data presentation adapted by joanna wolfe from marianne w. zawitz, bureau of justice statistics,...
TRANSCRIPT
Data Presentation
Adapted by Joanna Wolfe from Marianne W. Zawitz, Bureau of Justice Statistics, October 11, 2000
Presenting effective Tables and Figures
This presentation covers the principles of creating good technical graphs and tables.
1. Provide descriptive captions.
2. Consider your “story” and audience when presenting data.
3. Avoid non-data ink.
4. Be ethical: Persuade but do not deceive.
Always label tables and figures.
Figure 2: Recurrence of Gastric ulcers for the year after successful healing with ranitidine alone or triple therapy plus ranitidine. No maintenance was given.
Per
cent
wit
h U
lcer
Rec
urre
nce
Weeks
Graphs excel at emphasizing trends and relationships while tables display exact values in a small amount of space.
WeeksRanitidine
aloneTriple
therapy
4 8% 0%
8 20% 5%
16 42% 10%
24 60% 10%
Table 1: Recurrence of Gastric Ulcers by Therapy Type
Fig 2: Recurrence of Gastric Ulcers by Therapy Type
Line graphs are good for communicating change over time.
Bar graphs are good for communicating differences between values.
Pie charts are good for simple messages.
Bureau of Justice Statistics
Pie charts are bad for complex data or when precision matters.
Which slice is biggest?
Which is smallest?
By how much?
Bar charts allow readers to visualize differences between variables.
Which bar is biggest? Which is smallest?
By how much?
0 5 10 15 20 25
A
B
C
D
E
F
G
A 21B 19C 18D 14E 12F 9G 7
Tables allow readers to see EXACT differences between variables and exact values.
Which value is biggest? Which is smallest?
By how much?
The visualization expert Edward Tufte says "the only thing worse than a pie chart is two pie charts."
Which military branch grew the most between 1990 and 1998?
Bar charts simplify comparisons.
Which military branch grew the most between 1990 and 1998?
A 21B 19C 18D 14E 12F 9G 7
Which visualization is best for a technical audience?
0 5 10 15 20 25
A
B
C
D
E
F
G
Summary so far:
• Always supplement tables and figures with text: label axes and provide an informative caption.
• Use bar charts or tables to compare values.• Use line graphs to show trends.• Use tables when exact values are important.• Only use pie charts for simple data and non-
technical audiences.
Strive for simple designs that show the data.
Avoid non-data ink.
Avoid 3D.
1986 1987 1988 1989 19900
20
40
60
80
100
120
140
160
Administrative forfeituresMillions
3D charts add non-data ink and are more difficult to read than 2D charts.
Is 1987 above or below 60 million? What about 1988? 1990?
1986 1987 1988 1989 19900
20
40
60
80
100
120
140
160
Administrative forfeitures
Millions
Hidden data is a common problem with 3D presentations of 2D data
Bureau of Justice Statistics
A simple line chart in 2 dimensions avoids hiding data
West
South
NortheastMidwest
Bureau of Justice Statistics
The purpose of a chart is to convey data so avoid non-data ink.
Grids Ticks Frames Anything that does not convey data
1989 1990 1991 1992 1993 1994 1995 19960
10
20
30
40
50
60
70
Bureau of Justice Statistics
1989 1990 1991 1992 1993 1994 1995 19960
10
20
30
40
50
60
70
13-14
15
16
17
18 years old
Age of offender
Bureau of Justice Statistics
Excess tick marks can make charts hard to read.
13-14
15
16
17
18 years old
Age of offender
Bureau of Justice Statistics
Removing tick marks make trends easier to see.
ImplementationBefore implementation
After implementation
Average vehicle speed 45.5 mph 45.7 mphStandard deviations in vehicle speed
9.4 mph 7.2 mph
Implementation Before AfterAverage vehicle speed 45.5 mph 45.7 mphStandard deviations in vehicle speed 9.4 mph 7.2 mph
Tables also benefit from minimizing non-data ink.
Compressor Efficiency score
LZOP 1
PAR 1
PKZIP 1
JAR .99
WINACE .98
WINZIP .97
STUFFIT .90
DEEPFREEZE .91
HUFFMANCE .91
OPAQUE .71
BLINK .58
Tables also benefit from minimizing non-data ink.
Compressor Efficiency score
LZOP 1
PAR 1
PKZIP 1
JAR .99
WINACE .98
WINZIP .97
STUFFIT .90
DEEPFREEZE .91
HUFFMANCE .91
OPAQUE .71
BLINK .58
Criminal history of detained vs. released felony defendants, 1990
Series1
0 10 20 30 40 50 60 70 80 90
Detained
Released
Percent of felony defendants
Prior criminal history
Arrest
Any conviction
Felony conviction
Violent felony conviction
Avoid hatched fill patterns, which are hard to look at.
Se-ries1
0 10 20 30 40 50 60 70 80 90
Percent of felony defendants
Weapons arrest rate by age of offender
0
250
500
750
1000
1250
1500
1750
2000
2250
1965 1970 1975 1980 1985 1990
17-1815-1613-14
Stacked bars are also confusing.
Bureau of Justice Statistics
"Simplicity of design and complexity of data"
Edward R. Tufte