.2 9 8.8 8.6 8.4 8.2 8
Wednesday, March 21, 12
Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Jan Feb
Actual unemployment rate
9.2 9 8.8 8.6 8.4 8.2 8 Wednesday, March 21, 12
Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Jan Feb
Actual unemployment rate Doesn t that look a lot different?
Wednesday, March 21, 12
Edward Tufte A bit outdated... but still true
Graphical Integrity “Tell the truth”
Aesthetics “Do it well”
Wednesday, March 21, 12
Graphical Integrity
Wednesday, March 21, 12
Graphical Integrity According to Tufte
Well designed presentation of interesting data
Complex ideas communicated with clarity, precision and efficiency Wednesday, March 21, 12
Gives the viewer greatest number of ideas in the shortest amount of time, with the least ink and smallest space
Nearly always multivariate
What’s wrong with this picture?
New York Times, August 1978
Wednesday, March 21, 12
What’s wrong with this picture? Lie Factor
Size of effect in graphic Lie Factor = Size of effect in data
5.3in – 0.6in = 7.83 0.6in 27.5 – 18.0 18.0
= 0.53
Lie factor = 14.8
Wednesday, March 21, 12
Cleaned Up Version Better?
Wednesday, March 21, 12
Cleaned Up Version Better?
Wednesday, March 21, 12
Scale Reminds you of something?
Wednesday, March 21, 12
Scale Reminds you of something?
27%
1964 Wednesday, March 21, 12
16%
12%
1975
1990
Scale Reminds you of something?
27%
16%
+33%
+69%
1964 Wednesday, March 21, 12
12%
1975
1990
Principle 1 Don t distort. Use the appropriate number of dimensions (1 dimensional data = 1 dimensional graphic)
Wednesday, March 21, 12
Context is key Connecticut Traffic Deaths
325
before stricter enforcement
300 275 ’55 ’56
Wednesday, March 21, 12
after stricter enforcement
Context is key Connecticut Traffic Deaths, 1951-1959
325 300 275 250 225
Wednesday, March 21, 12
1951
1953
1955
1957
1959
Context is key
Wednesday, March 21, 12
Context is key
Wednesday, March 21, 12
Context is key
Wednesday, March 21, 12
Context is key Comparisons
Incarceration Rate (number of inmates per 100,000) United States 800 600 400 200 0 1992
1995
1998
2001
2004
2007
Data from http://www.prisonstudies.org/info/worldbrief/ Wednesday, March 21, 12
Context is key Comparisons
Incarceration Rate (number of inmates per 100,000) United States
Switzerland
800 600 400 200 0 1992
1995
1998
2001
2004
2007
Data from http://www.prisonstudies.org/info/worldbrief/ Wednesday, March 21, 12
Context is key Comparisons
Incarceration Rate (number of inmates per 100,000) United States
Switzerland
Mexico
800 600 400 200 0 1992
1995
1998
2001
2004
2007
Data from http://www.prisonstudies.org/info/worldbrief/ Wednesday, March 21, 12
Context is key Scale
http://www.flickr.com/photos/philgyford/2932821996/in/photostream/
Wednesday, March 21, 12
Context is key Scale / Baseline
http://www.flickr.com/photos/philgyford/2932821996/in/photostream/
Wednesday, March 21, 12
Context is key Scale / Baseline
Is this more accurate? http://www.flickr.com/photos/philgyford/2932821996/in/photostream/
Wednesday, March 21, 12
Principle 2 Show context and scale. Indicate baseline (not necessarily at 0)
Wednesday, March 21, 12
Aesthetics ...by Tufte s definition
Wednesday, March 21, 12
Data-Ink Ratio Maximizing the use of ink
Data-ink ratio =
Wednesday, March 21, 12
data - ink total ink used
Stripping away “distraction”
–
Wednesday, March 21, 12
=
Stripping away “distraction”
Wednesday, March 21, 12
Stripping away “distraction”
Wednesday, March 21, 12
Data-Ink •
Above all else, show the data
