Presentation Graphics
Pie Charts
In the period 1910 1920 there was a great deal of
discussion or the relative merits of pie charts and
divided bar charts in the Journal of the American
Statistical Society.
Eventually consensus was reached that divided bar
charts were a superior way of presenting proportions.
Since 1980, the study of graphical of graphical
perception has revealed why bar charts are preferable.
Humans are much better at decoding numbers presented
in the form of lengths or positions than they are at
decoding numbers presented as angles or areas.
Comments on Bar Charts I
Becker R., and Cleveland W. S. (1996).
The Splus Trellis Graphics User Manual.
Page 50.
Pie charts have severe perceptual problems.
Experiments in graphical perception have shown
that compared with dot charts, they convey
information much less reliably. But if you want to
display some data, and perceiving the information
is not so important, then a pie chart is fine.
Bill Cleveland is one of the worlds foremost authorities on
how information is extracted from graphs.
Comments on Pie Charts II
Tufte, E. (1983).
The Visual Display of Quantitative Information.
Page 178.
A table is nearly always better than a dumb pie
chart; the only worse design than a pie chart is
several of them, for then the viewer is asked to
compare quantities located in spatial disarray
both within and between pies . . . Given their low
data-density and failure to order numbers along a
visual dimension, pie charts should never be used.
Ed Tufte was Professor of Statistics, Political Science and
Graphic Design at Yale University. He has written some of the
best-selling books on information display.
Comments on Pie Charts III
Bertin, J. (1981).
Graphics and Graphic Information Processing.
Page 111.
Bertin describes multiple pie charts as
completely useless.
Jarques Bertin is one of the major names in semiotics (the
study of signs). He has written a number of very influential
books on graphical presentation.
Comments on Pie Charts IV
The Energy Information Agency (EIA) is part of the
U.S. Department of Energy and is charged with compiling
and disseminating information about energy to the
government and private sectors.
EIA maintains a large standards manual for graphical
presentation.
http://www.eia.doe.gov/neic/graphs/preface.htm
Comments on Pie Charts IV
William Eddy of Carnegie-Mellon University, formerly
vice chair of the American Statistical Association
(ASA) Committee on Energy Statistics, said of pie
charts at the April 1988 ASA committee meetings in a
session on the EIA Standards Manual, death to pie
charts.
Howard Wainer of the Educational Testing Service
stated in a 1987 Independent Expert Review of EIA
Statistical Graphs Policies that the use of pie charts is
almost never justified and that they ought not to be
used. Wainer recommended to EIA that dot charts be
used instead of pie charts in EIA products.
Comments on Pie Charts V
During revision for the STAT 120 (Information
Visualisation) exam in 2002, Ross Ihaka said:
If you want to fail this course, just show me a pie chart.
Drawing Pie Charts with R
A basic pie chart is produced from a vector of named values.
such a vector can be created as follows:
> meat = c(8, 10, 16, 25, 41)
> names(meat) = c("Lamb",
"Mutton",
"Pigmeat",
"Poultry",
"Beef")
> pie(meat,
main = "New Zealand Meat Consumption",
cex.main = 2)
New Zealand Meat Consumption
Pigmeat
Mutton
Lamb
Poultry
Beef
Annotating Pie Charts
Because it is so hard to decode values from pie charts, it is
common to include the values as text in the plot.
> meat = c(8, 10, 16, 25, 41)
> names(meat) = c("Lamb",
"Mutton",
"Pigmeat",
"Poultry",
"Beef")
> pie(meat,
labels = paste(names(meat),
" (", meat, "%)",
sep = ""),
main = "New Zealand Meat Consumption",
cex.main = 2)
New Zealand Meat Consumption
Pigmeat (16%)
Mutton (10%)
Lamb (8%)
Poultry (25%)
Beef (41%)
A Tabular Representation
In this case, the information is easier to extract from a table
than from a pie chart.
New Zealand Meat Consumption
Lamb
Mutton
Pigmeat
Poultry
Beef
8%
10%
16%
25%
41%
(This table has deliberately been kept simple. No boxes or
lines have been used.)
New Zealand Meat Consumption
Lamb
Mutton Pigmeat
20
Poultry
40
Beef
60
80
100
A Simple Bar Chart
> barplot(meat, ylim = c(0, 50),
col = "lightblue",
main = "New Zealand Meat Consumption",
ylab = "Percent in Category",
cex.main = 2, las = 1)
New Zealand Meat Consumption
50
Percent in Category
40
30
20
10
0
Lamb
Mutton
Pigmeat
Poultry
Beef
A Horizontal Bar Chart
> barplot(meat, xlim = c(0, 50),
col = "lightblue",
main = "New Zealand Meat Consumption",
xlab = "Percent in Category",
cex.main = 2, las = 1,
horiz = TRUE)
New Zealand Meat Consumption
Beef
Poultry
Pigmeat
Mutton
Lamb
10
20
30
Percent in Category
40
50
New Zealand Meat Consumption
Beef
Poultry
Pigmeat
Mutton
Lamb
10
20
30
Percent in Category
40
50
United States Energy Production I
United States Energy Production II
Year: 1985
Year: 1986
coal
19.5
coal
19.3
gas
16.9
gas
16.5
other
5.38
nuclear
4.15
other
5.4
nuclear
4.47
oil
18.4
oil
19
Year: 1987
Year: 1988
coal
20.9
coal
20.1
gas
17.1
gas
17.2
other
4.8
other
5.04
nuclear
4.91
nuclear
5.68
oil
17.7
oil
17.3
United States Energy Production III
25
20
coal
oil
gas
coal
+8%
oil
gas
9%
+2%
15
10
nuclear
other
other
nuclear
0
1985
1986
1987
1988
1989
+37%
11%
Decorating Plots
One common complaint about R is that the plots it
produces are plain or boring.
