Discussion


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
 
 
 


Reply
  Author   Comment  
DataInAction

Registered:
Posts: 12
Reply with quote  #1 
I was recently reading an article published in the McKinsey Quarterly (management consulting company newsletter) which was discussing the ability of a special kind of map graphic to allow you to “understand the impacts of borders, distances, and differences”.  It basically involves sizing geographic entities by the amount of a given metric.

After reading an article and viewing several examples  I must admit I'm completely perplexed as to their value.  It looks to me like one of those techniques that "looks cool" but relates no useful information.  

The problems with this technique are numerous. I ended up doing a full analysis of the problems (and even created a different version using scatter plots).   I would be interested in if anyone else has seen these diagrams used and if they have any value at all.

As I tried to use these maps to obtain the information referenced in the article I encountered a number of difficulties.  Here are a few:
  1. It’s very hard to perceive which values are higher than normal and which are lower except in extreme cases.  The United States and UK are big, but what about Australia.  In addition I can’t answer basic questions like “what are the three largest markets for world wide box office dollars?”  This is never clearer than when the author states “China is so much smaller in Map 2 than on the reference map”.  I find this assertion difficult to read from the maps shown.
  2. This diagram requires extremely high amounts of labeling to even be readable.  Where is Greece?  What is that country in Africa that is the only one visible?  
  3. If I’m trying to compare the base map to the rooted map (diagram 1 to diagram 2 in the original) I have no idea where opportunities truly exist and what the relationships between the variables are.  Where are the markets with high box office dollars but low US penetration?  The diagram doesn’t tell me this piece of crucial information.
  4. Country geography is deceiving.  Why is Alaska huge?  Is it because a great deal of box office $ come from Alaska.  No, it’s simply part of the US but that is not obvious from the map.  Is Northern Ireland bigger than it should be because it is part of the UK in this map?  I suspect so.
  5. Even if I’m trying to show distances, a global map is a poor way to do it.  If I am a Canadian company interested in the effect of distance on my product sales, it is impossible to tell from this diagram that Tokyo is 4,700 miles from my border/port but Istanbul is over 5000 miles away from Toronto.
  6. Many of the other flaws are also exposed based on one of the fundamental flaws of this type of map:  It is using the size of a country (its square miles essentially) as the basis for which importance should be measured against.  But this is a poor “base setting” for a business intelligence analysis.  In this type of analysis, Canada and Russia for example will almost always “shrink”, not because they are not important, but only because their land masses (relative to their population) are so huge.  A dense country like Germany will almost always show up bigger not because it necessarily contributes a huge absolute amount to the total, but rather that it’s (relatively) small size makes it more likely to be “inflated” by any given metric.
  7. One other issue which is not specific to the map chosen but bears mentioning as it makes the diagram harder to read is the apparently arbitrary selection of colors to show increasing domestic market share.   By not choosing a graduated scale that would intuitively show me values from highest to lowest the diagram forces me to constantly have to look back at my legend to know which values are high and which ones are low. 

Below is the link to the original article.  It's helpful to read this to understand the authors intentions.  He seems like a really smart guy which makes me question/wonder why I don't "get" this technique.

Here is my full analysis of the issue as well as thoughts on a "re-do":



The first two charts below are his "before" and the third one is my "after" version of the graphic (combines both charts values into one):





The "after" version as a scatter plot.  It shows market size as dot size, distance on the x-axis, % of box office US on the y-axis, and domestic percentage as color.  For full details go to:

Anders

Registered:
Posts: 18
Reply with quote  #2 

I've always thought the same about these maps; especially with regards to the density of the population. Yet another point is that a map is a two dimensional representation of a three dimensional globe, which makes the land areas eschewed to begin with. The classical example is the huge size of Greenland compare to Africa.

sfew

Moderator
Registered:
Posts: 814
Reply with quote  #3 

Geo-spatial displays of this type are called “area cartograms.” They are a silly invention. I’ve never encountered a justification for their use. When values are displayed on a map, geography provides important context, otherwise a map isn’t needed. By distorting geography, area cartograms make the context unrecognizable.

Your analysis of these examples is thorough and insightful. Unfortunately, even smart people often display data poorly, usually because they haven’t been trained to do it well or to assess the merits of a given display. Even without training, however, simply by using one’s eyes it should be easy to see that cartograms like these fail to tell their story effectively.


__________________
Stephen Few
DataInAction

Registered:
Posts: 12
Reply with quote  #4 
Thanks Stephen.  FYI, I sent a note to the editors at McKinsey Quarterly pointing out some of the problems with the maps and asking them to add my critique to their comments section.  So far, I've only heard "we'll think about it".  Oh well.  
Previous Topic | Next Topic
Print
Reply

Quick Navigation: