Pointing in Entertainment-Oriented Environments:
Appreciation versus Performance
J.H.D.M. Westerink and K. van den Reek
Abstract For these latter environments, research on pointing devices is
quite extensive: it includes the physical characterization of the
Multimedia applications for consumer entertainment often pointing devices (Whitfield, Ball & Bird, 1983, or Mackinlay,
employ a point-and-select interaction style, borrowed from Card & Robertson, 1990), although most attention has been
more task-oriented computer applications. The environment given to what is generally known as Fitts' Law (see Fitts &
of use and the pointing devices involved are so different, how- Peterson, 1964). Later refined by MacKenzie (1992) and also
ever, that a higher importance should be attributed to the users' by others, Fitts' Law describes the relation between the time
appreciation of the pointing device than to its efficiencyor any needed to point to a certain target and the size of and initial
other objective performance measure. We set up an experiment distance to this target: for some input device types - though
to investigate how appreciation and performance measures not for all (see Card, English & Burr, 1978) - it is found that
relate. The experiment involved six different input devices, two it is primarily the ratio of target distance to target width that
CD-i titles and 16 subjects making both voluntary and pre- determines the time-to-target, and that the latter is in fact lin-
scribed cursor control movements. For the mouse-like point- early related to the logarithm of that ratio. The slope of this lin-
ing devices we obtained a Fitts' Law-like dependence on target ear line is often considered to be an indication of the
width and target distance. This was, however, not replicated 'information capacity of the human motor system', in analogy
for any of the other input devices, mainly owing to a positive to the concept of channel capacity in Shannon's information
influence of cursor constraints. Concerning the relation of the theory. Various input devices can differ in the information
capacity they support, although the values found for a specific
performance measures with the users' appreciation, we found
type of input device, e.g., a mouse, can vary enormously across
that neither time-to-target nor relative-path-length on its own
investigations (MacKenzie, 1992).Thus most studies focus on
is a reliable indicator of the users' appreciation. Together, how-
an objective characterization of the efficiency of the pointing
ever, they might explain the appreciation scores to a consider-
Introduction Pointing in entertainment-oriented environments is different
from pointing in task-oriented environments in two respects.
Firstly, the pointing devices are designed to allow a relaxed sit-
Multimedia applications are gradually entering the consumer ting position at a substantial distance from the screen. Sec-
market, often meant for entertainment, such as Philips' system ondly, the efficiency of the pointing device is much less
for multimedia applications, CD-interactive. In many of those important than its appreciation by the user although, of course,
applications, a point-and-select paradigm is used as the basic efficiency can be one of the factors contributing to apprecia-
interaction style, adopted from more computer- and task-ori- tion. This can be generalized to other performance measures
ented environments. besides efficiency: also their use in evaluating a pointing device
SIGCHI Bulletin Volume 27, Number 1 January1995 41
is dependent on the extent to which they reflect the users' For entertainment software, we used two CD-i titles: 'Time/
appreciation. Life Photography Course' and 'Jigsaw', an electronic puzzle
game. Both employ a 'point-and-click' interaction style.
In this article we will therefore attempt to investigate both
(subjective) user appreciation and (objective) performance In order to simulate a typical living-room and leisure-like situ-
measures, and more specifically how the two relate. ation, the television set was placed on a low table. The subject
was seated in a comfortable armchair at a distance of some 2
Experiment Description metres from the television screen. Most of the pointing devices
could be held in the hand or on the lap. A small horizontal
An experiment was set up to investigate the 'subjective-objec- plate was attached on the arm-rest of the chair to serve as a base
tive' relation for a set of 6 different pointing devices, using two for the mouse devices, so that they, too, could be used in a
existing CD-i titles as multimedia applications. The experi- relaxed way.
mental design was a within subjects design with CD-i titles as
a between subjects condition: each subject worked with all six Procedure
pointing devices, but with only one of the two CD-i titles. The subject was made comfortable, and the experiment leader
explained the way CD-i works (basically the point-and-select
Subjects paradigm) using the trackerball.
16 subjects, aged between 20 and 40 years, participated in the
experiment, 50% female and eight for each title. Some were Then the subject was invited to explore the title and pointing
colleagues and some drawn from the Institute's subjects pool. device. The experiment leader stayed in the room and, in order
They were specifically selected for their interest in the content to create a relaxed atmosphere, adopted the role of a friend to
of the title they would work with, as previous experiments had talk to and enjoy CD-i with. After 5 minutes, the experiment
shown that without this interest no serious results could be leader asked the subject's opinion on the pointing device and
expected. to express overall appreciation in the form of a score on a scale I
from 1 to 10. After that, the subject was requested to perform
Equipmentand Environment a series of prescribed movements. They involved four repeti-
The core of our experimental set-up consisted of a professional tions of a pattern that consisted of 10 consecutive movements.
