Young Children'S Skill in Using A Mouse To Control A Graphical Computer Interface

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Computers Educ. Vol. 19, No. 3, pp. 19%207,1992 0360-1315/92$5.00+ 0.

00
Printed in Great Britain. All rights reserved Copyright 0 1992Pergamon Press Ltd

YOUNG CHILDREN’S SKILL IN USING A MOUSE TO


CONTROL A GRAPHICAL COMPUTER INTERFACE

CHARLES CRGGK
Department of Psychology, Durham University, Science Laboratories, South Road,
Durham DHI 3LE, England

(Received I.2 February 1992; accepted 13 March 1992)

Abstract-Preschoolers and children within the first 3 years of formal education are compared in respect
of their performance on four tasks that embody basic skills underlying mouse-based control of a graphical
computer interface. The children had not used such an interface before and were compared after a small
number of introductory and practice sessions. Results from one preschool child studied over a longer
period are included. The children’s performance is set against that of their own novice teachers and a group
of experts. Age differences are found on all tasks but the absolute level of control is impressive for all
children involved. Younger children are mainly disadvantaged by difficulties in repositioning the
instrument on the working surface while keeping the screen pointer fixed. No gender-correlated differences
were found at any age. It is suggested that the results encourage the exploitation of graphical interfaces
within the design of early educational software.

INTRODUCTION

Computers now commonly available for preschools and early education may readily support
graphical user interfaces. Some software authors have begun reflecting on the challenge of
programming within this new environment, e.g. [l]. Yet we know little of how easily very young
children respond to this form of interface. There are two broad issues to address in that respect.
Firstly, we might wish to clarify how readily children can manipulate the tools of the interface-
control pointers, make selections, relocate icons and so on. Secondly, we might wish to discover
how far children can interpret the syntax and semantic of the symbolic material making up this
environment-interpreting the meaning of icons, interrelating them, etc. The present paper
considers the first of these two issues: the more basic (and more tractable) of the two.
Some may feel confident that they already know whether or not young children can use a mouse
to control a graphical interface: there are informal accounts of preschool and classroom practice
that imply that they can, e.g. [2,3]. However, there is a need for more formal observations of what
is possible. We would gain from a clearer idea of just how early the rudiments of WIMP
environments are accessible to children; it would be helpful to identify those features of its control
that were relatively more demanding; we could also usefully explore just how quickly (and with
what degree of autonomy) children might acquire the necesssary basic skills. A clear view on these
questions would help guide practitioners make decisions about the feasibility of adopting materials
using such interfaces and would inform software authors as to what might be realistic ambitions
in the design of interfaces for youngsters.
There are two published studies that are relevant to these concerns. Wilton and McClean[4]
report that the ability to use a mouse to move targets around a screen improves across children
aged 7-9, 11-12 and 14-16. However, their targets were words and it is not clear whether age
differences in pointer control were confounded with age differences in literacy. A useful extension
of the study would be to look at a wider range of component tasks; to involve younger children;
and to specify the extent of practice that lies behind a given level of achievement. A study by Scaife
and Bond [5] does include data from younger children (from 5 years old) using a mouse. However,
neither of the tasks they used (tracking a moving target and tracing a letter outline) are typical
of actions commonly employed within graphical interfaces and their report does not go beyond
an account of these children’s very first encounters.
The present study reports data from four tasks that were designed to elicit certain basic skills
judged necessary to control typical graphical interfaces. The dimensions of target material and the
general character of tasks were chosen to reflect demands characteristic of software that employs
199
200 CHARLESCROOK

this kind of interface. Children were observed from the earliest age at which they might encounter
computers within an institutional setting and they were observed under circumstances that were
typical of their normal work with computers. Their performance in mastering these tasks is
presented alongside certain benchma.rks: adult novices (the children’s teachers) and experts (regular
users of such interfaces).

METHOD

Subjects
Four categories of subjects were involved. (i) Preschoolers: 6 children of average age 3 years 10
months (3 : 10) were arbitrarily selected from within a nursery school. They had been attending there
for no more than 6 months and participated in this study across a 3-month period. (ii) Primary
pupils: 36 children representing the first 3 years of schooling were involved from around the middle
of the school year. Six boys and 6 girls were randomly selected from each class. Their average ages
were: 5 : 3, 6 : 6 and 7 : 4. (iii) Adult novices: the three (female) class teachers of these children were
tested. (iv) Experts: 12 (6 male and 6 female) members of the postgraduate community within the
author’s own department volunteered. All had made regular use of graphical interfaces for at least
the preceding year. All subjects were judged to be right handed. All, except the experts, had no
previous experience with graphical interfaces; however, all could claim some experience of using
a computer keyboard.

