How to make graphs that work

Aleks pointed me to this good advice from Seth Godin on preparing graphs. Some snippets:

1. Don’t let popular spreadsheets be in charge of the way you look . . . when you show me something exactly like something I’ve seen a hundred times before, what do you expect me to do? Here’s a hint: Zzzzzz.

2. Tell a story

3. Follow some simple rules

– Time goes on the bottom, and goes from left to right

– Good results should go up on the Y axis. This means that if you’re charting weight loss, don’t chart “how much I weigh” because good results would go down. Instead, chart “percentage of goal” or “how much I lost.”

4. Break some other rules . . .

9 thoughts on “How to make graphs that work

  1. I don't see how rule 1 is of any help. If the defaults are bad, don't use them. If the defaults are good, then we should happily use them.

    Consider the strong use of conventions in print. What percentage of written material is printed in size 10 or 12 Times New Roman printed in black. The vast majority of documents obey similar conventions in margins, indentation, and other formatting. A shared set of conventions allows the document itself to become almost invisible, leaving the content itself to reach us with minimal barriers.

    A series of graphs printed with common graphing convention and designed for clarity doesn't put me to sleep. Instead, it allows me to follow the chain of argument without the distraction of chart junk, comic sans, or absurd color choices.

  2. 'Good results should go up on the Y axis. This means that if you're charting weight loss, don't chart "how much I weigh" because good results would go down. Instead, chart "percentage of goal" or "how much I lost."'

    I don't agree with this. The convention that the graph does up as quantities get greater has a physical basis, is perfectly sensible, and makes the graph more readily interpretable. The other issue is if you chart "percentage of goal" or "how much I lost" you're imposing your own structure on the weight data – either by picking a goal or a comparison date – which you may not want to do. It's a good idea to show people the raw data, without imposing your model of what's happening on it.

  3. It's pretty to think that the defaults are good. Unfortunately, the defaults for Excel graphs in Office 2003 seem to have been set by someone who's brother-in-law sells toner for laser printers. They did fix the gruesome gray background in Office 2007.

    Common conventions or style sheets for graphics ARE a good idea; I implement them with a macro that sets fonts, decimal places, background, and gridlines.

  4. What Mike said. That said, it's probably not a bad idea to mix things up a bit. In Red State, Blue State we used a pretty standard range of graphing options; maybe it would've been grabbier if we'd varied the style of graphs a bit more.

    Also, I wish we'd talked more in that book about the choices we made in modeling and graphing. Maybe readers would've gotten more swept up in the narrative if we'd involved them more in the statistical process.

  5. I agree with Alex. It would be odd if climate researchers published charts with the proportion of stuff in the atmosphere that's not greenhouse gas on their y-axes.

  6. I disagree with #3 as well. Thinking of weight loss, I think most people think first not about how much they have lost, but about what they actually weigh. And those who want to lose weight want this to go down – hence, down on the y-axis.

    With a lot of other data, there is no 'good' or 'bad'. Even with weight loss, we eventually want to reach a plateau. And there are some people who want to gain weight.

    I'd go instead with the (to me) more natural "Up is more, down is less".

Comments are closed.