•
Maximize the data-ink ratio
•
Erase non-data-ink
•
Erase redundant data-ink
•
Revise and edit
Wednesday, March 21, 12
Data-Ink •
Above all else, show the data
•
Maximize the data-ink ratio
•
Erase non-data-ink
•
Erase redundant data-ink
•
Revise and edit
h T
? s t h g u o
Wednesday, March 21, 12
Stripping away “distraction”
Wednesday, March 21, 12
Stripping away “distraction”
Wednesday, March 21, 12
Wednesday, March 21, 12
Chart Junk
Wednesday, March 21, 12
Chart Junk An example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk The same example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk The same example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk An example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk Another example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk Another example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk Another example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk Another example
examples from Bateman et al., 2010 Wednesday, March 21, 12
Chart Junk The Rise in Cigarette Prices
Average retail price per pack of cigarettes, in cents 90 81.9
60 54.3
56.8
1978
1979
60
63
30
0
Wednesday, March 21, 12
1980
1981
1982
Chart Junk The Rise in Cigarette Prices
Average retail price per pack of cigarettes, in cents 90 81.9
60 54.3
56.8
1978
1979
60
63
30
0
Wednesday, March 21, 12
1980
1981
1982
Chart Junk The Rise in Cigarette Prices
Average retail price per pack of cigarettes, in cents 90 81.9
60 54.3
56.8
1978
1979
60
63
30
0
Wednesday, March 21, 12
1980
1981
1982
Chart Junk The Rise in Cigarette Prices
+51%
Average retail price per pack of cigarettes, in cents 90
in 4 years
81.9
60 54.3
56.8
1978
1979
60
63
30
0
Wednesday, March 21, 12
1980
1981
1982
Chart Junk The Rise in Cigarette Prices
+51%
Average retail price per pack of cigarettes, in cents 90
in 4 years
81.9
60 54.3
56.8
1978
1979
60
63
30
0
1980
1981
1982
federal cigarette excise tax doubled in late 1982 Wednesday, March 21, 12
Principle 3 Avoid Chart Junk, but don t be bland
Wednesday, March 21, 12
Causation / Correlation Don t fool yourself.
http://www.businessweek.com/magazine/correlation-or-causation-12012011-gfx.html
Wednesday, March 21, 12
Causation / Correlation Don t fool yourself
http://www.businessweek.com/magazine/correlation-or-causation-12012011-gfx.html
Wednesday, March 21, 12
Principle 4 Don t imply what isn t
Wednesday, March 21, 12
Common Pitfall Population
http://www.npr.org/2011/07/03/137536111/by-the-numbers-todays-military
Wednesday, March 21, 12
Common Pitfall Population
http://www.npr.org/2011/07/03/137536111/by-the-numbers-todays-military
Wednesday, March 21, 12
Common Pitfall Population
http://www.axismaps.com/blog/2008/11/a-new-kind-of-election-map/ Wednesday, March 21, 12
Recap
Wednesday, March 21, 12
Show & Tell
Wednesday, March 21, 12
Labs Assessing effectiveness of Infographics
Wednesday, March 21, 12
http://www.flickr.com/photos/goodmagazine/3576589207/sizes/l/in/photostream/
Wednesday, March 21, 12
http://www.informationisbeautiful.net/visualizations/colours-in-cultures/
Wednesday, March 21, 12
http://www.economist.com/node/21524821 Wednesday, March 21, 12
https://votizen-upvote.s3.amazonaws.com/blog/facebook.jpg
Wednesday, March 21, 12
http://blog.massivehealth.com/post/19358238102/are-bananas-really-as-bad-for-you-as-cookies
Wednesday, March 21, 12
http://www.nytimes.com/interactive/2010/07/02/sports/soccer/facebook-worldcup.html Wednesday, March 21, 12
