Evaluation of flat, angled, and vertical computer mice and their effects on wrist posture, pointing performance, and preference

Abstract

Background:
Modern computer users use the mouse almost three times as much as the keyboard. As exposure rates are high, improving upper extremity posture while using a computer mouse is desirable due to the fact that posture is one risk factor for injury. Previous studies have found posture benefits associated with using alternative mouse designs, but at the cost of performance and preference.

Objective:
To develop new computer mouse shapes, evaluate them versus benchmarks, and determine whether there are differences in wrist posture, pointing performance, and subjective measures.

Method:
Three concept mice were designed and evaluated relative to two existing benchmark models: a traditional flat mouse, and an alternative upright mouse. Using a repeated measures design, twelve subjects performed a standardized point-and-click task with each mouse. Pointing performance and wrist posture was measured, along with perceived fatigue ratings and subjective preferences pre and post use.

Results:
All of the concept mice were shown to reduce forearm pronation relative to the traditional flat mouse. There were no differences in pointing performance between the traditional flat mouse and the concept mice. In contrast, the fully vertical mouse reduced pronation but had the poorest pointing performance. Perceived fatigue and subjective preferences were consistently better for one concept mouse.

Conclusion:
Increasing mouse height and angling the mouse topcase can improve wrist posture without negatively affecting performance.

Full text

Work 52 (2015) 245–253
DOI:10.3233/WOR-152167
IOS Press
245
Evaluation of flat, angled, and vertical
computer mice and their effects on wrist
posture, pointing performance, and preference
Dan Odella,∗and Peter Johnsonb
aSynaptics, Inc., San Jose, CA, USA
bSchool of Public Health, University of Washington, Seattle, WA, USA
Received 3 March 2014
Accepted 7 May 2014
Abstract.
BACKGROUND: Modern computer users use the mouse almost three times as much as the keyboard. As exposure rates are high,
improving upper extremity posture while using a computer mouse is desirable due to the fact that posture is one risk factor for
injury. Previous studies have found posture benefits associated with using alternative mouse designs, but at the cost of performance
and preference.
OBJECTIVE: To develop new computer mouse shapes, evaluate them versus benchmarks, and determine whether there are
differences in wrist posture, pointing performance, and subjective measures.
METHOD: Three concept mice were designed and evaluated relative to two existing benchmark models: a traditional flat mouse,
and an alternative upright mouse. Using a repeated measures design, twelve subjects performed a standardized point-and-click
task with each mouse. Pointing performance and wrist posture was measured, along with perceived fatigue ratings and subjective
preferences pre and post use.
RESULTS: All of the concept mice were shown to reduce forearm pronation relative to the traditional flat mouse. There were no
differences in pointing performance between the traditional flat mouse and the concept mice. In contrast, the fully vertical mouse
reduced pronation but had the poorest pointing performance. Perceived fatigue and subjective preferences were consistently better
for one concept mouse.
CONCLUSION: Increasing mouse height and angling the mouse topcase can improve wrist posture without negatively affecting
performance.
Keywords: Human-computer interaction, design, ergonomics
1. Introduction
The trend in computer input has shifted away from
keyboard input to pointing device input as user inter-
faces have evolved from text-based to graphical user
∗Address for correspondence: Dan Odell, Synaptics, Inc., San
Jose, CA, USA. E-mail: drdanodell@yahoo.com.
interfaces. Today, average active time on the mouse
exceeds average active time on the keyboard by almost
3 to 1 [4, 12, 22]. The number of hours per day working
on a computer has been identified as one of the key risk
factors for Musculoskeletal Symptoms (MSS) [18], and
research has linked intensive mouse use to MSS [6, 13].
High incidence of computer related MSS of the upper
extremity has been reported, for example in a college
1051-9815/15/$35.00 © 2015 – IOS Press and the authors. All rights reserved