CD-i player (a Philips CDi602), with additional software that Each movement was based on screens available in the particular
allowed us to event-log all cursor positions on the screen and CD-i title, and was indicated in terms of a target area to go to,
all button presses on the pointing device. The player's output while a button was to be pressed once the target was arrived at
was to a normal television set (Philips Matchline) for presenta- and this last position then fimctioned as the starting position
tion of both the visual and auditory content of the CD-i titles. of the next movement. The movements varied in target dis-
tance, target size and target direction. I f needed, the subjects
The six pointing devices are described in Table 1. The two could look up the intended movements on a reminder posted
mouse devices were identical as far as their physical character- next to the T V screen. During these prescribed movements the
istics were concerned, and the same holds for the two joysticks. experiment leader took care not to interfere with the subject's
The 'Airmouse' is an absolute device sensing its position and tasks unless a correction to a mistake needed to be pointed out.
orientation in space. As required for CD-i, all these pointing
devices had two different selection buttons. This same procedure was then repeated for the other five point-
ing devices. The order o f their presentation was trackerball,
mouse devices (pseudo-random), joysticks (pseudo-random)
and Airmouse. Though this order is not neatly counterbal-
Table 1: O v e r v i e w of pointing devices anced, it was deliberately chosen to allow the subjects to com-
pare impressions on similar pointing devices.
Characterization Typesof Data
The objective data were derived from our event-logging record-
airmouse am absolute pointing device, sensing ings. We selected two performance measures. The first is time-
its position and orientation in to-target, well known from Fitts' Law, which we defined as the
space time needed for a specific movement, starting with the button
press at the end of the previous movement, and ending with the
joystick j+ relatively high speed/control ratio button press on the intended target position. The other perfor-
mance measure we call relative-path-length, and we defined it
joystick j- relatively low speed/control ratio as the length of the path between the beginning and end point
of a certain movement, divided by the (minimum) distance
mouse m+ high resolution of 350 dots per between these two points. Thus both performance measures
inch are calculated for each repetition of each (prescribed) move-
ment for each pointing device and for each subject.
mouse m- low resolution of 250 dots per inch
trackerball tr standard CD-i remote control 1' This scale is familiar to all Dutch subjects as it is used in all pri-
mary and secondary schools.
4-2 January 1995 Volume 27, Number 1 SIGCHI Bulletin
.As described above, the experiment also yielded subjective data
on the appreciation of the pointing devices in the form of a
score on a 10-point scale. These judgements were obtained
from each subject for each pointing device.
Results and discussion
"= ~ J+
The appreciation scores for the pointing devices were averaged
over all subjects (and thus over titles) and are shown in Table 2. / . /J'/" . . TR
Although subjects did differ in their appreciation of the various
pointing devices, the difference between titles were minimal,
and the average scores for the pointing devices describe most of ~ MS
the variance. They show that the high-resolution mouse, the ~ ~ -ML
low-resolution mouse and the trackerball are appreciated J
equally and best, followed at quite some distance by the Air-
mouse and the low-speed joystick. The high-speed joystick was
considered worst of all.
I I r
210g2. ~ W ) 4
Table 2: Average appreciation judgements. Scores
were averaged over subjects (and titles) Figure 1: Time-to-target data for all prescribed
m.ov, ements in the Jigsaw title. The values areplo~ed in
Fitts Law format, that is according to the logarithm of the ratio
Pointing device Average Standard error-in- of twice the target distance A and the target width W. Thus var-
score the-mean ious combinations of distance and width can result in the same
mouse m+ 8.0 0.2
mouse m- 7.9 0.2 of approximately 200 ms/bit. For the other three pointing
devices we do not find Fitts' Law, a fact mainly due to the low
trackerball tr 7.8 0.1
values found for a long movement to a small target near the
airmouse am 5.4 0.5 edge of the screen (points on the extreme right in Figure 1).
This low time-to-target value reflects the beneficial influence of
joystick j- 4.8 0.3 the cursor constraint which prevents the cursor from moving
off the screen, and we observed that the subjects made active
joystick j + 3.4 0.4 use of that feature. For the Airmouse the constraints do not
help since a movement off screen has to be fully recovered by
an absolute pointing device, and for the mouse devices the con-
straints are apparently not needed.
