H F-XC A N GE

H F-XC A N GE

N bu

y .c

to k lic .d o

Senior-Friendly Input Devices: Is the Pen Mightier than the Mouse?

m

w

o

m o

c u-tr a c k

w

w

.d o

http://www.psy.fsu.edu/~charness/a… w

w

w

C

23/2/2009

C

lic

k

to

bu

y

N

O W !

PD

O W !

PD

c u-tr a c k

(Presented at the 103rd Annual Convention of the American Psychological Association Meeting in New York, New York, August 11, 1995) Neil Charness, Florida State University Elizabeth A. Bosman, University of Toronto Robert G. Elliott, Florida State University Kelley and Charness (1995) point out that computer technology may inadvertently pose human factors problems for older adults, particularly with respect to input devices. There have been remarkably few comparative studies of younger versus older users of computer systems, and virtually no comparisons of the efficacy of different input devices. Analysis of videotape records showed that novice older adults learning a graphic user interface (GUI) word processing package commit far more mouse-related errors than younger adults (Charness, Kelley & Bosman, in preparation). In our first study we compared the use of a mouse versus cursor keys for ease of target item selection by younger and older novice users. This study revealed that although the mouse was superior to cursor keys, older adults had particular difficulty acquiring smaller targets (age x target size interaction) and the gap between young and older adults grew with the use of a mouse (age x device interaction), suggesting that part of the problem was learning to control fine positioning aspects of movement (undershoot and overshoot, as well as holding the mouse still while performing a mouse click). One potential problem with using a mouse is in learning to map a desk-based mouse movement coordinate system to a screen-based cursor movement system. Translation processes are known to create difficulty for novice older adults doing digit-symbol processing tasks and even for skilled older typists preparing keystrokes (e.g., Bosman, 1993, 1994). In the second study, we test a light-pen as a target selection device. The light-pen is a direct addressing device, in that, unlike the mouse, there is no coordinate translation necessary to map the target selection response onto the display. In theory, direct address devices ought to be superior to indirect ones, such as a mouse or trackball. We find that the light-pen is far superior to the mouse for both target acquisition (move and click) and for drag operations (acquire and drag). Further, interactions with age show that older adults gain more speed when using the light-pen compared to younger ones, and are helped more when targets are made larger. Study 1.Cursor Key versus Mouse as an input device for Older Adults Participants: 27 computer novices, 9 in each of the following 3 groups, 20s, 50s, and 60s. Method Description of Task The task was to move the cursor, which always appeared at the bottom left hand corner of the screen to the target on the screen using either the cursor keys or the mouse. The target appeared at 12 different locations on the screen. On each trial a row of 4 targets was presented on the screen, and the center of one target was highlighted. This was the target the subject had to move the cursor to. The middle of the row was either 3, 6, or 9 cm above the starting position of the cursor. The middle of the target within the row was either 3.6, 9, 14.4 or 19.8 cm to the right of the starting position of the cursor. There were 3 target sizes, the smallest was a square 6 by 6 mm, the medium sized target was 9 x 9 mm, and the largest target was 12 by 12 mm. The size of the cursor corresponded to the smallest target size. During each trial, two measures were obtained; homing time and movement time. Homing time was the time that elapsed from the appearance of the row of targets to the time the subject began using one of the input devices. Movement time was the time that elapsed from the first movement of the input device until the enter key was pressed after the cursor had been properly positioned over the target. The subject initiated each trial by pressing the space bar. A row of targets would then appear on the screen and the center of one of the targets would be highlighted. Subjects completed 4 blocks of trials, 2 using the cursor keys, and 2 using the mouse. In each block there were 144 trials corresponding to 12 occurrences of each of the 12 target locations with each different target size occurring 4 times at each target location. Within each block the order of trials was random, and a different random order was generated for each block and each subject. The device order across blocks was either mouse, cursor, cursor, mouse, or cursor, mouse, mouse, cursor. Within each age group subjects were assigned to these conditions in alternation. Results: Results were analyzed with a repeated measures analysis of variance in which group (20s, 50s, 60s) was the between subjects factor, and device (mouse or cursor), and target size (small, medium or larger) were within subject factors. www.psy.fsu.edu/~charness/apa95/ 1/9

.c

H F-XC A N GE

H F-XC A N GE

N lic

k

to

bu

y

c u-tr a

.d o

o

w

m

C

m

o

.d o

23/2/2009 http://www.psy.fsu.edu/~charness/a… and device (mouse or cursor), and target size (small, medium or larger) were within subject factors. .c ck

w

w

w

w

w

C

lic

k

to

bu

y

N

O W !