246 D. Odell and P. Johnson / Evaluation of flat, angled, and vertical computer mice
student population [28]. The higher durational exposure
of the mouse relative to the keyboard, and the observed
link between mouse use and MSS [8] emphasizes that
mouse design should be a key focus when seeking to
improve comfort at the computer.
Several commercially available mice have been
designed with the intent of improving user posture
during mousing by making them more vertical. Early
efforts focused on providing sculpted, non-symmetrical
mice that were designed to improve comfort and wrist
posture [1, 10]. Some of these mice designs have been
shown to provide better posture during mousing [10,
11, 25]. Other studies have shown benefits in reduced
muscle load for more vertical mice [11, 25, 27]. These
benefits can translate into longer term comfort benefits.
For example, testing of one vertical mouse demon-
strated that the upright pronation-reducing design was
successful at reducing subjective pain over a short
period of use [1]. Unfortunately, of the ‘neutral posture’
mice reported in the literature, all have shown reduced
pointing performance [1, 10, 25, 26]. Subjective pref-
erence measures have also generally been low for
these inclinced mice [1, 26]. Reduced pointing perfor-
mance and preference can serve as significant barriers
to widespread adoption of a pointing device despite
other ergonomic benefits that alternative designs may
provide.
In this study, three concept mice with angled topcases
were developed and evaluated against two benchmark
mice: a traditional flat mouse, and a commercially avail-
able vertical mouse. The intent behind the designs of the
concept mice was to reduce non-neutral upper extrem-
ity postures that may be associated with increased
risk for musculoskeletal discomfort, while maintaining
pointing performance. Specifically, the concept mice
were designed to reduce: 1) forearm pronation, 2) wrist
extension, 3) ulnar deviation, 4) extended finger pos-
tures, and 5) change the location of the contact area
between wrist and the desktop.
These design criteria were informed by existing
research on carpal tunnel pressure and wrist angle
[5, 15, 19, 21]. Hydrostatic fluid pressure within the
carpal tunnel has been one means to show the potential
risks associated with various forearm, wrist and fin-
ger postures. Non-neutral wrist and forearm postures
can increase the hydrostatic fluid pressure within the
carpal tunnel and the increased pressure can impair
nerve function and can lead to long-term damage [19,
21]. For example, Rempel et al. [19] determined that
∼45◦of forearm pronation and ∼45◦of finger flex-
ion were postures both associated with reduced carpal
tunnel pressure (CTP). These postures correspond with
the targeted forearm and finger postures outlined in the
mouse design criteria cited above. Similar to prona-
tion and finger extension, elevated fluid pressure in
the carpal tunnel has also been associated with wrist
extension and ulnar deviation [15]. This prompted the
design criteria to include promoting more neutral wrist
postures.
Elevated CTP is also associated with external con-
tact pressure on the palmar surfaces of the hand and
wrist. Cobb et al. [5] showed that certain areas of the
hand and wrist, especially the the base of the palm over
the flexor retinaculum, were sensitive to external con-
tact pressure. Contact pressure in those regions elevated
CTP. Therefore, another potential benefit of more ver-
tical concept mouse design may be to reduce contact
pressure in those sensitive areas by shifting the contact
area off of the base of the palm and to the ulnar aspect
of the hand.
After the three concept mice were developed, it was
important to evaluate them to determine how effective
they were in meeting the design criteria, and therefore if
any of them should be produced commercially. The goal
of this study was to make this evaluation of the concept
mice and determine whether there were benefits in wrist
posture, pointing performance, perceived fatigue and
preference relative to the flat and vertical benchmark
mice.
2. Methods
2.1. Subjects
Twelve experienced computer mouse users (6 male,
6 female) who all self-reported to use the computer
at least 10 hours a week were recruited to participate
in this study. The mean age of the subjects was 32.7
years (range 20–52) and all subjects used the mouse
with their right hand. Experimental procedures were
approved by the University of Washington Human Sub-
jects Committee and all subjects gave their informed
consent.
Using the methods outlined in the ADULTDATA
anthropometric handbook [30], hand length and hand
breadth were measured from all subjects. Hand length
was measured from the distal wrist crease to the tip
of the middle finger and and hand breadth was mea-
sured from the medial side of the palm just below the
little finger to the lateral side of the palm just below
the index finger. The mean hand legnth of the subjects