Equipment
The same computer (Acorn Archimedes) and peripherals (Acorn 3-button mouse) were used in
all the observations reported involving schoolchildren and adults. This mouse measured 6 cm
wide x 10 cm deep. The buttons were offset 1 cm from the top of the instrument and each measured
1.5 x 2.8 cm. For all tasks described here, only the left button was used. The mouse mat was cut
from the wetsuit material commonly used for commercial mats; it measured 38 x 38 cm and was
placed to the right of the computer. A mouse was specially constructed for the preschoolers (to
fit more easily into their smaller hands). A model car was mounted on top of the variety of mouse
described above and a single button located at the front of the car. A mat of the same material
was used but of twice the size and positioned in front of the computer.

Tasks
Four computer programmes were written to present the four tasks (A, B, C and D) illustrated
in Fig. 1 and described fully below. The figures are not drawn strictly to scale and, in particular,
do not necessarily show the full screen.
In each case, a screen area of 28 cm wide x 18 cm high was available and the mouse gain was
fixed at 1: 1 (one unit of mat distance traversed corresponded to one unit of screen distance moved).
The timing of important user actions was monitored by the programmes and recorded in data files
stored on a common disc. These events were sampled at a rate of 100 Hz.
Task A. Pointing and selecting: the pointer must be positioned within a target and the mouse
button clicked or double-clicked. A session comprises six trials upon each of which the user selected
(and thereby removed) one of a number of green target rectangles, each measuring 4mm
high x 5 mm wide. Selections involved either a single or double button click while the tip of the
arrow pointer rested on the target, thereby causing it to break up into a brief spray of colour. The
screen was conceived as a 7 x 7 invisible grid: at the start, and after each selection, the pointer
appeared in the centre of this grid and remaining targets were positioned randomly within other
grid units such that no two were in the same unit and none was adjacent to the pointer starting
position. The current target rectangle was identified by incorporating either one or two pink 1 mm*
squares within it. One square signalled a single button click requirement; two squares, a double
click. A double click was defined as two successive button presses separated by no more than 0.4 s.
Figure 1A illustrates a portion of the screen following two successful selections. To remove the next
target the tip of the arrow must be placed in the rectangle containing two squares and the button
double clicked.
Young children’s mouse control 201

Fig. 1. Diagramatic representation of the four computer tasks under examination. The mouse controls
the arrow pointer. A-The pointer must be positioned within the marked target and the mouse button
clicked or double-clicked; B-a square target must be moved into an open frame; C-three circles must
be joined by dragging lines from one to another; D-the pointer must be placed within circles which are
then removed and the pointer repositioned.

Tusk B. Dragging: a square target must be moved into an open frame on the screen by placing
the pointer within this target and moving the mouse to the frame while the mouse button is held
pressed. A session of this task comprised three such trials. On all trials, the receiving frame was
randomly positioned on the horizontal axes with its base 3 cm from the top of the screen. For the
first trial, a large (1.6 cm) square target was positioned 9.5 cm from the top of the screen. For the
second, this large square was 17 cm from the top. For the third, a small (0.9 cm) square was
positioned 17 cm from the top. The horizontal starting position of all target squares was randomly
determined. Targets would follow the mouse as long as the tip of the pointer arrow was within
them and the button held down. A trial was concluded when the arrow tip was within the frame,
while still holding the target, and the button then released. Figure 1B illustrates the pointer close
to a target. Correctly positioning the target was marked by a short sound and change of target
colour (red to purple).
Task C. Fixing: three circles are joined by successively dragging a line from the inside of one
to the inside of another and then releasing and re-pressing the mouse button to fix the connection
drawn. A session comprised two trials. The first involved three white circles targets positioned to
form a small triangle; the second, three circles positioned to form a larger triangle. Circle diameters
were 1.5 cm. Triangle dimensions were designed to vary from trial to trial as follows. One circle
was always drawn with its centre 4 cm from bottom and 3 cm from the left of the screen boundary.
The distance to the next was randomly chosen to lie between 4 and 7 cm (small triangle) or 11 and
13 cm (large); the first two angles of the triangle were allowed to vary between 40 and 70”. Within
such a configuration, a red line could be picked up by positioning the pointer arrow tip within a
circle and holding down the mouse button. Any movement with the button held down was now
followed by a continuously redrawn line-at any moment the line always being shown between the
current arrow position and its circle starting point. To join two circles, the button needed to be
released within the second circle and then, before moving the pointer out of its area, pressed again.
This caused the single line to be instantly redrawn as a permanent join of three parallel lines. Figure
1C shows a trial in progress: two circles have been fixed and the moving line is being held in the
third circle awaiting the final fixing button press. The completion of a triangle was marked by
flooding the space within it with red.
202 CHARLESCROOK

