A Usability Study of Workspace Awareness Widgets
Carl Gutwin and Mark Roseman
Dept. of Computer Science, University of Calgary
Calgary, Alta, Canada T2N 1N4
Tel: +1-403-220-3532
E-mail: gutwin, roseman@cpsc.ucalgary.ca
ABSTRACT
Groupware systems that use large shared workspaces generally provide
only limited awareness information about other collaborators in the
workspace. We are designing a set of groupware widgets to provide
this missing information. This paper describes a usability study of a
number of such widgets. The study has both validated our intuitions
about the need for workspace awareness information, and revealed the
strengths and weaknesses of several current designs.
KEYWORDS:
Groupware, CSCW, awareness, usability
INTRODUCTION
Compared with physical shared workspaces such as table tops and
whiteboards, shared workspaces in groupware are greatly
impoverished. In particular, systems supporting a relaxed-WYSIWIS
(What You See Is What I See) view of large workspaces often fail to
convey information about workspace awareness, the up-to-the minute
knowledge about the location and actions of other collaborators.
We have designed a suite of groupware awareness widgets to address
this deficiency [2]. These widgets augment a user's view of the
workspace with information about the workspace, the location of other
collaborators, and their actions within the workspace.
This paper describes initial observations from a study carried out
to evaluate a shared workspace system that incorporated several
different awareness widgets. We had two goals in this study. First, we
wanted to confirm our intuitions that workspace awareness is used in
shared workspaces. Second, we wanted to evaluate how well our current
widget designs support the maintenance of workspace awareness. We were
particularly interested in knowing if the information in the widgets
was easy to interpret, if they distracted users from their tasks, and
if users thought they were worth the extra screen space.
METHODOLOGY
We constructed a groupware editor for manipulating the layout of a two
page newspaper spread, allowing users to move pictures, headings, and
columns of text. Eight pairs of subjects, primarily senior
undergraduate computer science students, worked on separate
workstations. Subjects were within speaking distance but unable to see
each other. Each user could scroll independently within the layout,
and their window was large enough to view about one third of the
workspace at a time.
Pairs completed two layout tasks, each limited to fifteen minutes.
One task was performed with limited awareness information about the
other person. Subjects used either the shared workspace by itself, or
combined with a miniature view showing only the locations of objects
within the workspace. Half the pairs completed this condition first.
In the other condition, subjects used the shared workspace along with
one of three awareness widgets we had built. The first was a
multi-user scrollbar, which shows the location of each user with a
colored bar beside the "thumb" of the conventional scrollbar.
The second was a radar view, which shows a miniature of the entire
document, a rectangle for the extent of each user's view, and a
telepointer showing their mouse location. In both of these widgets,
participants are differentiated by color. The third was a local view
widget, which shows the full scale but limited region immediately
around the other user's mouse cursor.
To collect data, we videotaped the tasks, asked subjects to fill out
questionnaires, and conducted interviews to follow up particular
aspects of the session.
RESULTS
All pairs completed their tasks and produced reasonable layouts, and
made use of workspace awareness in doing so. Several of the widgets
that we tested provided useful awareness information, and most
subjects greatly favoured the conditions that included these widgets.
Use of Workspace Awareness
We observed a variety of working styles, ranging from "divide and
conquer" to tightly coupled collaboration. Regardless of the style,
there was evidence that the pairs maintained an awareness of each
other's use of the workspace, and acted on that information to
collaborate with their partner and complete their task.
Many of our observations of the use of awareness echo previous
observational studies. For example, we noticed the frequent use of
gestures [3]. For gestures to be interpreted, the receiver must see
them, and therefore the sender must be aware of the receiver's
view. Gestures were most often communicated through the main
telepointers, but people sometimes gestured by moving objects in the
workspace or by using the telepointer in the radar widget. In addition
to gestures, we also noticed the regular use of deictic references
(e.g. "move this object). As with gestures, deixis depends on
the hearer being able to see what the speaker is pointing at, or the
hearer having a mental picture of their work area.
Use of Awareness Widgets
The widgets were well received by subjects, who used them, liked them,
and often requested even more awareness information than what was
available. Widgets were used in two ways. First, subjects used the
widgets to keep track of the locations of objects, and as a high level
overview of the entire layout. For example, many subjects used the
radar to check if text columns fit on the page. Second, subjects used
the widgets to keep track of their partner's location, activities,
and progress on the task. For example, widgets assisted subjects in
discussing placement of articles with their partner, who was working
on the other page.
Feedback from subjects showed that the radar and miniature view
widgets were most useful in their task. Although subjects could see
some use for the local view and multi-user scrollbar widgets, these
did not seem to support the task better than the plain
workspace. Below, we consider these results in terms of ease of
interpretation, distraction and perceived value of the widgets.
To explore how easily the information in the widgets was interpreted,
we considered the difficulty of shifting contexts between the main
view and the widgets, and the problem of mapping colors to users. The
context shift to the radar view proved not to be a problem - subjects
reported that it was easy to identify workspace objects in the radar
view. Users found it more difficult to integrate the two different
dimensions of the scrollbar than to interpret the view rectangles in
the radar. Interpreting the local view was extremely problematic for
all subjects who used it. Several remarked on its small size and
erratic motion. The mapping between colors and people in the radar and
the multi-user scrollbar proved difficult for some subjects. One
subject said "I couldn't figure it out; I just watched for
motion."
Distraction was an issue in some widgets but not in others. None of
the subjects found that the radar view or miniature stole their
attention. Only one of four subjects found the scrollbars
distracting. However, all users of the local view found it very
disruptive, due to its erratic motion.
Finally, we asked subjects about the value of the widgets in
completing their task, and whether they were worth their screen
space. All subjects using the radar and miniature said that they found
them valuable and worth the screen space - and even complained when
we took them away. Their assessment of the scrollbars and local view
was less positive; only two scrollbar users found them valuable, while
none of the local view users found it valuable.
DISCUSSION
There are a number of lessons that can be drawn from these
results. First, overviews are useful both for managing one's own
interaction with a large workspace, and for maintaining awareness of
other's locations and activities. The overview frees the user from
having to maintain a mental model of the ever-changing workspace found
in relaxed-WYSIWIS groupware. Second, if awareness information is to
be easily interpreted, it must be presented in a familiar context,
ideally that of the workspace itself [1]. Though the radar was
physically separate, it closely paralleled the workspace, providing an
easy transition between the two. Third, awareness widgets must try to
be as lightweight as the mechanisms of face-to-face interaction, or
they will not be used. For example, integrating the two separate
dimensions of the scrollbar required more effort than finding view
outlines in the radar.
CONCLUSIONS
Overall, this study has confirmed our belief that workspace awareness
is an important part of collaborating in large workspaces. Though all
of our pairs completed the task, we found that some of our widget
designs provided useful awareness information that would otherwise be
missing from a groupware system. This information allowed for more
natural interaction over the workspace. In some cases, our widgets
were remarkably effective, leading one subject to remark "it really
felt like you were working on the same big table." We expect the
issues raised here will motivate groupware designers to continue
exploring awareness with the goal of building more natural shared
workspaces.
REFERENCES
- Dourish, P., and Bellotti, V. (1992). Awareness and coordination in
shared workspaces. In Proc. Computer Supported Cooperative
Work (CSCW '92).
- Gutwin, C., Stark, G., and Greenberg, S. (1995). Support for workspace
awareness in educational groupware. In Proc Computer Supported
Collaborative Learning, Bloomington, IN, Oct 17-20. LEA Press.
- Tang, J. (1991). Findings from observational studies of collaborative
work. IJMMS. 34(2).