Note: Images may be inserted into your messages by uploading a file attachment (see "Manage Attachments"). Even though it doesn't appear when previewed, the image will appear at the end of your message once it is posted.
Register Latest Topics

  Author   Comment  

Posts: 2
Reply with quote  #1 
Hi Stephen,
I'm a budding data viz nerd, and I've been gobbling up your work as quickly as I can so I can make better judgement calls in my role as a UI/UX designer. I fell upon your article saying that dual axis graphs should probably not be used (https://www.perceptualedge.com/articles/visual_business_intelligence/dual-scaled_axes.pdf) in most cases. I would not be posting here if you had not concluded that article with the sentence "I invite you to propose viable exceptions, which I will welcome with open arms."

My situation is one that doesn't seem to be covered in your article linked above, and if there's a better solution to my problem, I can't seem to find it.

In my work, we look at telecommunications network traffic, so one of the big questions that our users need answered is "Why is my VoIP call quality performing poorly?"

One of those reasons might just be a capacity issue: too many simultaneous calls means call quality drops for some people. In this case, we'd want to show a graph that correlates call volume with percentage of poor calls.

This is the quick mockup I came up with to illustrate this example. Probably more thought needs to go into color choices and other aesthetic aspects. 
This indicates a correlation between call volume to poor call quality, and tells the technician that more network capacity is required to eliminate the poor quality in some of these calls.

Is there a better way to indicate this correlation?


Posts: 192
Reply with quote  #2 

Till Stephen will answer you, I would suggest to search this forum for "dual axis" topics / posts where the "correlation" idea using a dual axis design was extensively discussed. Your dual axis design is not allright and can be replaced by a scatter graph or side by side charts.

Posts: 853
Reply with quote  #3 

The essential problem with dual-scaled graphs is that they are confusing to many people. They encourage people to mistakenly read meaning into and to make comparisons between values with different scales. For example, if the line in your example intersected a bar at some point, that intersection would encourage people to read meaning into that event even though the intersection is arbitary. On the other hand, if your audience knows how to read a dual-scaled graph properly, your design works fine, assuming that people must see the values as time series.

As Danz mentioned, if your sole purpose is to show the correlation between trunk capacity and calls with poor quality, a scatterplot would show this more clearly, but it would not display the patterns of change through time.

If you have any concern at all that your audience might find a dual-scaled graph confusing, there is an easy solution: create two line graphs, one directly above the other. This would make the relationship between trunk capacity and calls with poor quality easy to see without creating the potential for confusion.

Stephen Few

Posts: 2
Reply with quote  #4 

Posts: 200
Reply with quote  #5 
Originally Posted by sfew
If you have any concern at all that your audience might find a dual-scaled graph confusing...

In my experience, this is almost always a safe bet.

In addition to the concern that it could be confusing to some users, there is a very good chance on most of these charts that the lines will cross or otherwise interfere with each other in a way that adds unnecessary clutter to the chart.

I find using two charts to be a much better solution in nearly any case I've ever encountered, using two different measures in two different units.

Previous Topic | Next Topic

Quick Navigation:

Easily create a Forum Website with Website Toolbox.