Department of Information Systems
University of Maryland, Baltimore County
Baltimore, Maryland 21228
410-965-6961
whisenand@umbc2.umbc.edu
Under OS/2 and Windows, default icons are
generally square and approximately 1.5 cm per
side (0.3 cm per side for minimized icons). The
use of icon-like targets, in size and shape, is a
viable alternative to actual icons (which could
carry confounding semantic meanings) to test
direct manipulation devices in human-computer
interactions.
This study examined the relationship between
movement time and angle of approach for a
mouse as the direct manipulation device for the
selection of icon-like targets.
Software. The study was run using software
developed using Microsoft's Visual Basic for
Windows, 2.0.
Amplitudes and angles were measured from the
center starting position to the center of the
target. Each trial presented an icon-like square
target (0.25, 0.5, 1.0, or 1.5 cm/side) at an
amplitude of 1, 2, 4, 8, or 10 cm, and at an angle
of 0, 45, 90, 135, 180, 225, 270 or 315 degrees.
Each subject completed four trial blocks. A trial
block consisted of 160 randomly presented,
successful trials representing each target size,
amplitude, and angle of approach (4X5X8 within
subjects design). The subject positioned the
cursor arrow in the starting block and, when
ready, depressed the mouse button; a target then
appeared at its angle and amplitude. The subject
moved the cursor to the target and depressed the
mouse button to complete the selection. If the tip
of the arrow was within the target, the subject
went to the next trial. If the tip of the arrow was
outside the target, an error was recorded and
that trial remained in the pool of trials to be run
within the block. A trial block was not completed
until all possible targets (160) were successfully
captured. At the end of each block, the subject's
performance feedback was displayed with the
number of errors and the mean movement time
for each block. Between blocks, subjects
commented on their interaction, and were told to
flex or stretch their arm and hand. The mean
participation time for each subject was 45
minutes with a one minute break between blocks.
The software recorded subject, block, size,
amplitude, angle, response and movement time
for hits. Response time was measured from the
appearance of the target on the screen to the
movement of the cursor off the starting block.
Movement time was measured from the time the
cursor moved from the starting point to the
selection of the target. Errors were written and
analyzed separately.
Figure 1Comparison of MT for each
angle averaged across target size and distance. (CAUTION: This is a very big image.
You may not be able to open it.
A MANOVA for repeated measures showed a
significant effect of angle on MT, F(7,15)=5.39,
p<.005. Pairwise contrasts (Bonferonni
corrected, p <.05) showed significant differences
between angle pairs 0/45, 45/270, 135/180,
45/180, and 45/315. A significant size by angle
interaction was obtained, F(16,6)=3.98, p<.05.
The interaction was interpreted by examining
the variability in MT across all angles within
target sizes. The variance for the .25 cm target is
more than twice the variance on any other target
size. Pairwise contrasts showed significant
differences between target-size pairs .25/1.0 cm
and .25/1.5 cm.
From these results, arranging icons, 0.5 to 1.5 cm
square, at a distance of 4 cm or less, and with an
angle of approach of 0, 180, or 270 degrees from
the starting point on a computer desktop would
improve performance of icon selection, where
movement time is an issue.
Further research into the effect of the angle of
approach on target selection tasks will attempt to
define its role in Fitts' model.
Abstract
Over the past decade, research related to various
aspects of human-computer interactions has
become increasingly prominent within the
human factors field. In that regard, the speed
and accuracy of human motor movements
associated with computer input devices has often
been modeled by Fitts' law. However, most such
analyses have not considered the angle of
movement as a factor. Accordingly, the present
study investigated the effects of the angle of
approach for a mouse as the input device to select
icon-like targets presented on a VDT. The angle
of approach had a significant effect on movement
time.
Keywords:
Icon selection, Fitts' Law, input
devices, human performance modeling.
Introduction
With the recent proliferation of windowing-type,
icon-based software systems, the mouse has
become the most utilized direct manipulation
interface device. While several studies have
compared the mouse with other devices, few have
examined the effect of the angle of approach on
movement time. Fitts' law [2] has proven to be
robust under a wide variety of conditions and
subject populations for target acquisition tasks;
however, it does not account for the angle of
approach. Selection of icon-like objects on a
screen is not limited to the horizontal movements
of reciprocal tapping (Fitts' task). For example,
higher mean movement times (MT) have been
reported for helmet-mounted sights [3] and
joysticks [1] along diagonal axes. For the mouse,
angle of approach had no effect on MT [1].
However, the angles used did not exercise the full
range of movement on a computer desktop
(screen), as movement was restricted to the right
of the vertical axis on the left-hand side of the
screen.
METHOD
Subjects
Twenty-two undergraduate students in
Information Systems participated for course
credit (9 male, 13 female, and 1 left-handed). All
subjects reported over 30 days experience with
personal computers (PCs) and mice.
Materials
System Configuration. This experiment used an
Eclipse PC 486/33, 4M of RAM, 5.25 and 3.5 inch
floppy drives, and an Orchid 280oF graphics
board with 1M RAM. Cursor control and target
selection were controlled by a Logitech two-
button serial mouse. The screen on the CTX
Proscan 17 inch monitor was 31x23.5 cm
(800x600 pixels). The background color for target
selection was black, the starting block was gray,
and the targets were yellow.
Procedure
Participants were tested individually. Subjects
were interviewed to gather demographic and
experiential data. Subjects read a single page of
written instructions and were walked through a
sample trial block (16 targets) to familiarize them
with the task and environment. Subjects were
instructed to move the cursor and capture the
target as quickly and accurately as possible.
RESULTS
Angle of Approach
The angle of approach had a significant effect on
mean movement time overall, in pairwise
contrasts, and in
interaction with the smallest target size.
Figure 1 presents a comparison of MT for each
angle averaged across target size and distance.
Error Rate
The error rate for all subjects across all blocks
was 2.96%. Card [1] reported mouse error rates
ranging from 3.5% to 5%. The lower error rate
could be attributable to the subjects' experience
with PCs and mice.
CONCLUSIONS
The angle of approach for the selection of icon-
like targets by a mouse had a significant effect on
movement time. Movement times along the
diagonals were slower than along the horizontal
and vertical axes. Upward movements (45, 90,
135 degree angles) were slower than downward
moves (225, 270, 315 degree angles), while
movement on the horizontal axis was the most
efficient.
