Scribd
Upload
99 views5 pages

Good and Bad Graphs

This document discusses good and bad practices for creating graphs. Good graphs clearly show important data features through a title, labeled axes, and key. They tell a story while minimizing non-essential visual elements. Bad graphs can mislead through scaling issues, uneven emphasis, or distorting areas. Pie charts in particular are often misleading and bar charts are usually better. Histograms should have equal bar widths. Real examples from newspapers demonstrate improperly sized pictograms and muddled datasets.

Uploaded by

kkk
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
99 views5 pages

Good and Bad Graphs

This document discusses good and bad practices for creating graphs. Good graphs clearly show important data features through a title, labeled axes, and key. They tell a story while minimizing non-essential visual elements. Bad graphs can mislead through scaling issues, uneven emphasis, or distorting areas. Pie charts in particular are often misleading and bar charts are usually better. Histograms should have equal bar widths. Real examples from newspapers demonstrate improperly sized pictograms and muddled datasets.

Uploaded by

kkk
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Good and bad graphs.

Good graphs clearly show the important features of the data. They should always
have:
 a title
 labelled axes
 a key.

In general they should tell a story and be memorable but also have a ‘low
information to ink ratio’ (junk kept to a minimum and no distracting features) and
not mislead the viewer. Some of the following examples of bad graphs also give a
corrected good graph. Choice of colour when designing charts and graphs is also
important to allow for colour blindness and black and white printing.

Graphs are often made misleading for advertising or other purposes, or even just
by accident, by:
• Leaving gaps/changing the scale in vertical axes
• Uneven shading/colours
• Unfair emphasis on some sections
• Distorting areas in histograms (bar widths should always be equal - if you
have different widths then the bar height must be adjusted so AREAS
reflect counts)
• Use of 3-dimensions instead of two
• Misleading use of pictograms
In particular, watch out for missing zero points on axes, spurious colouring and
annotation, and unjustifiable extrapolation. Pictograms are often misleading as in
the case of the following graph as areas or volumes (instead of heights) are used
to represent numbers exaggerating differences visually.

Example (from the Los Angeles Times, August 5 1979, p3)


The data show a reducion in the ratio from 27% to 12%. this is represented by a
change in the height of the doctor but your eye sees a change in the area.

1
Examples of bad graphs that could have be drawn in Excel

Pie charts

It is debatable whether pie charts ever need to be used as bar charts are almost
always a better representation of proportions in a data set. Unless properly
constructed pie charts can be very misleading.

For example, the two dimensional pie chart below has been constructed so the
party with the highest vote is in the front (and therefore inflated) but the second
highest is at the back (and looks less than it should). Also no proportions are given
on the graph.

Percentage of Vote

National Greens Labour Maori NZ First ACT Other

Pie charts like the one below (of the same data) that ‘explode’ by having the
sectors move apart from each other further exaggerate these misleading features
and should never be used.

Percentage of Vote Percentage of Vote

National Greens Labour


Maori NZ First ACT National Greens Labour Maori
Other NZ First ACT Other

A correct pie chart for this data is:

2
1%
1% Percentage of Vote
3% 2%
6%

53%
34%

National Labour Greens Maori


NZ First ACT Other

Bad pie charts created using UK CENSUSATSCHOOL data (problems with


each graph are listed below it)

 Sectors should be touching each other  Sectors should be touching each other
 No percentage labels  No key to say what each sector represents
 No indication of how many children  No indication as to how may children
participated participated

 Pie charts should always be 2-dimensional, in


3-dimensions the visual representation of the  All the sectors should be presented in the
sectors is distorted same pie
 A key should be used for the attributes
 A key should be used for the attributes
 No title

 Sectors don’t touch in the middle, (so angles


can’t be compared easily)

3
Bar charts

Both the (made up) graphs on the left mislead the reader by exaggerating the
differences in the heights of the bars by not starting the vertical axis at zero. The
correct graphs are given on the right.

Histograms

The graph on the left misleads the reader by doubling the width of some of the
bars. The correct graph on the right halves the heights of these bars so that the
area still represents the frequency.

4
Real life examples of bad graphs

The following graph from a survey of inner city apartment dwellers in Wellington,
New Zealand was published in the Dominion Post newspaper on Monday 13 April,
2009. Not only are the pictograms representing walking to work, going by bus or
driving incorrectly sized (with the 6% going by bus being larger than the 13%
driving) but also has two different sets of data muddled into the same graph (the
data displayed on the left hand size of the graph is not just about the proportion of
those cycling to work, 6%, but also about what cycle owners might do in certain
situations). The larger picture of a bike is there merely to display the proportion
that own a bike. It is not good practice to use a pie chart with just two values.

Many more examples of real life bad graphs can be seen on websites such as
Junk Charts http://junkcharts.typepad.com, The Top Ten Worst Graphs
www.biostat.wisc.edu/~kbroman/topten_worstgraphs and Eval Blog
http://evalblog.com/2012/01/23/tragic-graphic-the-wall-street-journal-lies-with-
statistics.

You might also like