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: 192
Reply with quote  #1 
I saw this treemap for the first time a few days ago when I read Bryan post.
They were too many reasons to redesign that graph, I will not repeat them all here.

Bryan redesign does a great job. He uses horizontal bar charts to encode years of lost life, % of change and, as a bonus, a third measure, all of them providing a much better interpretation than existing treemap. A treemap visualization usually serves for: grouping, part of a whole ratio - within group and global - and uses to encode an extra measure using colors. Comparison between lengths of bars sharing the same baseline will always be more accurate than comparison between areas, situation becoming even worse when the ratio width/height of the comparing rectangles is very different. More than that, the colors intensities encode very poor the % of change information. Bryan again uses bars to encode accurate that information. Eventually he is using colors for grouping purposes only, dropping any intensity variation making the whole design pleasant and effective.

Ignoring the unnecessary 3D effect of the original design, I could not see how Bill Gates could see at a glance "that while the number of people dying from communicable diseases is still far too high, those numbers continue to come down". Especially the second part of his statement. So I decided to come up with my own version using Bryan colors, labels and values using a different approach. My design is trying to be a reinterpretation of the original treemap goals, starting from the original remarks and Bill Gates comment. Here is my version:


Bellow I use YLL as abbreviation for years of lost life.
Shapes height encodes the part of a whole absolute percent of YLLs. They are ordered in Y direction in the descending order of YLLs for groups and elements within the groups.
Shapes Y position encodes the cumulative percent of YLLs
Shapes X position (center) encodes the other valuable measure (% of change of YLLs).
My solution improves the original solution in the following aspects:
1. The design tells the story.
2. Improved visual approximation of part of a whole for groups and elements (precise percentage is specified).
3. Improved visibility for % of change YLLs. See (almost) in block progress of the gray group (see Bill Gates comment).
4. Because the shapes do not overlap, useful information can be added for "relevant" elements (if height of available space is enough to fit the text)
5. Shapes are rendered top to bottom within group in the descending order of YLLs, allowing an easier comparison. Using bar heights only for comparison is easier then comparing different width/height ratio rectangle areas.
6. Colors are used just for logic grouping, dropping distracting intensities. 
7. Design is very compact, being suitable for any web site or magazine article. 
8. Solution is also suitable for animation starting from an stacked block horizontally centered in the graph.
1. Possible confusion in decoding shapes width. They do not relate to the % of change, but neither a bubble graph does relate the diameter to any of the axis scales.
2. It looses accurate comparison between elements. Still, it looks for me easier to compare (even not aligned) bar heights than the areas of the elements in a treemap.
My interpretation is that shape sizes are important more as order of magnitude than for precise comparison. Ordering them and mentioning (only the large) values next to the shapes do the right job IMO.

How does this work for you?



Posts: 200
Reply with quote  #2 
Personally I find the width of the bar far too confusing. 
Even after reading that it was only the center point that encoded anything, I subconsciously kept trying to work out what the start and end of the bar meant.

With a bubble it is more intrinsically apparent that the center is the point that encodes a value, rather than the edges, while the size of the bubble encodes something different.

I like that the design is compact and has fewer points, but I find Bryan's much easier to read and explore.

Posts: 192
Reply with quote  #3 
Bubbles or squares would not have the same cumulative sense while they would encode a measure by area. It is also a matter of vizibility, easier to see small parts. Might be that a white dot in the center of the shape would help a bit, but I don't find that difficult to ignore a constant width for all the bars. Eventually, the design can be interpreted like this: take a bar (as a whole), break it in pieces and move them to the left or to the right depending on the direction of change. I found this easy to imagine.

Posts: 200
Reply with quote  #4 
"Bubbles or squares would not have the same cumulative sense while they would encode a measure by area."

As it stands, I really don't get a cumulative sense from this any way.

And I guess all I can do is reiterate that I have a hard time making sense of the bars, and I have to think a lot of other people will.
Usually a bar of this nature encodes a range, and people will intuitively read it as a range, no matter how many times you explain that it isn't.