D. Odell and P. Johnson / Evaluation of flat, angled, and vertical computer mice 247
was 18.1 cm (range 16.6 to 20.1 cm) and the mean hand
breadth was 8.8 cm (range 7.6 to 9.6 cm). Using US
anthropometric data as a reference [30], hand lengths
spanned from the 18%tile female to the 86%tile male and
hand breadths spanned from the 37%tile female to the
96%tile, so a wide hand size range was represented.
2.2. Mouse models evaluated
Theexperimentwasarepeated measuresdesignwhere
subjects performed a series of standardized point-and-
click tasks with five different mice (Fig. 1). The five
models consisted of two benchmarks (Models E and F)
and three concept mice (Models H, I and J). The two
benchmarks were cast models of the Evoluent™Vertical
mouse (Model E), and the more traditional right-handed
mouse, the Microsoft IntelliMouse Explorer for Blue-
tooth (Model F). All tested mice were cast models made
out of BJB TC182 castable foam. The cast models were
made functional by harvesting wireless optical track-
ing sensors from commercially manufactured mice and
inserting them into the cast mouse bodies. All models
where made of the same materials, finishes, and track-
ing engines to minimize confounding factors. The left
buttons on the mouse models were made operational by
installing and connecting a tactile switch under the flex-
ible button surface.
The bodies of the three concept mice were similar in
design. They were all approximately 61 mm wide and
60 mm tall at the highest point on the left mouse but-
ton and the topcases were angled roughly 20 degrees
downward relative to a horizontal plane. The topcase
slope and extra height was designed to help reduce
forearm pronation. The concept mice all had thumb
grooves roughly 25 mm off of the table surface with
Model E Model F Model H Model I Model J
Fig. 1. Top and rear view images of the mouse models tested. Model
E (benchmark vertical mouse), Model F (benchmark flat mouse), and
the three concept mouse designs - Model H, Model I and Model J.
the thumb groove for Model ‘I’ being slightly lower
and less distinct.
The primary differences between the concept mice
were the overall length and the design of back of the
mouse where the thumb and palm of the hand rested.
Model H was the shortest (105 mm), ball-like in shape
and afforded the most flexibility in grip styles. Model
I was the longest (130 mm) and had a large flange sup-
porting the base of the thumb. Model J was a hybrid of
Model H and I, intermediate in length (120 mm) with a
smaller flange for the base of the thumb.
2.3. Experimental procedures
The test workstation was set up according to ANSI
HFS 100 standards [2] to match the subject’s stature.
Subjects were allowed to make slight adjustments in
table and chair height for comfort. Mouse performance
was measured while subjects performed a series of
standardized point-and-click tasks, as specified in the
international standards for evaluating pointing device
performance [29]. These tasks consisted of a series
of omni-directional pointing tasks which consisted
of alternately clicking on 18 evenly spaced round
targets arranged in a circle (Fig. 2). As illustrated in
Fig. 2, to perform the tasks, subjects would move the
cursor with the mouse and click on the first active,
black-highlighted target, the target would disappear,
and then the target on the diametrically opposite side
of the circle would become active, and the subject
had to move the mouse to acquire this target and
then click on it. This sequence continued until all 18
targets had been acquired. The series of tasks consisted
of performing: 1) six large pointing tasks requiring
gross movements with a center-to-center inter-target
distance of 142 mm and target width of 12 mm - this
target size was similar to the size of folders and icons
on a computer desktop; 2) six medium pointing tasks
requiring intermediate movements with a center-to-
center inter-target distance of 71 mm and target width
of 6 mm - half the size of the large targets and twice the
size of the small targets; and 3) three small pointing
tasks requiring fine movements with a center-to-center
inter-target distance of 28 mm and target width of
2 mm - these small targets approximated the size of
individual characters. The small, medium and large
tasks all had the same index of difficulty according to
Fitt’s Law [7].
While performing the point-and-click tasks, subjects
were instructed to move the mice as fast as possible
while maintaining a balance between speed and accu-

248 D. Odell and P. Johnson / Evaluation of flat, angled, and vertical computer mice
Fig. 2. Large, medium and small omni-directional pointing tasks. The numbers in large pointing task on the left shows the sequence of the pointing
movements.
racy. Using a different traditional mouse than the one
tested in the study, subjects were allowed to practice
the tasks (typically 2 to 4 minutes) until they were
comfortable with the how to complete the various tasks.
Mouse model and task order were randomized and no
instructions were provided as to how to grip or use the
mice.
2.4. Forearm and wrist posture
Right hand wrist angles were measured using an
electrogoniometer (Model XM-65; Biometrics; Gwent,
UK), forearm pronation/supination was measured with
an inclinometer (FAS-G; Microstrain, Inc.; Williston,
VT) mounted to the distal end-block of the elec-
trogoniometer, and all measures were collected and
stored at 100 Hz on a portable data logger (Muscle
Tester ME6000; Mega Electronics; Kuopio, Finland).
As prescribed by the American Academy of Orthopedic
surgeons [9], the neutral flexion/extension (F/E) posi-
tion of the wrist was defined at the position where the
horizontal plane formed by the back of the hand was
in line with the plane formed by the back of the fore-
arm. The neutral radial/ulnar (R/U) position was defined
as the position where the third metacarpal was in line
withthelongaxis of theforearm.Thiscalibration posture
accounted for and minimized any offset errors associ-
ated with forearm pronation [14]. With the inclinome-
ter, subjects alternated between a hand-shake position
with their hands perpendicular to the worksurface (0º)
and a fully pronated position with their palms parallel to
the worksurface (90º). Pronation/supination (P/S) mea-
surements were relative to the neutral position (0º).
2.5. Perceived fatigue and preference
Perceived fatigue levels while using each mouse
model were measured in the right hand, forearm, shoul-
der and neck using visual analog BORG CR-10 scales
[3] administered before and after using each mouse.
Before using the mice, subjects were asked to give
their preliminary preferences based on visual appear-
ance and touch. Then, after completing the series
of omni-directional point-and-click tasks with each
mouse, a series of eight mouse preference questions
were answered. These questions employed a 7-point
Likert scale ranging from 1- Strongly Disagree to 7-
Strongly Agree. Finally, after using all mouse models,
the participants ranked their final mouse preferences.
2.6. Data and statistical analysis
With the inclinometer and goniometry data, mean
postural values were calculated for each subject
and then group mean postural values were calcu-
lated for Pronation/Supination, Flexion/Extension and
Radial/Ulnar deviation. A program written in Labview
calculated movement times between targets in seconds
and the number of missed targets per trial. To reduce any
additive effects of time, Borg scale rating data were nor-
malized to the initial measurement so that comparisons
between mice could be made. Normalization involved
subtracting the Borg scale rating at the initial measure-
ment from end measurements, thus, the change in Borg
scale ratings relative to the initial measurement were
being compared. All posture, performance, perceived
fatigue and preference data were then tabulated and
analyzed using repeated measures analysis of variance

