When design goes awry
Feb 06, 2018
One can't accuse the following chart of lacking design. Strong is the evidence of departing from convention but the design decisions appear wayward. (The original link on Money here)
The donut chart (right) has nine sections. Eight of the sections (excepting A) have clearly all been bent out of shape. It turns out that section A does not have the right size either. The middle gray circle is not really in the middle, as seen below.
The bar charts (left) suffer from two ills. Firstly, the full width of the chart is at the 50 percent mark, so readers are forced to read the data labels to understand the data. Secondly, only the top two categories are shown, thus the size of the whole is lost. A stacked bar chart would serve better here.
Here is a bardot chart; the "dot" part of it makes it easier to see a Top 2 box analysis.
I explain the bardot chart here.
PS. Here is Jamie's version (from the comment below):
I like the bardot, but do you think it might be even better if you reversed the order of stacking, so the somewhat/very satisfied categories are stacked to the left? That way, you're reading the share that gave one of those two answers alongside the axis starting at 0, making it easier to know what to focus on. Right now it's almost like you're subtracting those responses from 100%.
Posted by: Camille | Feb 06, 2018 at 11:13 AM
I won't even comment on the donut chart.
The bar chart is not something that I would represent the way that you have.
I don't see the axis cutoff at 50% as a problem - I think in most cases, extending the axis to 100% is unnecessary, and in some cases counter productive.
The main problem that I see is that the two series are presented as if they should be compared to each other. I can't imagine it being useful to compare 'Very Satisfied' and 'Somewhat Satisfied' in this scenario - it would clearly be more useful to understand the total level of positive responses. So let's just plot those, as a standard bar chart with a single series (Percent Satisfied).
I remain unconvinced of the usefulness of the "bardot". In the two examples you've posted so far, I find the dot to be a potentially confusing complication. So far (to me at least) it seems that use of color on the bar segments is a better way to highlight the section of the bar that should be the focus.
Posted by: Jamie | Feb 06, 2018 at 01:21 PM
Kaiser - I came back to see if there were any replies to the comment I left yesterday, but it's gone :(
I had some comments and a link to an example - is that recoverable on this platform?
Posted by: Jamie | Feb 07, 2018 at 01:27 PM
Jamie - Thanks for the second comment. See my note on missing comments. It's an ongoing problem with Typepad's spam filter. There is a fair number of false positives, and I haven't been monitoring the spam folder.
The problem with not going to 100% is particular to this chart. I'd not recommend that if they had all categories plotted, that is, if the bar lengths actually added up to 100%. When they omitted some bars, then the reader loses the reference point.
Bardot: it's designed for people who care about Top 2, Bottom 2 type analyses. My philosophy is always to make it the easiest for readers to do the task they care about.
Camille: It was late and I was too lazy to try reversing the stacking. But I think you may be right.
Posted by: Kaiser | Feb 08, 2018 at 01:50 PM
Thanks for the update Kaiser.
Regarding: "always to make it the easiest for readers to do the task they care about". Couldn't possibly agree more! I strive to achieve this in my work, with varying degrees of success, of course.
But I disagree with the idea that the 'bardot' accomplishes that. Or, perhaps more precisely, I don't think that the bardot is the best to way to accomplish it, and I think that adding a dot is an unnecessary complication to an otherwise straightforward chart.
I've created a series of examples that I've put in one image, here: https://imgur.com/g9G1wlo
There are four charts.
On the top left, what I would consider a fairly typical display of survey results (the best way to display 'neutral' responses can wait for a different conversation).
Top right, the same chart with a dot added to mark the 'top 2'. The dot certainly draws the eye, and in that sense could be said to be successful in highlighting the top 2 mark as a data point.
I think that it does so in a way that makes it more difficult to decode the overall information in the chart - the bar segments are interfered with by a dot that the user needs to asses the purpose and meaning of, and the dot is obfuscated by virtue of the colored bar segments surrounding it.
On the bottom left and right are the two methods that I would use to highlight the results that need to be focused on. Which method to use depends on how much importance needs to be placed on the individual sections, vs. just highlighting the positive results in contrast to all others. I think this decisions is dependent on the particular circumstances when making the chart.
To me, if the goal is to focus on the top 2 or bottom 2, let's make that entirely the focus. The bardot seems like a 'half way' measure to me, which ends up not quite accomplishing the desired goal, while also making the chart more difficult to digest as a whole.
tl;dr - I prefer a subtractive method as opposed to an additive method when possible.
Posted by: Jamie | Feb 09, 2018 at 08:41 AM