My US GDP/”Population” Map
Yes, the quality on this is not super high, but the point is to experiment with an idea.
There has been a lot of controversy floating around about a map.
… to name a few.
The map being this: http://www.reddit.com/r/MapPorn/comments/1y371s/
I have a thing for trying out abstract ideas in spacial analysis. Like:
As a western Canadian, I’m very aware of the east-west dynamics in the US and Canada, so I thought I would follow a hunch and make this map. I think it demonstrates nicely how lop-sided the US is.
The major consultancies are turning out to be a goldmine for content for demonstrating the potential for improved data visualisation design. Today’s winner is PWC, though their error is not terribly sinful and with some effort one could probably find something worse.
This is a demonstration of the power and importance of left-to-right. The original violates two assumptions readers make about left-to-right:
- Things further to the right are later in time
- Things further to the right are higher, bigger, better.
In the original 2014 is shown to the left of 2013 and the bars are ordered high, medium, low.
Quick answers to basic questions:
- Are things good or bad? Bad. The bars make a download sloping shape.
- Are things better or worse than last year? Hard to tell.
What if we simply change the left-to-right ordering of everything?
So now we at least get the impression that things are good before we start studying the labeling in detail. We can also scan across, say, the red bars in order to see how things are changing with time without having to realize that the chart’s timeline is backwards.
I also experimented with a line design here:
- Maybe the “things are good” message is a little muted
- We can certainly see the trends over time, though. Generally flat, but with fewer doubters and a bit more optimists.
- The central story of the graphic becomes the crossed lines in Business headcount. Unfortunately this is probably indicating an error in the original graphic. Scroll up and see the probably incorrect data labels
Another day, another poor visual. Apparently I have a quest to complete against bad visuals from consultancies. Today it’s KPMG from this tweet:
A 4×4 grid of single-value donuts? Are you kidding?
Like last time, I will show you how this is bad with simple re-design.
First let’s try simply cleaning up the graphic a bit:
It looks better already! If I can improve your graphic by erasing the visual elements that you used to encode your data, this is a very bad sign.
And finally the obligatory bar-chart redesign:
Extra patterns become visible:
- EMEA generally more pessimistic than Americas, Asia-Pacific of Overall
- EMEA substantially less impressed by More sophisticated/granular insights
KPMG can join Accenture on the list of consultancies that should practice better data visualisation, especially when generating thought leadership on the topic of data and anatlytics.