Tusk D. Mouse re-positioning: the tip of the pointer must be placed within circles that,
thereby, are successively removed. The task is demanding because the pointer is continuously
replaced to a fixed starting point on the screen; this requires a periodic repositioning of the
mouse itself-which will otherwise move off the edge of the mat surface. A session comprises
five groups of four trials, these groups differing in the direction of required movement. A
trial involves placing the pointer tip within a 1.8 cm dia coloured circle, upon which this target
silently implodes (as a burst balloon) and its colour is instantly redrawn in a bank of squares
behind the target. The initial target and the pointer positions are randomly determined on a 12 cm
invisible line; these two lines are 11 cm apart. A visible line is placed 1 cm behind the pointer
starting space to highlight the current direction of movement. The first four groups of trials
differ in requiring movement to be either left-right, right-left, updown, or down-up; the ordering
of these groups being randomly determined on any session. The final group of four trials is
a random mixture of the four possible movement directions. There is no pause between
the completion of one trial and the starting position for the next. Figure 1D shows a session
underway: the current trial requires left to right movement, and the pointer is at its starting
position.

PROCEDURE

Preschoolers
The research was carried out in a quiet room where computer activities were normally located.
Introducing the interface to these very young children was necessarily less tightly controlled than
was possible with the other groups: there had to be allowance for abandoning an activity should
a child become distracted or frustrated. The data to be presented for these children do come from
a session of the four tasks that was identical to that reported for all other groups. However, the
period of preparation before that session was more variable. All children began with a short session
in which movement of the mouse caused a snake-like object to move on the screen. This snake
would change colour on moving through target squares that occasionally appeared. All children
picked up the association between hand and screen on this first exposure. Subsequently, they had
three sessions of guided experimentation with a simple painting programme whereby the pointer
acted as a pen for painting on the screen; different colours and thicknesses were available from
a palette at the top of the display.
These children were then introduced to the four tasks over seven sessions spread over a period
of 3 months (a pace reflecting their normal frequency of access to a computer activity). An adult
was always present and offered direction and encouragement when it seemed appropriate. Towards
the end of this period, a session could often comprise all four tasks and might involve minimal
adult intervention. Data here is taken from a final session upon which no explicit guidance was
given.
After this procedure was complete, one child was tested in her own home on tasks A and D on
a roughly daily basis for a period of 30 days.

School children
The work was conducted in a quiet corner of the classroom where computer activities were
typically located. In the first of five sessions, these children were also shown mouse/pointer
association through the snake activity. They were then introduced to the four tasks in a hands-on
manner. Not all of the youngest children were able to complete all four on the first session,
although they did so on the second. Sessions occurred on average once every 5 school days.
For the first the experimenter intervened verbally and manually where it seemed the child
needed encouragement or rescue. This was most likely in respect of (a) remembering to double
click and to do so with appropriate click separation, (b) repositioning the mouse as it wandered
off the mat and (c) keeping the button pressed in order to effect a dragging motion. The final two
sessions were without intervention but with encouragement to work at a quick but comfortable
pace. Data are reported from the final session. Sessions typically lasted about 10min with the
experimenter present throughout; no other use of the graphical interface occurred outside of this
testing.
Young children’s mouse control 203

Adult novices
The tasks were demonstrated and explained and subjects performed them in five sessions carried
out one per week. They were instructed to work at a steady and comfortabIe pace. Data are taken
from the final session.

Experts
Subjects were shown the tasks and asked to do five sessions in their own time with no more than
one on any given day. They were instructed to do the tasks at the pace they would use the interface,
one that was typical of their own working practices when con~ntrating and engaged. Data are
reported from their final session.