D. Odell and P. Johnson / Evaluation of flat, angled, and vertical computer mice 249
(RANOVA) methods using the statistical program JMP
(Version 7.0; SAS Institute; Cary, NC). Data are pre-
sented as means and standard error and differences were
considered to be significant when p-values were less
than 0.05.
3. Results
3.1. Posture
There were postural differences between the five
mouse models tested, and average postures during the
performance of the point-and-click tasks are summa-
rized in Table 1. Model E, the vertical benchmark
mouse, was operated with the greatest amount of wrist
extension and the least amount of hand/forearm prona-
tion. Model F, the flat benchmark mouse, required the
greatest amount of ulnar deviation and hand/forearm
pronation. Concept Model I was operated with the least
amount of extension and Concept Model J the least
amount of radial/ulnar deviation. On average, compared
to the flat benchmark mouse (Model F), the concept
mouse designs (Models H, I and J) reduced prona-
tion by 13.1◦±1.0◦, ulnar deviation by 6.9◦±1.5◦and
increased extension by 1.6◦±1.8◦.
3.2. Performance
As shown in Table 2, during the performance
of the omni-directional point-and-click tasks, there
were movement time differences between the five
mouse models tested (p= 0.02). Subjects consistently
performed tasks the slowest with vertical benchmark
mouse (Model E). There were only small differences in
movement times between the concept mouse designs
(Models H, I and J) and the flat benchmark mouse
(Model F). Despite the tasks having the same index of
difficulty, there were movement time differences based
on task size (p< 0.01) but there was no difference by
task size interaction (p= 0.53). Depending on the mouse
model, the smaller tasks took 10 to 20% longer to com-
plete. There was also an order (p< 0.01) and trial effect
(p< 0.01). When cycling through the five mouse mod-
els, the subject’s movement times were progressively
faster with each subsequent mouse. With respect to tri-
als, movement times decreased from the first to second
trial (p< 0.01) but then were relatively stable for the
remaining four trials.
Table 3 shows the accuracy of the mice in the form
of number of missed targets. No statistically significant
differences were found for missed targets.
3.3. Perceived fatigue and subjective preference
On average, the time to complete the suite of tasks
with each mouse was 4 minutes. Table 4 shows the
changes in perceived fatigue by body region after the
4 minutes of mouse use. In general, changes in per-
ceived fatigue were small and mouse Model H had the
lowest perceived fatigue levels for the concept mouse
designs (Models H, I and J). Mouse Model H scored
favorably in three out of four body locations while the
vertical benchmark mouse (Model E) was rated as the
least fatiguing in the wrist and shoulder regions.
Table 1
Mean (standard error) posture with each mouse. Means with different superscripts are significantly different [N= 12].
Lower values are generally preferred for more neutral postures
Benchmark Mice Concept Mice p-value
Vertical Flat
Model E Model F Model H Model I Model J
Extension 41.2◦a(±3.0◦) 36.1◦a,b(±3.4◦) 40.7◦a(±3.0◦) 35.0◦b(±3.3◦) 37.4◦a,b(±3.5◦)p= 0.004
Ulnar(+)/Radial(-) Deviation 3.9◦a,b(±3.0◦) 8.1◦b(±3.9◦) 1.6◦a(±3.7º) 2.1◦a(±3.2◦) –0.2◦a(±3.1◦)p< 0.0001
Pronation 40.4◦a(±2.1◦) 70.5◦b(±2.4◦) 61.3◦c(±2.5◦) 56.4◦d(±2.4◦) 54.5◦d(±2.1◦)p< 0.0001
Table 2
Mean (standard error) movement times in seconds by mouse model while performing the large, medium and small omni-directional
point-and-click tasks. Means with different superscripts are significantly different [N= 12]. Lower values are preferred for faster task completion
Vertical Flat Concept Mice p-value
Model E Model F Model H Model I Model J
Large Task 1.04a(±0.07) 0.89b(±0.07) 0.91a,b(±0.07) 0.92a,b(±0.05) 0.87 b(±0.06) p= 0.02
Medium Task 1.03a(±0.06) 0.88b(±0.07) 0.88b(±0.06) 0.91a,b(±0.06) 0.90 a,b(±0.07) p= 0.02
Small Task 1.21 (±0.12) 1.04 (±0.11) 0.99 (±0.07) 1.00 (±0.08) 1.04 (±0.06) p= 0.09