In fact, if you are prepared to put a little effort in, you
can produce a wide variety of interesting effects.
Of course, there is no substitude for having a graph
which shows that something interesting is going on.
United States Energy Production IV
25
20
coal
oil
gas
coal
+8%
oil
gas
9%
+2%
15
10
nuclear
other
other
nuclear
0
1985
1986
1987
1988
1989
+37%
11%
Filling a Plots Background
Colouring the background of the plot region is simple. After
setting the up the axis scales, determine the coordinates of the
edges of the plotting region and draw a filled rectangle which
fills the area completely.
plot.new()
plot.window(xlim = xlimits, ylim = ylimits)
usr = par("usr")
rect(usr[1], usr[3], usr[2], usr[4],
col = "lemonchiffon")
Thick Lines
Thick lines can be drawn by first drawing the lines n units
wide in black and then drawing them n 3 units wide in the
fill colour. This works for all colours.
> lines(x, y, lwd = 8, col = "black")
> lines(x, y, lwd = 5, col = "green4")
Drop Shadows
A drop shadow effect can be obtained by first drawing the line
in gray, offset down and to the left, and then drawing the line
itself.
> lines(x + xinch(.1),
lwd = 8, col =
border = NA)
> lines(x, y, lwd = 8,
> lines(x, y, lwd = 5,
y - yinch(.1),
"lightgray",
col = "black")
col = "green4")
Specular Reflections
It is also possible to create a three dimensional look by adding
what appears to be a specular highlight.
> lines(x, y, lwd = 8, col = "black")
> lines(x, y, lwd = 5, col = "green4")
> lines(x, y, lwd = 1, col = "white")
Combined Effects
It is of course possible to include all three of these effects in
in a single graph.
> lines(x + xinch(.1), y - yinch(.1),
lwd = 8, col = "lightgray",
border = NA)
> lines(x, y, lwd = 8, col = "black")
> lines(x, y, lwd = 5, col = "green4")
> lines(x, y, lwd = 1, col = "white")
Spaggetti Anyone?
2
0.0
0.2
0.4
0.6
0.8
1.0
A Useful Set of Line Colours
Red 2
Dark Orange
Green 4
Dark Cyan
Medium Blue
Dark Violet
New Zealand Meat Consumption
Lamb
Mutton
Pigmeat
Poultry
Beef
10
20
30
Percentage of Meat Eaten
40
50
New Zealand Meat Consumption
Lamb
Mutton
Pigmeat
Poultry
Beef
10
20
30
Percentage of Meat Eaten
40
50
Mosaic Plots
Hartigan, J. A., and Kleiner, B. (1981), Mosaics for
contingency tables, In W. F. Eddy (Ed.), Computer
Science and Statistics: Proceedings of the 13th
Symposium on the Interface. New York:
Springer-Verlag.
Hartigan, J. A., and Kleiner, B. (1984) A mosaic of
television ratings. The American Statistician, 38,
3235.
Friendly, M. (1994) Mosaic displays for multi-way
contingency tables. Journal of the American Statistical
Association, 89, 190200.
Who Listens To Classical Music?
The following table of values shows a sample of 2300 music
listeners classified by age, education and whether they listen
to classical music.
Education
High
Low
Classical Music
Age
Yes
No
Yes
No
Old
210
190
170
730
Young
194
406
110
290
This is a 2 2 2 contingency table.
Old Versus Young
The effect of age and education on muscial taste can be
investigated by breaking the observations down into more
homogenous groups. The most obvious split is by age. There
are 1300 older people and 1000 younger people.
Old
Young
56.5%
43.5%
This is almost certainly a result of the way in which the
sample was taken.
Education Level
Within the old and young groups we can now find the
proportions falling into each of the high and low education
categories.
Old
Young
High Ed.
Low Ed.
High Ed.
Low Ed.
30.8%
69.2%
60.0%
40.0%
The young group is clearly more highly educated than the old
group.
Music Listening
Finally, we can compute the proportion of people who listen
to classical music in each of the age/education groups.
Old
Young
High Ed.
Low Ed.
High Ed.
Low Ed.
52.5%
18.9%
32.3%
27.5%
The music-listening habits of younger people seem to be
fairly independent of education level. This is not true for older
people.
Summary
The result of our analysis is a series of tables. From these
tables we can see:
There are slightly more old people than young people in
the sampled group.
The younger people are more highly educated than the
older ones.
The likelihood of listening to classical music depends
on both age and education level.
