Yes, you are not dreaming, this really is a new blog post. Take a moment to compose yourself.
Ready? Okay will give you a minute more.
Right let's begin. So what may you ask has driven me to putting finger to keyboard and actually writing something? Well sadly its not a good thing. I was having a quick look on twitter in my lunch break when this came up in my feed.
Now this is pretty bad even by the Guardian Datablog standards (remember when they used to be a great example of dataviz?)
Its showing the consumption of chocolate per capita for a selection of countries. There are two major issues with this viz.
Using areas to compare values
As we all know by now, we are really bad at comparing area sizes. FACT. We are not designed to do this and so any viz that uses areas to compare sizes is not generally a good idea. (I say generally as the old Dataviz maxim of "It depends" means that there is always an outlier) In this case I assume the size of the square corresponds to the $ spend of the country in question. So I am being asked to compare area sizes to compare countries spend against each other. That in itself is pretty bad, but to make matters worse...
Using mapping when there is NO NEED
The default response to when a lot of people see some geographic data in their data is "I MUST MAKE A MAP". Seriously its like a short circuit to some geo-cortex buried deep in the brain. (side note must chat to mapping Zen Master Allan about that). Once they see that data every other viz option disappears and all they can see is oceans, continents.
Don't get me wrong, maps are great, they look great and Tableau has some really nice mapping functions, but not all data needs to be mapped. Mapping data allows us to see regional trends that you can't get any other way.
In this case there is zero location based information in the data, just the country name.
Then there are a few minor issues
Using a cartogram, badly
We are being asked to compare areas, which is hard, when they are not next to each other, harder, and doing it on a cartogram type map. What does this add to the party?
Colouring by range
They have coloured the boxes according to the $ spend, fair enough, but then they have used ranged discrete colour categories rather than a continuous measure. This means that i can't see any variations within each set of countries.
Lets make this over.
Heres the same data, in a simple bar chart. I did toy with making it a fancy one but that's not the point of this makeover. I did however change the colour of the bars to be chocolate colour, well you gotta do a little bit haven't you.
- The Swiss really really like chocolate.
- Norway, Ireland and the UK are pretty keen on it too.
- Norway likes it more than the other Scandinavian countries.
- The USA spends a quarter of the amount Switzerland spends.
So whats the take away message
Just because you have geographic data, it doesn't mean you should make it into a map. Experiment with other viz types, sometimes the simplest one is the best one for the job. Remember, if you are not conveying the information in the data, you are not vizzing it right.