RESULTS

Summary data from all groups are depicted in Fig. 2. All histograms report mean measures of
response speed. Bar lines above histograms indicate -I- 1 SE of the mean. Reading from left to right,
data is shown for the six preschoolers (PS), for the three primary school classes [from youngest
(1) to oldest (3)], for each of the three teachers of those classes separately (T,, T,, T,), and for
the group of experts (E).
Substantive quantitative analysis will be limited to comparisons of the three school-based groups,
as procedures were most uniform for them. The other groups exist as useful benchmarks for judging
the absolute levels of achievement. Analysis of variance is used extensively; in situations involving
repeated measurements on successive trials, the average values contributed by individual subjects

p 5

8 4

g 3

PS

6 4

r
g 6
t 3
:: 5
a,
x
cn 4 In 2
t
3

2 1

PS 1 2 3
i
PS
I rl
*1'zT3 E

Fig. 2. Performance data (latencies) appropriate to each of the four tasks studied. PS = preschoolers,
l-3 = three school classes, T,-T, = three novice teachers, E = experts. Means and their standard errors
are shown. A-Average time to complete a trial of pointing and selecting; shaded or dotted areas show
time to get close to the target; B-median time to complete dragging task; C-mean latency to complete
join task; shaded or dotted areas show times with larger targets. D--mean latencies for repositioning task.
204 CHARLESCROOK

are medians. This dampens influences upon time scores that arise from the occassional distractions
typical of classroom life. Effects associated with subjects’ gender were considered but not reported
here as no such differences were found in any performance variable.

Task A
Histogram 2A shows average times to complete a trial for pointing and selecting. Timing starts
from the first detected movement of the pointer and ends on an appropriate button press. The age
(class) effect is significant: F2,,, = 5.72, p < 0.01. As would be expected, trials requiring a double
click took longer than those requiring a single click: F,,?, - 5.77, p < 0.05. However, there was no
significant interaction between age and trial type, indicating that double clicking was no more
demanding for the younger than it was for the older. No irrelevant presses on the other two mouse
buttons were recorded.
The longer time taken by younger children might reflect the time taken to locate and approach
the correct target and/or difficulties associated with keeping the pointer still in the very small target
area while a click was made. Certainly, the experimenter noted that, in early sessions, the effort
of trying to press would sometimes move the pointer out of the small target space. However, by
the fifth session this does not seem to be a problem. The time taken to get close to a target was
analysed separately from the time between getting close and correctly button pressing on it. “Close”
here means being within a 1.5 cm radius of the target centre and the time to get there is that
proportion of the total time-shown as shaded (or dotted for teachers) in the histogram in Fig.
2A) (data not avilable for the expert group). This time does show a significant age effect:
F2,% = 4.44, p < 0.05. The time between being close and making the selection press is not sensitive
to the age variable.

Task B
Figure 2B shows, for this dragging task, the median latency to complete a trial, timed from the
point at which the target is first picked up (a button press made upon it). The age effect for school
children is significant: F,, 29= 6.69, p < 0.005. However, this difference is not due to an inefficiency
in holding the target during the drag: there was no age effect associated with the number of times
the target is dropped by releasing the button. The cumulative amount of time spent holding the
target did differentiate the age groups: medians of 5.42, 3.85, 3.51, F2,29 = 5.62, p < 0.01.

Task C
Figure 2C shows, for the drag-and-fix task, the mean latency to complete a join action. There
are always three successful joining sequences. If we label them: (l)--(2), (2)-(3), (3)--(l), then
performance data here are derived from the average latencies for (2x3) and (3)-(l). In the
histogram the data from the smaller triangle of targets are shown unshaded, while data from the
larger triangle are shown shaded. The effect of triangle size is significant: F2,29= 22.6, p < 0.001.
Also, younger children do take longer, F2,29= 6.54, p < 0.005. Moreover, there is an interaction
between age and triangle dimension: Ez,t9 - 4.99, p < 0.05. This appears to reflect a more striking
difficulty for the younger children in joining the larger figure.
As would be expected, the longer times taken by younger children reflect more failures to execute
the fixing action-pressing again after successfully releasing the line in the new circle. Failure on
this means restarting that join. If links were correctly executed on each occasion, the minimum
number of discrete pickups of joining lines would be 3 for each triangle. The average number per
trial is actually more than this for younger children, but for all age groups the absolute numbers
are still close to optimal: 4.0, 3.5 and 3.3 for the 5, 6 and 7 year-olds respectively. Thus, the overall
efficiency of executing this sequence is good for all age groups.

