Page 359 -
P. 359
12.4 Custom software 349
extent and granularity of the instrumentation, you might consider exactly what you
need to capture. In some cases, individual mouse movements and key strokes might
be needed, while other studies might need only higher-level user actions, such as
selection of the “paste” operation. Careful attention to the relationship between
your experimental hypotheses—what do you hope to learn?—and the data col-
lected may help increase your chances of success. Another successful strategy in-
volves associating each recorded action with one or more categories, allowing ease
of processing and filtering by criteria appropriate to an analysis. This approach
might allow comparison of keyboard and mouse-movement records to higher-level
task indicators.
12.4.2 RESEARCH SOFTWARE
Another class of custom software tools for automatic data collection involves soft-
ware that is explicitly created for the sole purpose of running an experiment. These
tools generally present users with a series of tasks to be completed and record data
regarding task completion time, errors, and whatever other data may be neces-
sary. The Fitts' Law, Children, and Mouse Control sidebar discusses an example
of a custom software package developed for a study of how well young children
use computer mice. Researchers interested in studying how well young children
use a mouse built a tool that tracked task completion time as well as the trajec-
tory of mouse movements in tasks that involved moving between two targets. This
study found that younger children were much less accurate mouse users than adults
(Hourcade et al., 2004).
FITTS' LAW, CHILDREN, AND MOUSE CONTROL
Full-size computer keyboards, keypads on phone and small devices, mice,
trackballs, jog wheels, and joysticks are familiar controls for computers
and other electronic devices, but familiarity does not necessarily imply
understanding. How do we use these tools? How efficient are we? What sort
of mistakes do we make? What are the factors that determine task completion
time, accuracy, and error rate? Although researchers—in cognitive psychology
and more recently in HCI—have been asking these and similar questions for
more than 50 years, detailed study of the human use of these devices can still
lead us to valuable insights.
Target selection is an important task in this area. Given multiple targets
that a user might want to select—keys on a keyboard or buttons on a graphical
user interface—what determines how quickly and accurately a user can move
from one to another? Studies of target selection performance guide the size and
selection of graphical icons, placement of buttons on a cell-phone keypad, and
many other aspects of interface design.
(Continued)