I had occasion to build some bubble charts the other day, which made me realise what an under-used resource they are. I almost never see them used! So here's a post to show how useful they can be. Here is just one use, of course.

The data I had was seven 'models' that were built across two states. A model in this context referring to a score given to every person in a given state signifying a property of that person: their likelihood to vote, their likelihood to support a given candidate or their likelihood to support a particular issue. One state had three such models, and another state had four. I wanted to prepare a PowerPoint slide comparing these models. In particular I wanted to compare the distribution of scores across the population. Enough words. Here's the traditional way that this data might have been displayed:


I made this as clear as I could in this format. No axes, clear labels, colour differentiation between states. But still its hard to compare, and there is a LOT of clutter required. Here is the same data presented with bubble charts:

Now you'll see that its perfectly possible to compare across states, and across models. Some are 'lumpy', some are smooth. Some reach right to the low scores and others are concentrated in the centre.

A quick post. But I hope an interesting one. Next time you have a problem like this, consider a bubble or two :)