Great graphs
A picture is worth a thousand words
A great graph can illustrate a point more succinctly than words. It can demonstrate the author’s intimate knowledge of their subject and persuade a reader by helping them understand the concepts it illustrates.
A poor graph, on the other hand, can undermine the author’s communication of their research by eliciting doubt and distrust in the reader. At worst, it can misrepresent the data or mislead the reader.
Authors often design the graphs they use in their publications. Knowing what key elements every graph should have and understanding the basics of data visualisation can improve presentation, usefulness and reader interpretation of graphs, leading to clearer and more professional communication of the author’s research.
Key elements of great graphs
These key elements allow the reader to read and interpret the data in a graph:
a title or caption that clearly identifies what the graph is about
data represented as bars, lines, dots or other shapes in the data region
axes that define the data region
labels and (if the data requires it) units on both axes
a scale to supplement the information on the axes labels
a key or legend if the elements of the graph are not labelled in the data region
alt text if the graph is published online
a source that explains where the data or information came from
notes for further information, if required.
Once a graph contains all the key elements, the next step is to ensure the graph can be interpreted.
Data visualisation
A picture is worth a thousand words. The hard part of drawing graphs is ensuring those words are the right ones. A great graph should:
simplify comparisons of data or information
use concepts that take into account limits of human perception and cognitive load
reduce the potential for misinterpretation
avoid misleading the reader
demonstrate the message or key point intended by the author.
A great graph should make data easier to interpret, not harder. There is no point representing data visually if it complicates the information. Look at the following example:
The two graphs above show the population of Australia’s states and territories. The first graph orders the data alphabetically. The second graph orders the data from smallest to largest. The author’s intended purpose is to show the difference in population between different states. This is easier to interpret when the data are ordered by value.
When creating a graph, it is important to understand the concepts of human perception and cognitive load. Human perception is the ability of the mind to correctly interpret stimuli and to accurately interpret it. Cognitive load is the amount of information, especially visually and aurally, we are able to process at one time before the information overloads our senses.
The first graph above requires effective interpretation of the relative size of the portions of the ring as well as correct visualisation of each colour and its relation to a country in the legend. The reader also has to correlate the numbers with each ‘wedge’ of the graph. It requires the reader to perceive colour and relative size and match colours with numbers. (Remember that colour blindness can make the interpretation of colours difficult for many readers.)
The second graph is much less cluttered, creating less of a cognitive load. It orders the data values from highest to lowest, and there are fewer colours and numbers to interpret. It is easy to see which country has had the most COVID-19 cases. Yet both graphs represent exactly the same data.
A badly designed graph can lead to misinterpretation of the data by the reader.
In the first graph above, the y-axis starts at zero. In the second graph, the y-axis starts at 2,000,000. This changes the relative heights of the bars. The reader’s interpretation of how many voters are satisfied with the Prime Minister’s handling of COVID-19 could vary, from one-third more people saying ‘Yes’ than ‘No’ to twice as many.
This brings us to the final purpose of a great graph. It should successfully demonstrate the author’s message or key point. The author should carefully consider their main point and design the graph to convey that message as simply and succinctly as possible.
Both graphs above represent book sales on a monthly basis. The first graph orders the data values from lowest to highest, just like the population graph discussed earlier. However, the author’s intended purpose is to show the effect of national holidays like Mother’s Day (May), Father’s Day (September) and Christmas (December) on book sales. This concept is easier to see in the second graph, because we are more used to visualising monthly data in chronological order. This helps us visualise the data and decreases the cognitive load associated with interpreting the information in the graph.
If you make sure that your graph includes the necessary key elements and think about how readers will visualise and interpret it, you are several steps closer to ensuring it has a clear, concise and straightforward message that will be understood correctly.
Resources
The online Australian Government Style Manual (2020) guidelines for images.
Australian Government (revised by Snooks & Co) (2002) Style manual for authors, editors and printers, 6th edn, John Wiley & Sons Australia, Brisbane.
Kirkham H & Dumas RC (2009) The right graph: a manual for technical and scientific authors, John Wiley and Sons, Inc., Hoboken, NJ.
Tufte E (2001) The visual display of quantitative information, Graphics Press LLC, Cheshire, Conn.
United Nations (2009) Making data meaningful, Part 2: a guide to presenting statistics, United Nations Economic Commission for Europe, Geneva.
