This is the second blog in a series which focuses on the occasional difficulties around user acceptance of bullet charts. In my experience, bullet charts are a bit ‘marmite’ – some people ‘get’ them completely, but other people struggle and do not ‘get’ them. In my own opinion, I am a bullet chart fan, since I think that they are a good way of expressing data concisely in a way that’s compatible with our current understanding of human perception.
In order to try and help people make the mental leap towards understanding and using bullet charts, each blog will focus on a particular reason that I’ve heard from users who are not comfortable with bullet charts. I hope to balance each blog by attempting to provide a resolution to try and help users to understand bullet charts a little bit better.
In this blog, I would like to address one particular reason that people have said which means that they don’t ‘get’ bullet charts – they don’t always grasp the meaning of the linear scale used as a background to the bullet graph. In the Bullet Graph Design Spec, the linear scale is described as ‘From two to five ranges along the quantitative scale to declare the featured measure’s qualitative state (optional)’ (Few, 2010). What does this actually mean? Here’s an analogy, which might help to clarify what this means.
Let’s take the case where you would like to check the quality of your password strength. In other words, you would like to check how weak, medium or strong is your password. Microsoft provide a free online password checker, which provides a visual guide of the strength of your password. Here is an example image which denotes the quality of a weak password:
Here is the feedback image provided when the quality of the password is considered to be fair:
Finally, here is the visual feedback provided if the password is judged to be strong:
The farther right along the coloured bar travels, then the password is considered to be higher quality as it proceeds from the left to the right. This is familiar, since this is also used for progress bars during software installs, for example.
The Microsoft free online password checker also double-encodes the quality of the password by using traffic light colouring i.e. red denotes danger, amber denotes fair, and green is ‘go’ or acceptable. As an example, if we wanted to ‘label’ the aforementioned password strength example with percentages, the values 0%-40% may denote a ‘red’ or weak password; 41 – 69% may indicate an ‘amber’ or ‘fair’ password, and values over 70% – 80% may specify a ‘strong’ password; values 81% or over may indicate the ‘best strength’ password.
What has this got to do with bullet charts, I hear you ask? Well, the linear scale at the background of the bullet chart also helps to denote quality in two ways:
– The ‘left to right’ mechanism, familiar from progress bars or the above example, is also deployed in bullet charts. The qualitative state of the measured metric is considered to increase as the measure travels from the left to the right.
– The colour of the bullet chart helps to denote quality by helping to ‘label’ the quantity. As in the aforementioned example where values can be represented by a ‘traffic light’ colour, the bullet chart can also use colour to indicate value, thereby enhancing the ‘qualitative’ understanding of the chart. Here is an example bullet chart from Reporting Services 2008 R2, which can help to illustrate this point:
We can see the values in the darker grey colour are used to denote metrics that fall into the range 0 – 60%, the mid-grey is used to identify 61% – 80%, and the lightest grey is used to denote 81% and over. Stephen Few makes the point that colour-blind people may find the ‘traffic light’ colours confusing. To get around this issue, which affects approximately 10% of males, Few suggests that ranges of intensities of the same hue are easier for people to visually digest, with the darker colour representing ‘poor’ quality, progressing through to lighter intensity to represent ‘good’ quality.
In the last example, I’ve used darker blue to denote 0% – 60%, and lighter blue from 61 – 80%, and finally the lightest blue from 81% until 100%. The colour helps the viewer to identify the quality of the result e.g. poor, medium or high.
The comparative measure (the little vertical blue line) indicates the overall ‘success’ and could be called the ‘target’. In a real life example, this may indicate whether a particular department met a recycling target, or made a certain percentage of total projected sales, for example.
We can see that the top bullet chart slightly exceeded the comparative measure, or target, which can be found at 75% or thereabouts. We can see the performance measure of the top bullet chart (the thick blue horizontal line) ends at 80%, and ends at 100% in the bottom bullet chart. The top bullet chart has a performance measure that does not travel as far as the bottom bullet chart, because it is a lower percentage; this ‘left to right’ aspect makes it easier to read, and the linear colours help to attach a ‘quality’ label to the data i.e. poor, medium, good.
Even if only an ‘at a glance’ measure is needed, the colour helps users to add a qualitative label to the values. Since the bullet charts are located one on top of the other, this leads to easy comparison; in the above example, the top chart didn’t exceed the target as much as the bottom chart, and the bottom chart succeeded in obtaining 100% as the performance measure.
To summarise, I hope that this helps to make the value of the linear scale more clear to users; it isn’t obvious to everyone, and not everyone picks it up intuitively. However, when it is explained, I think that there is value in emphasising the comparative aspects of the bullet graph. The colour and ‘left to right’ aspects of the bullet graph can assist viewers to make qualitative understandings of the data that they see before them, and the neatness of the bullet chart location can help to facilitate comparisons between bullet charts.
I hope I’ve got you wondering how you can create bullet charts? More on this to follow!