Top Five Mistakes People Make When Creating Graphs
Below are the top five mistakes people most often make when creating a graph. The good news is that avoiding them is easy if you apply just a few of data visualization principles.
1. Choosing the wrong type of graph
The whole purpose of using a visualization tool – for example graphs – is to in an efficient way communicate the whole of a message to an end-user. This means you should be aiming for simplicity, and your end-user should get the whole message, and nothing but the whole message. Not another message, or parts of the message.
In the example below, you want to show your company’s market share and compare it with that of your competition. There are seven companies on the market, and they are similar to each other in regards to size.
If you select a pie graph, even with data labels in the slices, it does take some work to read the data. You need to move your eyes between the graph and the legend to interpret which color represent which company, and which one is our company.
In addition to that, the graph may be directly misleading, as the purple slice looks bigger as the orange or light blue (17.00%) slice. Your end-user not only has to work to get the message, he/she might be getting the wrong message.
What if you use a bar graph, as below, instead?
I opted not to use the data labels in this graph, but the message is still loud and clear: we are third. That´s the main point number one. In addition to that, there is a ranking among all the companies, and the differences in size is much more visible. Our company is highlighted, so it is easy to see our ranking and relative size. Even if we make the graph smaller it is still easy to understand it.
We only have one set of variables and that’s the market share. When most people think of data visualization, they tend to think that it is only done by using graphs. When you have only one set of variables, and the number of the individual values is few, then you are many times better off using a table instead, especially if you are more interested in looking up the exact values.
As simple as that…
2. Question and answer do not match
This one is more common than you might think.
In your role as communicator, it´s important to know the question, and then answer that specific question, and not to anotehr question. Your end-user needs to know what you are measuring. There must be an intuitive understanding between you and the end-user on what the specific question is, and that the answer is the answer to that specific question with the measure that you have agreed upon.
A very typical example in some situations is when the question is asked for a parts-to-whole in percentage but the answer given is in the measures of absolute numbers. Another example is in the case of deviations. For instance, you want to show the operational expenses by month compared to the corresponding month last year.
Have a look at the line graph below. While the graph, at the very first glance, looks clean and clear and gives the actual differences between the operational expenses current year against last year, there is a mismatch between the graph title and the graph itself: the question being asked is the variance between the expenses this year compared with last year.
Obviously, because we are talking about variance here, the measure used should not be the absolute amounts in dollars, but rather an index of the performance of the expenses vs last year, in percentage.
What we have done below is that we have used the amounts in the above graph, and set an index of 0% for last year, the year we are comparing the current year with. Not only does this graph match the graph title, it makes it also easier to see the differences between individual months. A difference is $28 of $832 in January is more difficult to see in a graph of this size than a variance of 3.4% in a graph of the same size, hence interpreting the message more clearly.
3. Forgetting the important stuff
You see a red Ferrari in a parking lot, surrounded by other cars, and you notice the red Ferrari. It stands out.
When you want to draw attention to your most important message or conclusion, you must make sure it stands out, like the red Ferrari. If you line up your graphs the way it is depicted in the example below, most end-users will not automatically know which message is being the most important. All of them have been given the same importance as they are all of the same size and use a range of colors.
Rank your graphs by the order of importance. Usually, in a report, if the objects are of the same size, we look at the top left corner and work our way down, or we start from the center. Highlight what is important by font size or a different color saturation or hue.
4. Not showing enough data
Showing too much data is one thing, what is even worse is showing not enough data.
Most people are not mind-readers. The graph above tells us nothing about if this was a survey that was carried out, and if so, where and when was it carried out, and what is the source. The x-axis does not reveal if the numbers are absolute, or percentage. If the numbers are percentage of a population, what about the rest as the numbers for each category do not add up to 100%. And if the number of everyday radio listeners is 20, it is good or bad, and how good or how bad.
Include a title to your graph. Specify the measure and the units of measurement. Make sure you have not left out any values. If applicable, name your source.
5. Go easy on color
Most people tend to overuse colors in graphs. Why? Partly, because they think it makes the graphs more appealing, and partly because when you create a graph in Excel, the software by default assigns a different color for each set of values.
Using colors sparingly is essential if you want to highlight something specific in a graph so that not all sets of values are given the same importance, or when you have a set of graphs and you don´t want to tire the eyes of your end-user. Beauty is in simplicity and when you are creating graphs as a mean of communication, your goal is to create graphs that are functional.
In the example below, you want to look at your sales figures, but your priority is to pay more attention to the gloves sales. The graph below does a good job of emphasizing on each category equally, but that´s just about that.
If you want to draw attention to or emphasize something in a graph, or show a relationship, colors are a very powerful tool. Opt to work with a few colors as you can, and avoid bright colors.
In the example below, it is easy to see the seasonality of the gloves sales. It is also easy to see that while gloves sales were the least in the summer months, they ended the year as number in December months sales. You could still get the same conclusion in the above graph with more colors, but the below example illustrates these facts more easily.
And simplicity is what you want for your end users.