250 D. Odell and P. Johnson / Evaluation of flat, angled, and vertical computer mice
Table 3
Mean (standard error) number of missed targets (out of 18 targets) by mouse model while performing the large, medium and small
omni-directional point-and-click tasks. [N =12]. Lower values are preferred for higher accuracy
Vertical Flat Concept Mice p-value
Model E Model F Model H Model I Model J
Large Targets 1.0 (±0.1) 0.7 (±0.1) 1.1 (±0.2) 1.1 (±0.2) 1.0 (±0.2) p= 0.31
Medium Targets 1.5 (±0.2) 1.3 (±0.2) 1.7 (±0.2) 1.6 (±0.2) 1.4 (±0.3) p= 0.51
Small Targets 3.1 (±0.4) 2.8 (±0.4) 3.3 (±0.3) 2.7 (±0.3) 3.6 (±0.4) p= 0.21
Table 4
Mean (standard error) change in Borg CR-10 scale ratings after using each mouse by body region. The larger the number the more fatiguing
mouse use was perceived to be [N= 12]
Vertical Flat Concept Mice p-value
Model E Model F Model H Model I Model J
Hand 0.64 (±0.25) 0.46 (±0.38) 0.00 (±0.22) 0.30 (±0.13) 0.75 (±0.26) p= 0.11
Wrist 1.00 (±0.50) 0.28 (±0.31) 0.21 (±0.34) 0.96 (±0.36) 0.53 (±0.26) p= 0.32
Forearm 0.74 (±0.33) 0.58 (±0.29) 0.15 (±0.16) 0.79 (±0.35) 0.77 (±0.27) p= 0.36
Shoulder 0.90 (±0.28) 0.01 (±0.28) 0.08 (±0.17) 0.62 (±0.44) 0.57 (±0.26) p= 0.15
Prior to the pointing tasks, the mice were ranked to
determine preferences based on feel and appearance and
the flatbenchmark mouse (Model F) was the most pre-
ferred mouse (Table 5). After subjects had performed
the pointing tasks with all of the mice, the models
were ranked again. After use, mouse Model F was still
the most preferred mouse but mouse Model H had the
greatest change in preference, starting as the 4th most
preferred mouse and ending as the 2nd most preferred
mouse. The vertical benchmark mouse (Model E) was
the least preferred both pre and post use.
Table 6 shows that there were significant differences
in preference across the mice tested. The flat bench-
mark mouse (Model F) received the most favorable
response in all 8 questions while concept mouse Model
H had the second most favorable response in 7 out of
8 questions and had the highest ratings of the concept
mouse models. The vertical benchmark mouse (Model
E) received the least favorable responses in all eight
questions.
4. Discussion
4.1. Posture
The concept mice (Models H, I and J) promoted more
neutral hand/forearm pronation and radial/ulnar devia-
tion postures compared to the flat benchmark mouse
(Model F). Mouse Models F, I, and J all had similar
wrist extension values (average 36◦), whereas Models
H and E had slightly higher values for wrist extension
(average 41◦). The vertical benchmark mouse (Model
E) had the greatest effect on reducing pronation.
As mentioned in the methods section, no instructions
were given on how to use or hold any of the concept
mice. These mice were designed so the front of the
mouse would point slightly inward towards the center-
line of the user’s body if held with no inward/outward
rotation of the arm and the wrist in a neutral radial/ulnar
posture. Most subjects were observed to not hold the
mouse with the front pointing in as designed, but rather
with the front of the mouse parallel to the right edge of
the keyboard, a posture which required more ulnar devi-
ation and extension than intended orientation/posture.
So, it seems that ulnar deviation and wrist extension
may be further reduced with some instruction on how to
position and hold the mouse, and this has been verified
in other work [11].
4.2. Performance
All of the mice used identical optical tracking
engines, so the observed performance differences are
attributable to the differences in shape/design of the
mice. The flat benchmark mouse (Model F) and the
concept mice (Models H, I, J) had similar pointing
times during the performance of the point-and-click
task. However, the vertical benchmark mouse (Model
E) was significantly slower than the other four mice.
The fact that the concept mice performed as well
as the flat benchmark mouse in pointing performance
was encouraging from a design standpoint. The lack