Mosaic Plots
Mosaic plots give a graphical representation of these
successive decompositions.
Counts are represented by rectangles.
At each stage of plot creation, the rectangles are split
parallel to one of the two axes.
Everyone
Old
Young
Low
Education
High
Old
Young
Age
Old
Young
No
Yes
Low
Education
High
Yes
Age
No
The Perceptual Basis for Mosaic Plots
It is tempting to dismiss mosaic plots because they
represent counts as rectangular areas, and so provide a
distorted encoding.
In fact, the important encoding is length.
At each stage the comparison of interest is of the
lengths of the sides of pieces of the most recently split
rectangle.
Creating Mosaic Plots
In order to produce a mosaic plot it is neccessary to
have:
A contingency table containing the data.
A preferred ordering of the variables, with the
response variable last.
Data Entry
> music = c(210, 194, 170, 110,
190, 406, 730, 290)
> dim(music) = c(2, 2, 2)
> dimnames(music) =
list(Age = c("Old", "Young"),
Education = c("High", "Low"),
Listen = c("Yes", "No"))
Data Inspection
> music
, , Listen = Yes
Education
Age
High Low
Old
210 170
Young 194 110
, , Listen = No
Education
Age
High Low
Old
190 730
Young 406 290
Producing A Mosaic Plot
The R function which produces mosaic plots is called
mosaicplot. The simplest way to produce a mosaic plot is:
> mosaicplot(~ Age + Education + Listen,
data = music)
It is also easy to colour the plot and to add a title.
> mosaicplot(~ Age + Education + Listen,
data = music,
col = "darkseagreen",
main = "Classical Music Listening")
Classical Music Listening
Old
Young
No
Yes
Low
Education
High
Yes
Age
No
Example: Survival on the Titanic
On Sunday, April 14th, 1912 at 11:40pm, the RMS Titanic struck an
iceberg in the North Atlantic. Within two hours the ship had sunk.
At best reckoning 705 survived the sinking, 1,523 did not.
The Data
There is very good documentation on who survived and
who did not survive the sinking of the Titanic.
R has a data set called Titanic which gives data on the
passengers on the Titanic, cross-classified by:
Class: 1st, 2nd, 3rd, Crew.
Sex: Male, Female.
Age: Child, Adult.
Survived: No, Yes.
Adults
Survivors
Male Female
Non-Survivors
Male Female
1st Class
2nd Class
3rd Class
Crew
57
14
75
192
118
154
387
670
Children
Survivors
Male Female
1st Class
2nd Class
3rd Class
Crew
5
11
13
0
140
80
76
20
1
13
14
0
4
13
89
3
Non-Survivors
Male Female
0
0
35
0
0
0
17
0
Producing a Mosaic Plot
The following command produces the mosaic.
> mosaicplot(~ Class + Sex + Age + Survived,
data = Titanic,
main = "Survival on the Titanic",
col = c("lightblue", "darkseagreen"),
off = c(5, 5, 5, 5))
Note the use of col= to produce alternating coloured
rectangles green for survivors and blue for non-survivors.
Also note that the off= argument is used to squeeze out a
little of the space between the blocks.
Survival on the Titanic
2nd
Child Adult
Child
3rd
Adult
Female
Yes
No
Sex
Male
Yes
No
1st
Child Adult
Class
Child
Crew
Adult
Example: Sexual Discrimination at Berkeley
In the 1980s, a court case brought against the University
of California at Berkeley by women seeking admission
to graduate programs there.
The women claimed that the proportion of women
admitted to Berkeley was much lower than that for men,
and that this was the result of discimination.
Gender
Male
Female
Admitted
1198
557
Rejected
1493
1278
%Admitted
44.5
30.4
It is clear that a higher proportion of males is being
admitted.
The University Case
The Dean of Letters and Science at Berkeley was a famous
statistician (called Peter Bickel) and he was able to argue that
the difference in admissions rates was not caused by sexual
discrimination in the Berkeley admissions policy, but was
caused by the fact that males and females generally sought
admission to different departments.
The Dean broke the admissions data down by department and
showed that within each program there was no admission
discrimination against women. Indeed, there seemed to be
some admissions bias in favour of women.
Admitted
Rejected
% Admitted
Department A
Male
Female
512
89
313
19
62
82
Department B
Male
Female
353
17
207
8
63
68
Department C
Male
Female
120
202
205
391
37
34
Department D
Male
Female
138
131
279
244
33
35
Department E
Male
Female
53
94
138
299
28
24
Department F
Male
Female
22
24
351
317
6
7
Producing The Berkeley Mosaic
We relabel the Admit/Reject levels so that the labels will fit
across the plot.
> x = UCBAdmissions
> dimnames(x)[[1]] = c("Ad", "Rej")
> mosaicplot(~ Dept + Gender + Admit,
data = x,
col = c("darkseagreen", "pink"),
main = "Student Admissions at UC Berkeley")
Student Admissions at UC Berkeley
Rej
Ad
Rej
Ad
Rej
Male
Female
Gender
Ad
Dept
Ad
Rej
Ad
E
Rej
Ad
F
Rej
