Switching Between Tools in Complex Applications
Journal of Usability Studies, Volume 3, Issue 4, August 2008, pp. 173-188
Article Contents
How to read the charts
Task 1 is a problem without a formulaic solution (e.g., a programming problem). Task 2 is simple plotting task amid a large array of possible tools (e.g., plotting and formatting data). In the programming Task 1, users create, step-by-step, a function or script that can find palindromes in text. They employ very few of the available tools. The complexity in this case lies in deciding when to switch from one tool to another. In the plotting Task 2 users load data from a file and create plots for export or printing. In this case the complexity is tool choice; there are many ways to accomplish Task 2 in MATLAB®.
We recorded 14 users doing each of the two tasks. This gave a feeling for the variation in tool use across users (see below) and a baseline against that we can measure our next set of design changes. The charts show the user's progress through the task in terms of tool use as well as indicating frequency and direction of tool changes so as to give a quick, overall picture of what tools were applied, when the tools were applied, and how long the tools were applied. The following is a description of how to read each line in the figures that follow:
- The x-axis is a timeline in minutes. Each plot begins when the user finishes the first reading of the task and begins to work with the software.
- Row 1 (the bottom row) contains the helper tools that support both the command window and the editor (file browser, several ways of looking at variables, and a command history). It’s a measure of the simplicity of the programming task that they get so little use.
- Rows 2 and 3 (from the bottom) of the timeline charts show the (legacy) Command Window (row 2), from which MATLAB® was originally run, and the Editor (row 3), where users write more elaborate code to manipulate the same component mathematical functions. Each of these makes its own contribution to the product’s capability. Working between them and using each most effectively has been difficult for users from the start.
- Row 4 is Help and Documentation. Switching to help is present (in varying degrees and with different severity) at all levels of expertise, and in every task. Users’ aversion to this instance of tool switching is well documented (Grayling, 2002). We anticipate new design work in this area. There is no problem accumulating large amounts of data.
- Success of the help-seeking step is (relatively) easy to measure. (The user’s evaluation is likely to be accurate.)
- The contradiction that provokes switching from a work-piece to help is comparatively easy to observe—users discover a need for information that they must then search for—and therefore to measure and evaluate.
- Row 5 (the top row) shows a group of GUI tools used for plotting, including the window where the actual plot appears.
This visual overview of test progress allows us to locate switching problems in a test recording, and affords a description of tool use that we can compare with tests of the same task using other designs.
Figure 1. User 6 working through the programming task (top) and the plotting task (bottom).
Figure 1 shows User 6 doing Task 1 and then Task 2, using different tools for each task. In Figure 2, user 2 plots with the same toolset user 6 used for programming. User 2's style (in the period from 10:00 to 50:00) demonstrates that user 6 might have done the plotting task with the same tool use tactics used to complete the programming task, but chose not to.
Figure 2. User 2 does the plotting task (top). User 6 does the programming task (bottom).