D. Odell and P. Johnson / Evaluation of flat, angled, and vertical computer mice 251
Table 5
Preliminary mouse model preferences before use and final mouse model preferences post-use. Bottom row is the mean rank for each mouse
model [N =12]. Lower values represent stronger preference
Preliminary Final
Vertical Flat Concept Mice Vertical Flat Concept Mice
Model E Model F Model H Model I Model J Model E Model F Model H Model I Model J
1st 17211 1st 18111
2nd 0 4 2 3 3 2nd 0 0 7 3 2
3rd 0 0 2 3 7 3rd 2 4 2 1 3
4th 1 1 6 4 0 4th 1 0 1 5 5
5th 10 0 0 1 1 5th 8 0 1 2 1
Mean Rank 4.6 1.6 3.0 3.1 2.7 Mean Rank 4.2 1.7 2.5 3.3 3.2
Table 6
Mean (standard error) responses to mouse preference questions using a 7-point (1 =Strongly Disagree, 7 = Strongly Agree) Likert scale. Scores
within each row with the different superscripts are significantly different [N= 12]. Higher values represent stronger preference
Vertical Flat Concept Mice
Model E Model F Model H Model I Model J p-value
1. This mouse was easy to use 3.3a(±0.54) 5.7b(±0.31) 5.2b(±0.27) 4.4a,b(±0.50) 4.4a,b(±0.47) p= 0.004
2. This mouse feels comfortable 3.0a(±0.49) 5.5b(±0.31) 4.8b,c(±0.35) 3.6a,c(±0.48) 3.7a,c(±0.53) p= 0.002
3. The overall shape of this mouse looks appealing. 3.3a(±0.49) 5.4b(±0.34) 4.3a,b(±0.41) 4.8b(±0.32) 4.3a,b(±0.46) p= 0.004
4. It was hard to make errors using this mouse 3.0a(±0.43) 5.2b(±0.47) 3.9a,b(±0.47) 3.5a,b(±0.52) 3.6a(±0.36) p= 0.006
5. I like using this mouse. 2.8a(±0.55) 5.3b(±0.38) 4.3a,b(±0.37) 3.8a(±0.45) 3.8a,b(±0.54) p= 0.001
6. I can quickly complete tasks using this mouse. 3.2a(±0.47) 5.4b(±0.42) 4.2a,b(±0.44) 3.8a(±0.46) 4.0a,b(±0.46) p= 0.003
7. I quickly adjusted to using this mouse. 4.1a(±0.54) 6.0b(±0.41) 5.1a,b(±0.36) 4.4a,b(±0.45) 4.4a,b(±0.47) p=0.02
8. I would buy this mouse. 2.7a(±0.61) 5.2b(±0.39) 3.4a,b(±0.56) 3.0a(±0.52) 2.9a(±0.42) p= 0.005
Average 3.2 5.5 4.4 3.9 3.9
of the performance decrement with the concept mice
indicated that it was possible to alter wrist and forearm
posture during mouse use without the negative impact
on performance that has been seen in previous studies
with other alternative mouse [1, 10].
Model E’s relatively slower pointing speed is likely
due to its vertical orientation. The difference was that
mouse Model E primarily required flexion and exten-
sion of the wrist to move the mouse side-to-side. In
contrast, wrist deviation was used to move the tradi-
tional flat mouse side-to-side, and a combination of
wrist deviation and extension was used to move the
concept mice side-to-side. It seems that either this wrist
flexion/extension motion is inherently slower, or some-
thing else (such as the larger size of Model E) accounts
for the reduced pointing speed.
4.3. Subjective preferences
Subjective measures showed preference for the tradi-
tionally designed flat mouse (Model F) both pre and post
mouse use. This is not a surprising finding as this mouse
was chosen as the benchmark due to its high comfort
ratings amongst flat mice, and was most similar to the
traditional mice which participants were most familiar
with. In contrast, the other mouse designs (Models E,
H, I and J) were unfamiliar. Model H ended up as the
second most preferred overall, and the most preferred
of the concept mice. Model H had the greatest change
in preference pre and post use going from the 4th to
the 2nd most preferred mouse. Subjects ranked mouse
Model H as the most preferred concept mouse because,
due to the shorter length and ball shape, this mouse was
the best fit for their hand and was more easily controlled
than the other concept mice. Subjects felt Model I was
too big and Models I and J did not fit their hands as well
as Model H. The vertical benchmark mouse, Model E,
was disliked overall. It was seen as too big, didn’t fit the
hand well, required the most drastic change in control
strategy and was noticably slower and more difficult to
control than the other mice.
4.4. Final mouse selection
In the end, all three concept mice had relatively
similar performance in posture, pointing speed, and

