UsableMarkets

… everyone copies the bejeezus out of it.

Take the periodic table of elements. It’s not unusual to see variations on this tabular format seemingly everywhere. If you doubt me check out the work of bloggers more patient than I who have collected samples from across the internet here and here and here. Everything from periodic tables of desert, to pokemon characters, to search operators (yeah, I don’t get it either).

What is it about this tabular chart that works so well?

Here’s perhaps the best version I could find on the internet.

click to increase the atomic mass

The chart, while somewhat impenetrable to the layman (like me!), is actually organized by some very simple principles. The rows are organized by the number of electron shells for each type of atom. The columns by the properties that those atoms exhibit (alkali metals, alkaline earth metals, halogens, etc.) Overall, going from left to right, and then down, the atoms are arranged by the overall number electrons it has. And finally there are larger groupings organized by color (metals, noble gases, etc.).

These are four variables nicely displayed in a two-dimensional table. Is this the only way to display this information? No, but it is the best known, and ubiquitous in high school chemistry labs everywhere. Hence, people feel the need to do things like this to it.

click to get to the bottom of the glass

Of all the periodic charts of xyz, this is probably the one that tries the hardest to incorporate multiple, meaningful variables to its organization. However, while it’s clear that the columns are beer styles, and families of beer (ale, lager, etc.) are collected into larger groups, it’s hard to figure out the how the rest of the table is organized. But no matter, its got the right spirit. Drink up!

Thanks for reading.
~alex

It’s really frustrating to see the number of data visualizations that use maps stupidly. For example, take this crap map from the FT.

It only allows the user to look at one variable at a time. Surely there must be a better way than clicking back and forth between the offered variables to see if there’s some correlation between them. Say, a little dynamite icon () to indicate recent protests, or something like that. Better still, perhaps some other type of chart, not a map, that allows the user to see all four variables at once. It’s not hard to do. Really. But instead we get the map.

Why? I can only guess it’s because designers think maps are visually interesting, and therefore should get a free ride on the chart junk train. And the FT is not the only one. Mint, among many others, has opted for using a map visualization when none was required.

Seriously, is it important for me to view the cities geographically here? No, the important comparison is not state by state, but rather the change in discretionary spending over time, and compared to each other. The map adds nothing of value. It doesn’t even help you compare the cities easily.

But instead of spending this whole blog post ranting about negatives, lets look at positive examples and what makes them work.

From Radical Cartography we have this example.

The story is pretty clear, eh?

Here the map does three things quite well. It encodes rich, non-binary, data — in this case value of land per square foot; it organizes the data in a meaningful way (i.e. the data points listed independently are less meaningful than when listed together); and it works in a labeling capacity (i.e. how tedious would it be to see labels on this chart?).

The real test, of course, is if you tried to chart this data without using the map. If you don’t know the answer, let me tell you: it would be dreadful.

Here is another excellent example, this time from NPR, of how to use a map to present data.

Again, the data itself is geo-specific (and I mean here related to features of geography, like coasts, not arbitrary state lines and such). And again there is rich data (wind speed), and multiple layers of data that work together to tell a story. In this case, to rely on wind power, we’ll need to build more power lines from places where wind speed is in abundance, to places where it is not.

In summary, to make your map based chart not chart junk, make sure:
- the geography is actually meaningful to the data
- the data is rich (i.e. multivariate)
- the presence of the map also labels the data

Thanks for reading.
~alex