Factors Influencing User Performance with Pointing Devices

Abstract

We are working toward ways of optimally selecting and configuring input devices for people with physical impairments. This exploratory study examined the effect of the following five factors on user pointing performance: (1) Gain setting; (2) Enhance Pointer Precision (EPP) setting; (3) Target size; (4) Target distance; and (5) Input device. For this group of 17 subjects, a lower gain combined with EPP On provided significantly better performance, although the gain effect was more variable across subjects. The type of input device used had the largest effect on pointing performance. BACKGROUND An important part of computer access interventions is appropriately choosing and configuring the user's pointing device. There are many pointing devices to choose from, ranging from "standard" mice to trackballs to head controls. Once a given pointing device is selected, tuning it to the user's strengths and limitations may yield significant performance and comfort benefits. Windows XP provides two adjustable settings, Gain and Enhance Pointer Precision, that affect the behavior of the mouse and many mouse alternatives. The gain setting determines how far the mouse cursor moves on the screen for a given movement of the pointing device. Changes in gain setting may help to accommodate physical impairments, although there is very little literature in this area (1). The Enhance Pointer Precision (EPP) setting enables a complex algorithm controlling the velocity and acceleration of the mouse cursor. By default, EPP is turned on. We have found no research on how the EPP setting affects user performance. The pointing environment also influences user performance. Both intuition and Fitts' Law tell us that larger, closer targets can be selected more quickly (2). The sizes of objects such as menu items and toolbar buttons can be manipulated through the Windows control panel and from within some applications. We have found no literature on how such manipulation might enhance pointing performance for a user with physical impairments. The long-term goal of this work is to determine ways of selecting and configuring a pointing device and operating system settings to provide optimal user performance, both initially and over time. As part of this work, we conducted an exploratory study to get a sense for how the following five factors affect user performance: (1) Gain setting; (2) EPP setting; (3) Target size; (4) Target distance; and (5) Input device. METHODS Protocol The design of this study was relatively informal, to allow us to explore the influence of a large number of factors within a single data collection session. Seventeen subjects with upper extremity physical impairments performed at least four test runs of target acquisition trials. The target acquisition task used was the Aim test in the Compass software package (3). Each trial presented a single target, and the user moved the mouse cursor inside the target and clicked to select it. Each test included 32 trials: four targets at each combination of four different sizes (18, 24, 40, and 100 pixels) and two different distances (50 and 512 pixels). Each test run with a given input device used one of four combinations of gain (low or default) and EPP (on or off) settings. After completion of all four combinations, the protocol was repeated with a second input device. Subjects

Full text

RESNA 2006 Proceedings
Factors Influencing User Performance with Pointing Devices
Heidi Horstmann Koester, Edmund F. LoPresti, Koester Performance Research;
Richard C. Simpson, University of Pittsburgh
ABSTRACT
We are working toward ways of optimally selecting and configuring input devices for people with
physical impairments. This exploratory study examined the effect of the following five factors on
user pointing performance: (1) Gain setting; (2) Enhance Pointer Precision (EPP) setting; (3) Target
size; (4) Target distance; and (5) Input device. For this group of 17 subjects, a lower gain combined
with EPP On provided significantly better performance, although the gain effect was more variable
across subjects. The type of input device used had the largest effect on pointing performance.
BACKGROUND
An important part of computer access interventions is appropriately choosing and configuring the
user’s pointing device. There are many pointing devices to choose from, ranging from “standard”
mice to trackballs to head controls. Once a given pointing device is selected, tuning it to the user’s
strengths and limitations may yield significant performance and comfort benefits.
Windows XP provides two adjustable settings, Gain and Enhance Pointer Precision, that affect
the behavior of the mouse and many mouse alternatives. The gain setting determines how far the
mouse cursor moves on the screen for a given movement of the pointing device. Changes in gain
setting may help to accommodate physical impairments, although there is very little literature in this
area [1]. The Enhance Pointer Precision (EPP) setting enables a complex algorithm controlling the
velocity and acceleration of the mouse cursor. By default, EPP is turned on. We have found no
research on how the EPP setting affects user performance.
The pointing environment also influences user performance. Both intuition and Fitts’ Law tell us
that larger, closer targets can be selected more quickly [2]. The sizes of objects such as menu items
and toolbar buttons can be manipulated through the Windows control panel and from within some
applications. We have found no literature on how such manipulation might enhance pointing
performance for a user with physical impairments.
The long-term goal of this work is to determine ways of selecting and configuring a pointing
device and operating system settings to provide optimal user performance, both initially and over
time. As part of this work, we conducted an exploratory study to get a sense for how the following
five factors affect user performance: (1) Gain setting; (2) EPP setting; (3) Target size; (4) Target
distance; and (5) Input device.
METHODS
Protocol
The design of this study was relatively informal, to allow us to explore the influence of a large
number of factors within a single data collection session. Seventeen subjects with upper extremity
physical impairments performed at least four test runs of target acquisition trials. The target
acquisition task used was the Aim test in the Compass software package [3]. Each trial presented a
single target, and the user moved the mouse cursor inside the target and clicked to select it.
Each test included 32 trials: four targets at each combination of four different sizes (18, 24, 40,
and 100 pixels) and two different distances (50 and 512 pixels). Each test run with a given input
device used one of four combinations of gain (low or default) and EPP (on or off) settings. After
completion of all four combinations, the protocol was repeated with a second input device. Subjects