PD

O W !

PD

c u-tr a c k

Movement Time: Time to position the cursor on the target once started yielded a significant main effect of Group (F(2,24) = 12.89, p < .0002) indicating that the older adults were slower. There was also a significant main effect of Device (F(1, 24) = 472.15, p < .0000) indicating that the mouse was faster than the cursor keys. We also found a significant main effect of Size (F(2, 48) = 206.99, p < .0000) indicating that the largest size was the fastest, and the smallest size was the slowest. As well we found a significant Group by Size interaction (F(4,48) = 14.20, p < .0000) indicating that the older adults were disproportionately slowed by the smallest target sizes. There was also a marginal interaction of age and device, F(2,24) = 2.95, p = .0714, indicating that the speed difference between young and older adults increased with the mouse. Movement Time: Group Means and Standard Deviations Device

Target Size

Age: 20s

Age: 50s

Age: 60s

Mouse

1

1.98 (.56)

4.86 (2.33)

6.13 (2.17)

Mouse

2

1.40 (.36)

3.07 (1.51)

4.17 (1.89)

Mouse

3

1.20 (.31)

2.65 (1.43)

3.25 (1.48)

Cursor

1

7.81 (.76)

8.99 (1.53)

10.80 (1.09)

Cursor

2

7.00 (.71)

7.38 (1.02)

8.66 (.67)

Cursor

3

6.33 (.61)

7.01 (1.05)

7.79 (.74)

The speed advantage for the mouse was fairly predictable, given that movements are magnified by the "gain" of the device. A 1cm movement of the mouse on the desk is translated into a much greater size movement of the cursor on the screen, even with the relatively low (slowest setting) acceleration pattern. In contrast, depressing the cursor key yields a small constant movement with much less gain. A major problem with this study is the very small sample size, meaning that there is very little power to detect small effects, such as higher order interactions. Study 2. Mouse versus Light-pen Although the mouse was a much more effective device than cursor keyes for all users, young and old alike, the ratio of young to old performance with that device was about 3:1, much above the typical slowing constant of 1.2 to 2.0 for reaction time studies comparing old with young response time (e.g., Cerella, 1990; Myerson et al., 1990). For cursor keys the ratio of old to young movement time was closer to 1.3:1. Such divergent ratios argue that there may be important differences in the way that some older adults homed in on targets. There are two potential problems with mouse use for older adults. One problem with a mouse is fine-positioning movements, such as controlling under and overshoot. Hand tremor and difficulty controlling movements due to arthritis are more likely to affect older than younger users. The second problem is with translating coordinates in the desk space (lateral movement of the hand in the plane of the desk surface) to those in screen space (cursor movement across the screen). If problems are more attributable to translation processes, then use of a direct positioning device such as a light-pen or touchscreen should be differentially beneficial to older users. In the following study we contrast light-pen movement time with mouse movement time. We also contrast button acquisition, moving between buttons and clicking at the end points of movement trajectories, with "dragging" a button between two positions, since dragging is slower and requires more effort to keep the mouse button or pen tip down during the movement. Methods Subjects A total of 29 subjects have participated to this point in the experiment. (We expect to have 16 young and 16 old shortly.) The young adult group consisted of 14 subjects ranging in age from 17-25 (Mean = 19.7). These subjects participated in return for class credit required in an introductory psychology course. The 15 older adults ranged in age from 54-79 (Mean = 67.1). These subjects participated for five dollars and a certificate of participation. All subjects reported minimal to no experience with the mouse in interviews before the experiment began. All subjects reported normal or corrected to normal vision and had no physical impairments or pathology which would www.psy.fsu.edu/~charness/apa95/ 2/9

.c

H F-XC A N GE

H F-XC A N GE

N y bu

bu

y

N

O W !

