Background - Measuring the User Experience: Collecting, Analyzing, and Presenting Usability Metrics - page 33

GENERAL TIPS FOR DATA GRAPHS

Label the axes and units. It might be obvious to you that a scale of 0 to 100%
represents the task completion rate, but it may not be obvious to your audience. You
might know that the times being plotted on a graph are minutes, but your audience
may be left pondering whether they could be seconds or even hours. Sometimes
the labels on an axis make it clear what the scale is (e.g., “Task 1,” “Task 2,” etc.), in
which case adding a label for the axis itself would be redundant.
Don't imply more precision in your data than it deserves. Labeling your time data with
“0.00” seconds to “30.00” seconds is almost never appropriate, nor is labeling your
task completion data with “0.0%” to “100.0%.” Whole numbers work best in most
cases. Exceptions include some metrics with a very limited range and some statistics
that are almost always fractional (e.g., correlation coefficients).

Don't use color alone to convey information. Of course, this is a good general principle
for the design of any information display, but it's worth repeating. Color is
used commonly in data graphs, but make sure it's supplemented by positional
information, labels, or other cues that help someone who can't clearly distinguish
colors to interpret the graph.
Show confidence intervals whenever possible. This mainly applies to bar graphs and line
graphs that are presenting means of individual participant data (times, ratings, etc.).
Showing 95 or 90% confidence intervals for means via error bars is a good way to
visually represent the variability in data.

Don't overload your graphs. Justbecauseyou can create a single graph that shows the
task completion rate, error rate, task times, and subjective ratings for each of 20 tasks,
broken down by novice versus experienced users, doesn't mean you should .
Be careful with 3D graphs. If you're tempted to use a 3D graph, ask yourself whether it
really helps. In many cases, the use of 3D makes it harder to see the values being plotted.

2.7.1 Column or Bar Graphs
Column graphs and bar graphs ( Figure 2.10 ) are the same thing; the only differ-
ence is their orientation. Technically, column graphs are vertical and bar graphs
are horizontal. In practice, most people refer to both types simply as bar graphs,
which is what we will do.

Bargraphsareprobablythemostcommonwayofdisplayingusabilitydata.
Almost every presentation of data from a usability test that we've seen has
included at least one bar graph, whether it was for task completion rates, task
times, self-reported data, or something else. The following are some of the prin-
ciples used for bar graphs.

• Bargraphsareappropriatewhenyouwanttopresentthevaluesofcon-

tinuous data (e.g., times, percentages) for discrete items or categories
(e.g., tasks, participants, designs). If both variables are continuous, a line
graph is appropriate.