Measurement Validity of Compass Software
used at least one and as many as four input devices, depending on how many tests they could
complete in an hour’s data collection time. Table 1 shows the order of test conditions.
Test Input Device Gain/EPP Setting Target Size (pixels) Target Distance (pixels)
1 ID1 Default/On 18/24/40/100 50/512
2 ID1 Default/Off 18/24/40/100 50/512
3 ID1 Low/On 18/24/40/100 50/512
4 ID1 Low/Off 18/24/40/100 50/512
5 ID2 Default/On 18/24/40/100 50/512
6 ID2 Default/Off 18/24/40/100 50/512
7 ID2 Low/On 18/24/40/100 50/512
8 ID2 Low/Off 18/24/40/100 50/512
Table 1. Layout of experimental conditions. Note that some subjects used only 1 input device, and
some used more than 2.
The two EPP settings used (on and off) are the only two available in Windows. For gain, there
are 11 different settings between 1 and 20, too large a range to examine completely. We used the
default setting of 10 as a common baseline. An earlier study with a similar subject pool suggested
that we explore lower-than-default settings further [4]. The first four subjects used 4 for the low gain
setting, but this was changed to 6 for the remaining participants.
Subjects were assigned pointing devices from the following list, as long as they were able to use
it: standard mouse, trackball, trackpad, mini joystick, head control, or MouseKeys. The mouse,
trackball, and trackpad were more frequently assigned, as shown in Table 2, since they are more
commonly used.
Input Device N
Mouse 8
Trackpad 8
Trackball 7
MiniJoy 5
Head control 6
MouseKeys 4
Table 2. Number of participants who used each pointing device in the study.
Data Analysis
For each trial, 15 dependent variables were measured. This paper reports on results from two
primary variables: (1) Time – the time required to select the target; and (2) Cursor entries – the
number of times the mouse cursor entered the target.
The trial-by-trial data set included 3136 trials. Data from head control and MouseKeys trials
were not included, since their custom settings interact with the gain and EPP settings in a poorly
understood way. Also excluded were data from devices used for only one gain/EPP setting
combination. Time and cursor entries were each modeled as a function of subject, as a random
effect, and the following fixed effects: input device, gain, EPP, target size, target distance, and the 2-
way interactions of these factors. This mixed ANOVA model allows the determination of main
effects while controlling for the effect of subjects. Effects were considered significant at the p=0.05
level. Post-hoc comparisons used model-estimated means to compare performance at different factor
levels.
An averaged data set was also constructed using the performance measures averaged across all 32
trials in each test run. The resulting data set had 144 observations for each dependent variable,
corresponding to the 144 test runs completed. Averaged data were used to explore the effect of
gain/EPP settings and input device for each subject.

Measurement Validity of Compass Software
RESULTS
Statistical Analysis of Factors
Table 3 summarizes the significance of all effects examined in the mixed ANOVA model. All of the
factors had a significant effect on target acquisition time, while all factors except EPP and target
distance had a significant main effect on cursor entries.
Time Cursor Entries
Factor p Effect Notes p Effect Notes
Device .000 Mouse fastest; Tball 38%, Tpad
59%, MJ 111% slower
.000 Mouse fewest; Tpad 21%, MJ 32%,
Tball 34% more entries
Gain .000 G=6 fastest by 10% .000 Fewest entries for G=4; G=6 15%, G=10
49% more
EPP .000 EPP On fastest by 17% .909 No main effect of EPP
Distance .000 Short fastest by 60% .978 No main effect of distance
Target Size .000 18 slower than 24 by 10% .000 18 more than 24 by 17%
Dev x Gain .000 Gain effect weaker for Mouse and
Tpad
.002 Increase from G=6 to G=10 is less for
Mouse than other devices
Dev x EPP .060 EPP has large effect for Mouse;
almost no effect for Tpad
.378 Very consistent for all devices
Dev x Dist .000 Dist a bit stronger for Tball, Tpad .920 Very consistent for all devices
Dev x Size .000 Size effect weaker for Mouse (24
only 2% faster than 18)
.000 Size effect weaker for Mouse
Gain x EPP .000 EPP effect strongest for G=4 .001 Gain effect stronger when EPP is Off
Gain x Dist .000 Gain stronger for short distance .140 Nothing to note
Gain x Size .001 Size stronger when G=10 .002 Gain stronger for smaller targets
EPP x Dist .013 Dist weaker when EPP is On .270 Nothing to note
EPP x Size .800 Nothing to note .002 EPP only matters for small targets
Dist x Size .037 Size stronger for farther targets .639 Nothing to note
Table 3. Results from the mixed model analysis of 3136 trials. Tball = Trackball; Tpad = Trackpad;
MJ = MiniJoystick.
Effect of Factors on Target Acquisition Time
Across all subjects, Gain=6 was the fastest condition. Looking only at each subject’s best performing
device, time was about 14% faster with G=6 compared to the default of G=10. However, as Figure 1
shows, this significant main effect masks the fact that gain had different effects for different subjects.
EPP On generally provided better performance, with time about 23% faster than EPP Off, when
looking at each subject’s best input device. Additionally, EPP On had a beneficial effect for almost
every subject and gain setting, as shown in Figure 2, in contrast to the more variable effect of gain.
Effect of Gain (at each level of EPP)
(Positive values for default faster than low )
-60.0
-40.0
-20.0
0.0
20.0
40.0
60.0
P25 P1 5 P1 3 P1 6 P28 P2 4 P2 7 P22 P1 8 P14 P2 1 P23 P2 0
Par ticipant
(for P15, P13, P16, P14: Lo = 4; for all else Lo=6)
% Time Difference between
different Gain/EPP conditions
De f v s L o , EPP Of f
De f v s L o , EPP On