It's much like taking a standard bar chart and telling people to ignore the length of the bars, IMO.


Posts: 192
Reply with quote  #5 
Actually, is more like taking a standard bar and tell people to ignore the thickness. The length of the vertical bars has a meaning, the thickness, not.

But I do understand your point and I respect it.

Posts: 853
Reply with quote  #6 

Essentially, you've taken a stacked bar with segments that represent year's of lost life using height and repositioned the segments horizontally to encode the percentage change in lost life. Unfortunately, both encodings suffer from perceptual problems. It is difficult to compare the heights of the bar segments when they don't share a common baseline and, as jbriggs suggested, it is difficult to focus on the horizontal center of those segments when they are so wide. Whenever I redesign a visual display of quantitative data because it isn't encoded in ways that work well for human perception, I always begin by trying to replace ineffective encodings with those that are effective. When Bryan and I redesigned the original tree map, we used the lengths of bars that share a common baseline to encode every variable because this approach works well perceptually. There was no reason to use an encoding method that suffered from perceptual problems.

Stephen Few

Posts: 1
Reply with quote  #7 

Bill Gates: Kids are dying, but we have the solutions
1 Gates quote and graph.PNG

Related information

Washington Post             27 Dec 2014    Source of Bill Gates’s quote and a copy of the Wired graphic.

Wired by Lee Simmons   15 Nov 2013    Source of the graphic and a good introduction to the topic.

@BillGates                       18 Nov 2013    Tweet the graphic to a large following

GBD Compare                                           Source of the data; site supports interactive data exploration.

Perceptual Edge              10 Jan 2014     Bryan Pierce and Stephen Few redesign the graph.

Key elements of Gates’s story

  1. The number of kids dying from preventable diseases… continue[s] to decline.

  2. Those numbers are still far too high.

  3. Fewer kids are dying.

  4. We have the solutions.  

  5. But we need to keep the up support where they're being deployed, and pressure to get them into places where they're desperately needed. 

  6. More diseases are on their way to being eliminated. 

  7. More kids are going to school

Key elements suggest possible graphs.

Reviewing the key statements and related information generated the following.   

  1. Continue[s] to decline:  Suggests a time-series: line plot deaths x year.

  2. Still far too high:  Needs a comparison group.  Try Developed vs Developing countries.

  3. Fewer kids are dying:  Filter to show only kids data.  Age < 15.

  4. We have the solutions:  Low death rates for Developed countries supports this.  They also provide an achievable target.

  5. Solutions deployed vs. desperately needed:  Countries differ in commitment.  Map showing countries color coded based on outcomes.  Variation in adjacent regions suggest governance issues.  Immunizations might add support to governance being an important factor.

  6. Diseases being eliminated:  Time-series of WHO diseases targeted for elimination.  Line per disease by developed vs. developing.  This is a component of 1 and may add more noise than signal.

  7. More kids are going to school:  This is an unrelated story.  Drop it.

Points 1-4 tell the main story focus on them.  Then see if 5 & 6 add to or distract from the main story.

Sub story

The original Wired article by Lee Simmons has a secondary focus: introducing the IHME site and its data exploration capabilities.  The availability of a good interactive data site allows the redesign to focus on cleanly telling the core story.  The redesign can direct readers to the IHME GBD Compare tool to dig into the details.  Treemaps are the primary visualization tool in GBD Compare which suggests thatPorostocky may have used a treemap to provide a clean transition to the sub story.

What I see when I read the key story elements.

2 two lines with text.PNG 

 While this is how I conceptualize the graph it is not how I would present it.  The following graph is my publishable version paired with a loose restatement of the Gates quote.  

3 publishable graphs and text.PNG 
See Wired for an introduction to these data and IHME for detailed data exploration.


Posts: 33
Reply with quote  #8 
I prefer the 'conceptual' version because it has the story clearly stated and needs no interpretation from the viewer. I would have thought I would prefer the 'publishable' version but I don't. Interesting
Peter Robinson
in Brisbane, Australia
Previous Topic | Next Topic

Quick Navigation:

Easily create a Forum Website with Website Toolbox.