PD

O W !

PD

to

k

Equipment The experiment was conducted on two Gateway 2000 P60 IBM compatible computers. 15 inch SVGA monitors running at a resolution of 800 x 600, 256 colors and a refresh rate of 72 megahertz were used to display the interface. The monitors were approximately 45 to 65 cm. away from the subjects, at a height of 80 cm. (bottom of visible screen) and an approximate angle of 104 (± 1) degrees. Brightness and contrast were set to the maximum possible values. The input devices used were a standard Microsoft mouse and an FTG Data Systems model FT-356H light pen. The mouse was controlled via the FTG light pen software and was set to the slowest possible tracking speed (mouse to cursor movement ratio of 2.5/12 cm.). The mouse was used directly on the table top 65 cm above the floor, in an area which was area approximately one foot square. Subjects were seated in armless chairs to allow for unimpeded motion of the arm. The monitors, mice and table were cleaned before each subject performed the task. Design A mixed design was used with age (old or young) being the only between subjects factor. Additional within subject factors included device (mouse and light pen), task (buttons and scrollbars), size of button (large and small), position of button (1 - 9), orientation of scrollbar (horizontal and vertical), length of scroll (5, 7.5 and 10 cm.). Dependent measures included movement time (MT) to target in milliseconds, MT for the actual scroll (scrollbars only), and number of errors, defined as clicking on a non-target screen area, during movement to the target. The various tasks were counterbalanced such that the button tasks were always followed by the scroll tasks. This was done because the scroll task required skill at moving to a target (the scroll button) the size of a small button and the movement of interest at that task was performing the actual scroll. Performance at getting to the scroll was thus facilitated by the button task, but there should have been no transfer to the scroll task itself. In addition the target size (large and small) and orientation (horizontal and vertical) order of use for the mouse and the light pen was balanced across subjects. This resulted in two task orders for each device: Large then small buttons followed by horizontal then vertical scrolls, or small then large buttons followed by vertical then horizontal scrolls. Procedure Upon arriving at a large common area subjects were greeted and placed in a small lab room containing a single computer. The experimenter activated the program and input various information regarding subject demographics and experimental procedures. After this, the experimenter brought up the introductory interface (Figure X). Subjects were told that this screen represented a typical experimental screen, with the exception that there were fewer buttons and scroll bars. The experimenter then explained each of the various displays across the bottom of the screen. These told the subject what block of practice they were on, the movement time (or scroll time) in milliseconds, the trial number, the target number and the device to be used, from left to right respectively. The basic functioning of the mouse in terms of movement and the concept of clicking was explained. Subjects then were allowed to practice the button task and the scroll task three trials each. Afterwards instructions for the light pen were given and again the subjects performed each type of tasks three times. Key points presented to subjects were that the mouse and pen must be clicked while keeping the tip of the cursor within the target, and that the light pen should be held near the tip and perpendicular to the screen. Also, subjects were shown that if they deviated from a scroll bar during a drag they did not have to go back and grab the button again as long as they did not let up on the mouse key or lift the pen from the screen. The targets on this interface were both considerably larger than the ones on the actual test interfaces. In addition subjects were not prompted to perform in a speeded manner during this phase, the purpose was only to instruct them in the basic functioning of each device and task. After the introduction the experimenter proceeded to the first task which was either large or small buttons (Figs. X and X). The buttons were numbered 1 through 9, left to right across the screen from the top. The distance to a button was held constant across both sizes of button ( Positions 1 and 3 = 16 cm., 4 and 6 = 12 cm., 7 and 9 = 9 cm., 2 = 13.5 cm. , 5 = 8 cm. and 8 = 2.5 cm.). The subject was then told that they would engage the timer by clicking on the start button. They were then to proceed to the appropriate target as quickly as possible and stop the timer by clicking within the target. They were informed that they would do this five times for a given target, at which point the tone would sound indicating that the subject should move to the next target. After completing all nine targets, the subject was prompted by the computer to switch devices and repeat the process. This procedure occurred a total of five times with each device. After completing one size of target the entire procedure repeated for the remaining target size. The same procedure occurred with the scroll bars. There were only three scroll bars on the interface (Figs. X and X). The task in this case was to engage the timer and move to the scroll bar, grab the button, and drag it to the end as fast as possible. If a subject released the button in the bar they would have to grab it again. If the button was released outside of the bar the subject would have to repeat the grab and drag from the beginning of the bar. The scroll bar buttons were the same size as the small buttons and were in the same position as the top three for vertical bars or the left three small buttons for horizontal bars. All scroll drags were from top to bottom for vertical bars or left to right for horizontal bars. Before subjects began the scroll task, the experimenter reminded them of how the scroll bars functioned. The number of trials per target, alternation of devices and number of practice blocks remained the same as during the button www.psy.fsu.edu/~charness/apa95/ 3/9