The time-to-target values were averaged over subjects and rep- Although the results for the 'Time/Life Photography Course'
etitions in order to yield one value for each prescribed move- title turn out to be slightly more noise they are similar to the
ment with each pointing device in each title. For half (5) of the results for the 'Jigsaw' title: Fitts' Law applies for the mouse
movements in each title, these averages turned out to be con- devices and the Airmouse, but not for the joysticks, and a ben-
siderably (1 to 2 s) higher than for the other half, as a result of eficial effect o f cursor constraints is found for all relative point-
small delays due to system processing. Although these delays ing devices.
increased time-to-target values, subjects hardly complained
about this, mainly because the application feedback was ade- In a similar way we calculated the relative-path-lengths for
quate in signalling the disabled interaction. those movements that were unaffected by extra waiting times.
For most movements, the average actual path length appeared
Because we want to describe the pointing devices and not the to be between 1 and 3 times the minimum required distance,
CD-i system's processing, we limited our analyses to those and showed little relation to the time-to-target values for the
movements that were unaffected by the waiting times. Figure same movements.
1 shows the average time-to-target values for the 'Jigsaw' title
as a function of the ratio of target distance and target width (in Comparisonof Appreciationand PerformanceMeasures
the direction of movement). The graph shows that for both Figure 2 shows the scatter diagrams of both performance mea-
mouse devices (whose results were not significantly different) sures versus the appreciation scores, and indicates that there is
Fitts' Law applies as expected, with a slope of about 130 ms/ certainly some relation between the two. In a way this is sur-
bit, which is well within the range given by MacKenzie (1992). prising because aspects of grip and handling (see, e.g., Neer-
Also for the Airmouse Fitts' Law proves to apply with a slope voort & McClelland, 1992) must also have had their part in
SIGCHI Bulletin Volume 27, Number 1 January 1995 43
the overall pointing device judgements, but can only be indi- considered to be so much better than the Airmouse and about
rectly reflected in the cursor control patterns and the values just as good as the mouse devices. The relative-path-length
derived from them. On the other hand, the difference between measure, on the other hand, is better at reflecting the clustering
the two joysticks - which are identical as far as grip and han- of appreciation scores for the three best pointing devices (m+,
dling are concerned - confirms that gain characteristics can m- and tr) and relating them to the Airmouse but fails, at least
also influence the user's appreciation and performance. for the 'Jigsaw' title, to explain why the slower joystick (j-) is
generally found to be worse than the Airmouse. Thus the data
suggest that the two performance measures complement each
Neither the time-to-target measure, nor the relative-path- other in reflecting the appreciation scores.
length measure, however, fully reflects the users' appreciation
judgements. The time-to-target measure reproduces the rela- Conclusions
• tion between the appreciation scores of joysticks and Air-
mouse, but fails to provide a reason why the trackerball is The main conclusions can be summarized as follows. The
time-to-target measure present in much of the research involv-
ing Fitts' Law overemphasizes the importance of efficiency in
Figure 2: Comparison of appreciation and performance mea- leisure-oriented environments. Indications of this overempha-
sure for both titles.The upper graph concerns the time-to-target sis are the lack of complaints in waiting-time situations and the
data, the lower concerns the relcztive-path-length data. Both are contribution that the relative-path-length measure can make to
plotted against the average judgement scores.
explaining users' appreciation judgements.
• Time-Life Acknowledgments
_ • Jigsaw The authors would like to thank Cor Luijks and Frank Stevers
of Philips CE - Interactive Media Systems for their contribu-
tion in writing and implementing the event-logging module.
nJ- Card, S.K., English, W.K. & Burr, B.J. (1978) Evaluation of
2-- mouse, rate-controlled isometric joystick. Step keys and
text keys for test selection on a CRT. Ergonomics, 21,
O J+ OJ-
Fitts, P.M. & Peterson, J.R. (1964) Information capacity of
•AM discrete motor responses. Journal of Experimental Psy-
ML chology, 6, 103-111.
0 "= • I I MS~ Mackinlay, J., Card, S.K. & Robertson, G.G. (1990) A seman-
2 4 6 8 tic analysis of the design space of input devices. Hu-
Averages c o r e
man-Computer Interaction, 5, 145-190.
MacKenzie, I.S. (1992) Fitts' Law as a research and design tool
in human-computer interaction. Human-Computer In-
teraction, 7, 91-139.
• Jigsaw Neervoort, P. & McClelland, I.L. (1993) CD-I remote control
handling behaviour. Philips Corporate Design Internal
Report No. AE 93-07.
Whitfield, D., Ball, R.G. & Bird, J.M. (1983) Some compari-
• J+ sons of on-display and off-display touch input devices
for interaction with computer generated displays. Ergo-
nomics, 26, 1033-1053.
1 .x+ Institute for Perception Research
PO Box 513
.~ J-m _ . m~ TRml 5600 MB Eindhoven
0 "~ l "J- I MsMTLI~ The Netherlands
2 4 6 Average score 8 Email: firstname.lastname@example.org, nl :Ht
44 January 1995 Volume 27, Number 1 SIGCHI Bulletin