Focus on Your Data!
Keep Graphics Clean and Simple
Edward Tufte has
written several books about the
visual display of information. They contain a wealth of detail about how
to display data visually in a clear and honest manner - way more information
than I can provide here.
Posters could be improved greatly by adhering to a few principles
(see Tufte for details - and he'd likely scream to see these in a bullet list).
Please keep in mind that this panel only skims the surface.
There's much more that can be done!
- As with everything else on a poster, your job is to communicate
clearly and directly with your audience.
You should work to eliminate anything that distracts from this.
- First, be sure you plot the relationship you want to show. If you
want to see the change in populations through time, the X-axis should be
time and the Y-axis should be population. Nothing else makes sense.
- The ratio of information ink to total ink
in a graph should be as large (close to one) as possible.
Information ink includes
the markings on the graph that show the data.
Anything else is just distracting - Tufte calls it "ChartJunk".
In the above example, only the two lines show your data.
The axes give the reader a sense of the range of your data.
Everything else is ChartJunk: the grey background, the grid lines,
the legend, and the overly-abundant axis labels.
- Maintain graphic integrity - tell the truth about your data!
- - physical area on graph should be proportional to numbers represented
- - data should not be displayed out of context
- - explanatory material and labels should be included on the graph
- - time series with money should be inflation-adjusted
Here's a very simple example showing the relationship between two populations through time.
The main point to be made is that the lynx (predator) and hare (prey) populations
oscillate through time in a somewhat predictable manner.
[ This is a classic example of predator-prey oscillations from the ecology literature,
and has been interpreted as evidence of the regulation of prey populations by predators. ]
Straight out of Excel, here's what you get:
There's a lot of ink here that doesn't convey information relevant to
the main point you're trying to make. Plus it's ugly.
- Grey background: not only does it provide absolutely no information, it's
also unsightly. After you remove it, you will likely have to darken
some of the lines.
- Grid lines: it's very unlikely that your audience cares about the
exact values at each data point - it's the pattern that matters.
The grid lines compete with the pattern you're trying to show.
- Legend: it's taking up space that would be better spent on the graph
- X-axis: The labeling between tick marks is confusing.
- Axis scales: Because the pattern is the main focus, we don't need
to have such finely detailed scales - just enough to provide a sense
of the range of values plotted.
- Y-axis label: Why make the reader tilt his or her head to read?
- Legend: Why make the reader look back and forth
between lines and legend? Just label the lines - then eliminate the legend.
- Line types: Use color and line type to differentiate - this will help
people who have color impaired vision, and also any grey-scale copies of the
poster you make (as for handouts).
It's easy to make a graph that looks cleaner and has a higher
ratio of information-to-total ink:
Without all the distractions, the relationship comes through loud and clear.