o

m

lic

m

C

w

w

o

C

lic

k

to

23/2/2009 http://www.psy.fsu.edu/~charness/a… w experiment began. All subjects reported normal or corrected to normal vision and had no physical impairments or pathology which would .c .c .d o .d o c u-tr a c k c u-tr a c k have an effect on the performance of the experimental tasks. w

w

w

H F-XC A N GE

H F-XC A N GE

N y bu

bu

y

N

O W !

PD

O W !

PD

to

k

o

During the procedure the experimenter repeatedly urged the subjects to move as quickly as possible. Subjects were also informed that they could take a brief rest if they began to feel tired or physically stressed during the experiment. Afterwards subjects were paid or given experimental credit depending upon their age and older adults were given their certificate. All subjects were shown a Microsoft Word interface with scrolls and buttons as well and given a written debriefing before being dismissed. RESULTS Since the trial did not end until the final click on a button or drag of a button to the end of the scroll bar, errors are accommodated within total movement time scores. Thus, we report only total time as the dependent variable. Errors, defined as clicks not on target buttons, or drags away from the boundaries of the scroll bar were quite infrequent in both age groups (<.1 error per condition). Move and click task There was a significant effect of age (F (1,27) = 23, Mse = 1991527, p < .0001, with older adults (M= 1587 ms) slower overall than younger ones (M = 1020 ms). The ratio of old to young time is ~ 1.6, what would be expected from typical age-related slowing. A significant size effect, F (1,27) = 65, Mse = 293486, showed that small buttons (M= 1497 ms) took longer to acquire than large buttons (M= 1130 ms). Trial block was also a significant factor (F (4,108) = 26.04, Mse = 58836, p < .0001, with time to perform the task decreasing across blocks, as seen below.

There was also a significant device effect, F (1, 27) = 189, Mse= 592750, p < 0001, with the the mouse (1759 ms) roughly twice as slow as the light-pen (868 ms). These main effects were mediated by interactions: device by age, size by age, block by age, block by size by age, block by size, and block by device by size. We will concentrate here mainly on interactions with age, since the point of this study was to look at the relative utility of these devices across adult age. The device by age effect, F (1,27) = 19.01, Mse 592750, p < .0002, showed that the gap in speed between young and old was greater for the mouse (845 ms) than the light-pen (287 ms).

www.psy.fsu.edu/~charness/apa95/

m

lic

m

C

w

w

o

C

lic

k

to

23/2/2009 http://www.psy.fsu.edu/~charness/a… w functioned. The number of trials per target, alternation of devices and number of practice blocks remained the same as during the button .c .c .d o .d o c u-tr a c k c u-tr a c k task. w

w

w

4/9

H F-XC A N GE

H F-XC A N GE

N bu

y to k lic

.c

.d o

m

w

o

m o

c u-tr a c k

w

w

.d o

http://www.psy.fsu.edu/~charness/a… w

w

w

C

23/2/2009

C

lic

k

to

bu

y

N

O W !

PD

O W !

PD

c u-tr a c k

Phrased in terms of Brinley ratios, the slowing factor of old to young drops from 1.64 for the mouse, to 1.40 with the light-pen. That is, no matter what your metric, the light-pen goes a long way to closing the age gap in target acquisition performance for computer use. The age by size effect, F (1,27) = 7.34, Mse = 293485, p < .02, showed that the gap between young and old diminished as targets shifted from small (688 ms) to large (444 ms). In terms of Brinley ratios, they dropped from 1.60 to 1.49. Clearly, having larger buttons as targets is disproportionally helpful for older adults.