252 D. Odell and P. Johnson / Evaluation of flat, angled, and vertical computer mice
Fig. 3. The commercially manufactured mouse based on concept
mouse Model H.
learning, but one model, mouse Model H, differenti-
ated itself due to more favorable perceived fatigue and
preference ratings. Mouse Model H was also observed
to best accommodate a variety of grip styles. These ben-
efits led to mouse Model H being selected as the final
model for commercial development as the Microsoft
Natural®Mouse 6000/7000 (Fig. 3).
Previous alternative mouse designs have also demon-
strated improvements in posture and discomfort.
However, these alternative mouse designs have often
resulted in reduced pointing performance and/or low
marks for subjective preference [1, 10, 25, 26]. The final
concept mouse model selected from this study reduced
these pitfalls by providing posture and comfort benefits
along with a pointing performance and learning curves
comparable to the commercially sold mice similar to
the flat benchmark mouse (Model F).
It is anticipated that the postural improvements
promoted by the final production mouse will fol-
low the trends of the postural and musculoskeletal
health improvements associated with split keyboard
designs. With split keyboard designs, early research
demonstrated improvements in posture and subjective
comfort [17, 20, 24]. These improvements were later
linked with long term comfort benefits and a reduction
in the risk of musculoskeletal disorders and symptoms
[23].
4.5. Study limitations
The main limitation of this study is the short expo-
sure time that each participant had with each mouse –
roughly 4 minutes with each mouse. This means that
only initial performance was captured from users who
were novices with all of the mice, except for the tradi-
tional mouse, Mouse F, was the most familiar. Another
potential limitation, is that only a single Index of Diffi-
culty [16] was used in the pointing performance tasks.
In the future, testing an array of indexes of difficulty
could be desirable.
4.6. Future study directions
The concept mouse design selected for manufacture
and sale was designed to address a number of iden-
tified postural risk factors which may be associated
with intensive mouse use. However, to what degree
this new design will benefit mouse users in the long-
term is a research question that should be answered
by future studies. To best answer the question of long-
term effects, a longitudinal, randomized control trial
that evaluates musculoskeletal symptoms and disorders
with mouse use over an extended period of time would
be desirable. A separate line of research could evaluate
the other biomechanical risk factors addressed by the
design. In particular, muscle loads, carpal tunnel pres-
sure and the contact area/pressure over the hand and
wrist associated with operating the mouse may be worth
evaluating with the new mouse design. It would be
interesting to see if further design modifications might
address the issue wrist extension as described in 4.1.
5. Conclusions
The concept mice in this study offered less pronated
and deviated postures relative to the flat bench-
mark mouse without negatively affecting pointing
performance. These postural benefits were created by
increasing the mouse height and angling the topcases
of the concept mice. In contrast, the vertical bench-
mark mouse was successful in reducing wrist deviation
and forearm pronation, but pointing performance was
adversely affected and preference ratings were lower.
Therefore, some pronation reduction has been shown to
be beneficial. But, it is possible to go too far and nega-
tively affect performance and preference. The concept
mouse with the best overall performance in this study
was selected for commercial production.