Measurement Validity of Compass Software
Figure 1. Effect of Default vs. Low Gain on target acquisition time, at each level of EPP. Time
effects shown are for the best-performing input device for each participant.
Effect of EPP (at each level of Gain)
( Pos it iv e v a lu e s f or EPP Of f f a s ter t han On)
-60.0
-40.0
-20.0
0.0
20.0
40.0
60.0
P25 P15 P13 P16 P28 P2 4 P27 P22 P18 P14 P21 P2 3 P20
Participant
(for P15, P13, P16, P14: Lo = 4; for all else Lo=6)
% Time Difference between
different Gain/EPP conditions
Off vs On, Def Gain
Of f vs On, Lo Gain
Figure 2. Effect of EPP On vs. Off on target acquisition time, at each level of Gain. Time effects
shown are for the best-performing input device for each participant.
Across all subjects, there was a significant main effect of target size, with larger sizes requiring
less acquisition time. Figure 3 shows the average effect of target size on target acquisition time for
three combinations of Gain/EPP settings. When using the default settings, 24-pixel targets were
selected about 7% faster than 18-pixel targets, while 40-pixel targets had a 23% advantage over 24-
pixel targets.
Effect of Target Size acros s Participants
0
5
10
15
20
25
30
35
40
Best Def-On Def-Off
Gain-EPP Condition
Time Effect Size (%)
18v24
24v40
18v40
Figure 3. Effect of target size on target acquisition time, at three combinations of Gain-EPP settings.
Time effects shown are for the best-performing input device and averaged across all participants.
In Figure 4, target acquisition times for each subject’s best-performing input device are compared
to their second-best input device, using Windows default values for Gain/EPP settings. The
differences in times were generally larger than for any other factor studied, averaging 122%, with a
minimum of 33% and a maximum of 218%.

Measurement Validity of Compass Software
Effect of Best vs. Next-Best Input Device
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
160.0
180.0
200.0
P25 P15 P1 3 P28 P24 P27 P22 P1 8 P14 P2 3 P20
Par ticipant
% Time Difference btwn
ID1 & ID2 using Default/On
Figure 4. Effect of Best vs. Next-best input device, when gain and EPP were set to their default
values.
CONCLUSIONS
The results of this exploratory study cannot be definitive, but they strongly suggest the following:
1. The gain setting for a pointing device often makes a difference, but that difference needs to be
assessed for each individual, using a tool like Compass.
2. Enhance Pointer Precision should be On for just about everybody.
3. Increased target size may have a role as a further enhancement to pointing performance.
4. The combination of appropriate settings and increased target size can yield a definite
improvement in target acquisition time and cursor control, supporting the value of an agent that can
help establish the appropriate combination for each unique individual.
5. Choosing the right input device to begin with, however, is at least as important, if not more so.
REFERENCES
1. LoPresti, E.F., Brienza, D.M. (2004). Adaptive Software for Head-Operated Computer Controls.
IEEE Transactions on Rehabilitation Engineering. 12(1):102-111.
2. Balakrishnan R. (2004). "Beating" Fitts' law: Virtual enhancements for pointing facilitation.
International Journal of Human-Computer Studies, 61(6):857-874.
3. Koester H., LoPresti E.F., Simpson R.C. (2005). Toward Goldilocks’ Pointing Device:
Determining a “just right” gain setting for users with physical impairments. ASSETS 2005: 7
th
Int’l
ACM SIGACCESS Conference. Oct 2005.
4. Koester, H.H., LoPresti E.F. (2003). Compass: Software for Computer Skills Assessment.
CSUN's 18
h
Annual Conference "Technology and Persons with Disabilities". March 2003.
ACKNOWLEDGMENTS
This work was funded by the National Institutes of Health, grant #1R43-HD045015, as an SBIR
award to Koester Performance Research and by the NSF (grant #0133395), as a CAREER award to
the University of Pittsburgh. We thank the participants for their time and effort.
Heidi Koester, Ph.D.; Koester Performance Research; Ann Arbor MI 48105; hhk@kpronline.com