The age by block effect (F(4, 108) = 7.59, Mse = 58836, p < .0001), showed that the gap between young and old diminished with practice (773, 577, 535, 486, 460 ms, in blocks 1 to 5 respectively). Practice was differentially beneficial to older adults, perhaps because the young were closer to optimal performance to begin with.

www.psy.fsu.edu/~charness/apa95/

5/9

.c

H F-XC A N GE

H F-XC A N GE

N bu

y to k lic

.c

.d o

m

w

o

m o

c u-tr a c k

w

w

.d o

http://www.psy.fsu.edu/~charness/a… w

w

w

C

23/2/2009

C

lic

k

to

bu

y

N

O W !

PD

O W !

PD

c u-tr a c k

The block by age effect was mitigated by an age by size by block interaction (F (4, 108) = 3.2, Mse = 79743, p < .02. The beneficial effects of practice were greater in older adults for small than large buttons.

As well we observed an age by block by device interaction. Indicating that past the initial block, the gap between young and old was narrowed for the lightpen, but remained constant for the mouse.

www.psy.fsu.edu/~charness/apa95/

6/9

.c

H F-XC A N GE

H F-XC A N GE

N to k lic .d o

m

w

o

.c

bu

y

m o

c u-tr a c k

w

w

.d o

http://www.psy.fsu.edu/~charness/a… w

w

w

C

23/2/2009

C

lic

k

to

bu

y

N

O W !

PD

O W !

PD

c u-tr a c k

It is worth briefly examining one non-age interaction, the block by device by size interaction. The biggest advantage for the light-pen is on the very first block with the small targets. The light-pen's advantage then remains more or less constant across blocks for each target size. The Scroll Bars Task The significant factors for the scroll bar task were age, blocks, blocks by age, device, device by age, orientation (vertical versus horizontal), blocks by orientation, device by orientation, and block by device by orientation. Again we will concentrate primarily on age effects and their interactions. As expected, the main effect of age (F(1,25) = 17.83, Mse = 66726704, p < .0003, was attributable to slower scrolling by old (5987 ms) than young (3016 ms) users, with the Brinley ratio being 1.99. As well the blocks effect F (4, 100) = 9.9, Mse=7263141, p < .0001, indicated that people became faster over blocks (5750, 4487, 4414, 3761, 3823 ms respectively for blocks 1-5). The device effect, F (1,25) = 30, Mse = 37209509, p < .0001 showed that the light-pen (3027 ms) was much faster than the mouse (5866 ms). The orientation effect (F (1,25) = 7.82, Mse = 19714469, p < .01) indicated that people dragged the cursor button faster in the vertical (3921 ms) than the horizontal (4972 ms) direction. (Gravity helps.) These main effects were mediated by two age interactions. The device by age interaction (F(1,25) = 4.677, Mse = 37209509, p < .015), showed that the age gap was wider with the mouse (4107 ms) than with the light-pen (1834 ms). Again a Brinley ratio showed the mouse to be higher (old/young = 2.06) than the light-pen (old/young = 1.86).

www.psy.fsu.edu/~charness/apa95/

7/9

.c

H F-XC A N GE

H F-XC A N GE

N to k lic .d o

m

w

o

.c

bu

y

m o

c u-tr a c k

w

w

.d o

http://www.psy.fsu.edu/~charness/a… w

w

w

C

23/2/2009

C

lic

k

to

bu

y

N

O W !

PD

O W !

PD

c u-tr a c k

The blocks by age interaction (F(4, 100) = 3.275, Mse = 7263141, p < .02) indicated that the gap between young and old tended to get smaller with practice (4431, 2813, 3229, 1982, and 2397 ms for blocks 1-5 respectively).

