1/6/2020
Truncating the Axis and a Idea Redesign
Final week Mona Chalabi posted a chart First Time Home Buyers Average Age. Beneath is Mona’s chart.
This chart unfold shortly by means of the information visualization neighborhood. Inside quarter-hour of her posting this graph, somebody messaged me, « Mona truncates the y-axis » and different feedback began to shortly pop up on Twitter. Pekka Taipale wrote « I’m most impressed by the 30 underground stories that all these houses have. #scale #barchart ».
Chris Ganowski wrote « Nice concept, but that y axis needs to start at zero – otherwise you’re exaggerating the differences. ». He later writes about consultants agreeing that the axis on bar charts needs to be set at zero, to which Edward Tufte responds.
Within the Huge Ebook of Dashboards, we addressed this kind of situation quite a few instances. One instance is that this callout field on web page 44 in Chapter 2 – the Course Metrics Dashboard. Our recommendation is that when utilizing size, peak, or space to encode the information for comparisons, all the time begin the axis at zero.
So what’s the situation right here? The first situation is that the information is being encoded with peak, on this case the peak of the home is representing the typical age of the firt time dwelling purchaser. The y-axis begins at 30 years previous. In consequence, the peak of the home in 2008 seems to be double the dimensions of the home in 2007, and 2017 is greater than 3 times as tall. It does not take somebody two or 3 times longer to purchase a home, so there’s a disconnect between the precise information and the information that’s being plotted. The graph is definitely plotting « How a lot older than 30 is the typical age of a primary time dwelling purchaser. »
Quite a few a individuals identified that nobody will purchase a home on the age of zero and in lots of locations there could also be a authorized age required to enter right into a contract to personal actual property (for instance, 18 years previous). Whereas that is true, this doesn’t change how the information is being encoded. The peak of the homes are nonetheless encoding « the variety of years over 30 ».
Truncating an Axis
Ought to we ever truncate the axis? Sure, there are various situations the place truncating an axis can be a good suggestion. If the distinction within the information is small and that distinction is vital, and you are not encoding the information utilizing peak, size or space, then truncating the axis may very well be very helpful. For instance, the distinction on the earth file instances for the 100 meter backstroke brief course during the last 10 years is about one second. In November 2009, Nick Thoman (from Cincinnati, OH) set the world file 100 meter backstroke brief course at 48.94 seconds. Up to now 10 years, it has been damaged 3 times, every by 2/one centesimal of a second.
If we present this as a bar chart or a line chart with the y-axis beginning at zero then there isn’t a seen distinction.
Aternatively, we will change it to a stepped line and truncate the y-axis to one thing that references the information in a significant approach.
Let’s return to Mona’s chart. Might we plot that as a line chart and truncate the y-axis? Effectively, in truth, that’s precisely what the BBC did once they printed this information. Right here is the original chart from the BBC published on January 31, 2019.
Discover it is a line chart beginning at 30, utilizing the identical .5 intervals, and over the identical time frame. This chart works very properly. In addition they included the abstract in a sentence proper above the chart, « The typical age of first-time patrons has risen from 31 to 33 over the identical 10 years. » This makes the purpose, easy and clear.
Idea Redesign
One choice for a redesign can be to vary the chart kind. Utilizing a line chart, dots or different encoding that doesn’t encode with peak or size can be one choice. This could be a serious redesign and would utterly change the design and elegance of Mona’s visualization. This could most likely be an excellent time so as to add that I’m an enormous fan of Mona Chalabi’s work. Be sure you follow her on Twitter in case you are not already. Her designs are spectacular and provoking, so the very last thing I wish to do is take away from that. Certainly one of my previous professors used to say, « Do not throw the child out with the bathtub water. » In an effort to maintain as a lot of the design as potential, another choice could be to vary the story barely to match how the information is encoded.
Beneath is an idea redesign. I modified 4 issues.
1. For the reason that information is encoding « how a lot older than 30 is the typical age », I modified the title of the visualization to match the encoding, « How lengthy after 30 ? »
2. I added an annotation to the visualization to make the message clear « The typical age of a primary time dwelling purchaser elevated 2 years from 2007-2017, ~ 31 to 33 years previous. » This can assist the reader perceive the magnitude of the distinction shortly.
3. I prolonged the y-axis right down to 29.5 to assist the reader see that the y-axis begins beneath 30, however the homes are plotted beginning at 30.
4. I added a line to the chart in an try to convey out the development with out encoding peak.
Would a line chart be higher? In all probability, however then we lose the entire design components on this visualization. Nonetheless, by making a couple of minor modifications, we will preserve the design components and assist the reader perceive what’s going on within the information. The encoding matches the story, there are clues to assist the reader see that the y-axis does not begin at zero and the primary message is introduced out explicitly in order that there isn’t a confusion within the magnitude of the change over time.
Here’s a variation of the redesign, muting the colours of the home a bit and bringing out the road. General, my aim was to maintain as a lot of Mona’s design components as potential, so I did not wish to mute the colours far more than this.
Notice – This can be a idea redesign. I did this in Adobe in a couple of minutes time simply to point out the ideas that I’ve mentioned on this weblog publish. I’m positive others might do a way more completed and polished redesign.
I hope you discover this data helpful. In case you have any questions be happy to electronic mail me at Jeff@DataPlusScience.com
Jeffrey A. Shaffer
Observe on Twitter @HighVizAbility