D. Odell and P. Johnson / Evaluation of flat, angled, and vertical computer mice 253
Acknowledgments
Many thanks to Monique Chatterjee for her excellent
design work and collaboration. Thanks to Dick Comp-
ton, Steve Buchanon, Vince Jesus and the Microsoft
modelshopfor generating thefunctionalprototypes used
in this study, and to Jim Ploger, Han Chen and Janet
Blackstone at the University of Washington who ran the
experiments. This study was funded by the Microsoft
Corporation and support was also provided by the Wash-
ington State Medical Aid and Accident Fund.
References
[1] Aaras A, Horgen G, Ro O. Can a more neutral position of
the forearm when operating a computer mouse reduce the
pain level for visual display unit operations? A prospec-
tive epidemiologic intervention study. International Journal of
Human-Computer Interaction 1999;11:79-94.
[2] American National Standards Institute. American National
Standards for Human Factors Engineering of Computer Work-
stations. Santa Monica (CA): Human Factors Society 1988;
No. 100-1988.
[3] Borg E, Borg G. A comparison of AME and CR100 for
scaling perceived exertion. Acta Psychologica (Amsterdam)
2002;109(2):157-75.
[4] Chang C, Menendez C, Amick B, Johnson P, Katz J, Robert-
son M, Dennerlein J. Increased computer usage is associated
with the reporting of MSD symptoms in 27 university students.
Proceedings of the 16th Triennial Congress of the International
Ergonomics Association 2006;797.
[5] Cobb T, An K, Cooney W. Externally applied forces to the
palm increase carpal tunnel pressure. Journal of Hand Surgury
1995;20:181-5.
[6] Cook C, Burgess-Limerick R, Chang S. The prevalence of neck
and upper extremity musculoskeletal symptoms in computer
mouse users. International Journal of Industrial Ergonomics
2000;26:347-56.
[7] Fitts PM. The information capacity of the human motor system
in controlling the amplitude of movement. Journal of Experi-
mental Psychology 1954;47:381-92.
[8] Gerr F, Marcus M, Ensor C, Kleinbaum D, Cohen S, Edwards
A, Gentry E, Ortiz D, Monteilh C. A prospective study
of computer users: I. Study design and incidence of mus-
culoskeletal symptoms and disorders. American Journal of
Industrial Medicine 2002;41:221-35.
[9] Greene WB, Heckman JD. The clinical measurement of joint
motion. Rosemont, IL: American Academy of Orthopaedic
Surgeons 1994.
[10] Gustafsson E, Hagberg M. Computer mouse use in two
different hand positions: Exposure, comfort, exertion and pro-
ductivity. Applied Ergonomics 2003;34:107-13
[11] Houwink A, Oude Hengel K, Odell D, Dennerlein J. Pro-
viding training enhances the biomechanical improvements
of an alternative computer mouse design. Human Factors
2009;51(1):46-55.
[12] Jacobs K, Johnson P, Dennerlein J, Peterson D, Kaufman J,
Gold J, Williams S, Richmond N, Karban S, Firn E, Ansong
E, Hudak S, Tung K, Hall V, Pencina K, Pencina M. Univer-
sity students’ notebook computer use. Applied Ergonomics
2008;40:404-9.
[13] Jensen C, Borg V, Finsen L, Hansen K, Juul-Kristensen B,
Christensen H. Job demands, muscle activity and muscu-
loskeletal symptoms in relation to work with the computer
mouse. Scand Journal Work Environ Health 1998;24:418-24.
[14] Johnson P, Jonsson P, Hagberg M. Effects of pronation and
supination on wrist goniometer measurement accuracy.Journal
of Electromyography and Kinesiology 2002;12(5):413-20.
[15] Keir P, Bach J, Rempel D. Effects of finger posture on carpal
tunnel pressure during wrist motion. Journal of Hand Surgery
1998;23A:1004-9.
[16] MacKenzie IS. Fitts’ law as a research and design tool in
human-computer interaction. Human-Computer Interaction
1992;7:91-139.
[17] Marklin R, Simoneau G. Design features of alternative com-
puter keyboards: A review of experimental data. Journal of
Orthopeadic Sports and Physical Therapy 2004;34(10):638-
49.
[18] Nakazawa T, Okubo Y, Suwazono Y, Kobayashi E, Komine
S, Kato N, Nogawa, K. Association between duration of
daily VDT use and subjective symptoms. American Journal
of Industrial Medicine 2002;42:421-6.
[19] Rempel D, Bach J, Gordon L, Tal R. Effects of forearm prona-
tion/supination on carpal tunnel pressure. Journal of Hand
Surgery 1998;23:38-42.
[20] Rempel D. The split keyboard: An ergonomics success story.
Human Factors 2008;50(3):385-92.
[21] Rempel D, Dahlin L, Lundborg G. Pathophysiology of nerve
compression syndromes: Response of peripheral nerves to
loading. Journal of Bone and Joint Surgury (American)
1999;81:1600-10.
[22] Richter J, Slijper J, Over E, Frens M. Computer work dura-
tion and its dependence on the used pause definition. Applied
Ergonomics 2008;39:772-8.
[23] Tittiranonda P, Rempel D, Armstrong T, Burastero S. Effect
of four computer keyboards in computer users with upper.
Extremity musculoskeletal disorders. American Journal of
Industrial Medicine 1999;35:647-61.
[24] Zecevic A, Miller DI, Harburn M. An evaluation of
the ergonomics of three computer keyboards. Ergonomics
2000;43(1):55-72.
[25] Quemelo P, Vieira E. Biomechanics and performance when
using a standard and a vertical computer mouse. Ergonomics
2014;56(8): 1336-44.
[26] Jung K. Effects of slanted ergonomic mice on task performance
and subjective responses. Human Factors 2014;45(3):450-5.
[27] Chen H, Chun-Tong L. The effect on forearm and shoulder
muscle activity in using different slanted computer mice. Clin-
ical Biomechanics 2007;22:518-23
[28] Hupert N, Amick B, Fossel A, Coley C, Robertson M, Katz
J. Upper extremity musculoskeletal symptoms and functional
impairment related to computer use among college students.
Work: A Journal of Prevention Assessment and Rehabilitation
2004;23:85-93.
[29] International Standards Organization (ISO) Technical Specifi-
cation. 9241-411:2012
[30] Peebles L, Norris BJ. ADULTDATA: The handbook of adult
anthropometric and strength measurements. London: Depart-
ment of Trade and Industry; 1998.

CopyrightofWorkisthepropertyofIOSPressanditscontentmaynotbecopiedoremailed
tomultiplesitesorpostedtoalistservwithoutthecopyrightholder'sexpresswritten
permission.However,usersmayprint,download,oremailarticlesforindividualuse.