Briefly, the highest order interaction, blocks by device by orientation, indicated that trend of decreasing differences between light-pen and mouse across blocks depends on orientation, with more speedup across blocks for horizontal than vertical with the mouse. The most difficult condition initially showed the most improvement across trials. CONCLUSIONS Across the two studies, we can suggest a rank ordering for input devices of light-pen, then mouse, then cursor keys for fast acquisition of targets on a display. This is true for young and old. Movement time to a target takes twice as long with a mouse than with a light-pen. Drag time goes up by 50% for novices using a mouse compared to a light-pen. The pen is indeed mightier than the mouse, particularly for novice older adults, given the interactions of device with age. Although a typical light-pen device costs about $300 compared to mouse devices as inexpensive as $15, serious consideration should be given to making this device more widely available with computer systems, and particularly in cases where older users are involved. Other direct positioning devices, such as a touchscreen, might also be considered. The size of target manipulation indicates that small targets pose problems for all age groups, but particularly for older adults. By increasing the size of visual icons, such as buttons, software designers can be particularly helpful to older novice adults. Also, older adults derive 8/9 www.psy.fsu.edu/~charness/apa95/

.c

H F-XC A N GE

H F-XC A N GE

N y bu

bu

y

N

O W !

PD

O W !

PD

to

k

o

Designers of software interfaces need to take these findings into consideration, particularly concerning button (target) size. There is a tendency to use many small icons on toolbars. Icons (buttons) are favored over menu items since more of them can be squeezed into the same screen space. But, their advantage over menu items may diminish if movement time is taken into consideration. Menu items can be easily accessed by holding down a shift key (e.g., alt) and the first letter of the menu-triggering item, followed by another unique first letter to trigger the option within the menu. A skilled typist does not need to take her hands off the keyboard to do this, compared to activating a button with a pointing device. CAVEATS Whether users will always derive a strong benefit from a light-pen over a mouse is open to a variety of other factors that we did not manipulate, and our conclusions are limited by one important constraint. First, one major constraint is that we set mouse movement speed to the slowest possible setting. Pilot research showed that the mouse was difficult to control for novices, old and young alike, and a faster movement speed would make it excruciatingly difficult. Skilled users however, would be inconvenienced by the slow speed. Thus, the relative advantage of a light-pen over a maximum movement setting for the mouse is unknown. However, for this type of movement task, we doubt that the mouse will ever catch up given the problems of acquiring small targets when speed increases due to problems such as overshooting the target. Second, we were testing only a small part of the complex array of activities of computer users. So, although the light-pen is superior for novices, and particularly for older adults in pointing tasks, this advantage needs to be weighed against the disadvantage inherent in moving your hands away from the keyboard when you must both enter text with the keyboard and click on buttons and scroll bars. Obviously, the smaller movement necessary to reach a mouse near a keyboard than a light-pen which must be moved a greater distance to the screen should make the mouse somewhat more attractive for mixed editing operations. We need to carry out further research with such tasks to determine the relative tradeoffs. Third, we do not know how far apart the two devices will be when in the hands of experienced users operating near asymptotic performance. References Charness, N., Kelley, C. A., & Bosman, E. A. (in preparation). Training novice adults in word processing: effects of age and interface. Cerella, J. (1990). Aging and information-processing rate. In J. E. Birren and K. W. Schaie (Eds.), Handbook of the psychology of aging, (3rd Ed.) (pp. 201-221). San Diego: Academic Press. Bosman, E. A. (1993). Age-related differences in motoric aspects of transcription typing skill. Psychology and Aging, 8, 87-102. Bosman, E. A. (1994). Age and skill differences in typing related and unrelated reaction time tasks. Aging and Cognition, 1, 310-322. Kelley, C. L., & Charness, N. (1995). Issues in training older adults to use computers. Behaviour and Information Technology, 14(2), 107-120. Meyerson, J., Hale, S., Wagstaff, D., Poon, L. W., & Smith, G. A. (1990). The information-loss model: A mathematical theory of agerelated cognitive slowing. Psychological Review, 97, 475-487.

www.psy.fsu.edu/~charness/apa95/

m

lic

m

C

w

w

o

C

lic

k

to

23/2/2009 http://www.psy.fsu.edu/~charness/a… the size of visual icons, such as buttons, software designers can be particularly helpful to older novice adults. Also, older adults derive w . d o .c .c .d o c u-tr a c k c u-tr a c k more benefit from practice than do younger adults (in part, probably because of floor effects for the young), so they need to be encouraged to practice and not compare their initial performance to that of their younger counterparts. w

w

w

9/9