Task D
Figure 2D illustrates differences among experimental groups in respect of the median latency for
the pointer simply to contact the target, measured from that moment on a given trial from which,
after contacting a target, the pointer is first moved. The differences among the school children are
significant: F2,29= 10.09, p < 0.001. These differences are true at all directions of pointer movement,
but there is a main effect of this variable: F,,, = 5.41, p < 0.005. Examination of the averages
Young children’s mouse control 205

associated with moving over the four directions indicates that the only consistent effect is that the
left to right direction is the fastest at all four ages. This may reflect the left to right experience of
controlling writing implements.
Thus, younger children seem less efficient in simply moving to a target-at least, where such
movement requires periodic repositioning of the mouse on its surface. Examination of changes in
pointer position at very short sampling intervals suggested that discontinuity of movement became
apparent at durations of pausing (no change in pointer coordinates) longer than 0.1 s. Using this
criteria, younger children did spend more time actually moving the mouse on a trial: F,,,, = 7.61,
p < 0.01. Moreover, their pattern of movement was less smooth. On the average trial, there were
more pauses longer than 1 s in duration, 2.73, 2.12, 1.75, for the three age groups respectively:
F,,, = 9.09, p < 0.001. And there were more pauses in the range 0.1-1.0 (5.8, 5.02, 3.39):
F,, = 3.58, p < 0.05. In contrast, the performance of experts was characterized by virtually no
pauses longer than 1 s and an average of only 0.2 trials had any pauses at all in the 0. l-l .Os range.

Tasks A and D, practice data


Figure 3 shows changes in the performance of one preschooler on Tasks A and D across a 30-day
period of almost daily practice. The dependent variables are the same as those reported in Fig. 1A
and D. The graph suggests that the latencies typical of early performance (Fig. 1) for the youngest
children are liable to change over a more lengthy period of practice- although they seem to reach
fairly stable values after about 10 days. For Task A, that value is around 2.75 s and for Task D,
around 1.25 s.

DISCUSSION

This study has successfully monitored young children’s early competance with a number of
activities basic to the control of graphical interfaces. The presentation of results has dwelt upon
significant age differences in performance on these tasks. However, our first conclusion should be
that, despite real age differences, the general level of fluency is very impressive. After only five short
sessions, second and third year primary school pupils are performing to the same standard as their
equally novice teachers. It is also worth stressing that no significant gender-correlated differences

i ” ’ ” ” ” ” ” ” ’ ” 4
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38

Ordinal trail

Fig. 3. Changes in response latency for one preschooler across 30 practice trials on Tasks A (point and
select) and D (reposition).
206 CHARLESCROOK

were found on any of these performance tasks-despite an informal expectation among the teachers
that girls might find controlling the mouse more difficult.
In respect of the age differences, they are consistently in the expected direction. In a comparison
of the three school-aged groups, the largest jump in performance appears to be between Class 1
pupils (5 year-old) and Class 2 pupils (6 year-olds). However, examination of results from pre-
schoolers, who had a rather longer period of familiarization and practice before testing suggest that
these younger children would soon become yet more efficient in their performace. In particular,
observations of the preschooler who had prolonged practice (Fig. 3) suggest that even the youngest
children can approach levels of performance on these tasks that are very close to those of expert
adults.
Yet there are age-correlated differences: what is it within these tasks that the younger children
find more difficult at first? Being clear about this may help us give more effective support to children
at the time that they are learning these skills.
Expectations of what might make the tasks more difficult have not been borne out. Thus, the
size of the conventional mouse is not a serious problem. A smaller mouse was built for the
preschoolers but informal observations at the end of the study indicated that they had no particular
difficulty with the larger instrument once the general skills had been acquired. Neither did these
children have any difficulty with situations where fine control of the mouse was needed or situations
where a more complex pattern of actions was involved. More specifically, there was no age-related
difficulty in keeping the pointer in a small target area while the button was pressed (Task A). Neither
did the demand of double clicking differentiate the older and younger school children. These children
were skillful at dragging icons: the younger children did take longer, but they were no more likely
to drop them. These children also experienced very little difficulty with the more complex dragging
and fixing actions of Task C. Rarely did they have to restart a join because they forgot to adequately
complete the last sequence. It is true that the younger children did have more restarts of this kind
but their absolute number remained small.
Performance on Task D gives an indication of what the prevailing difficulty was that differentiated
the age groups. This task might seem the simplest of all: merely moving to a target, no button press
being required. However, a real difficulty arises from the way in which the pointer consistently
returned to its screen starting position (while the mouse itself obviously stays where the hand has
left it). To keep this task going, the user must strategically reposition the mouse on the mat (without
moving it on the surface itself-that would cause the pointer to move again and further confuse
the situation).
In a sense, Task D isolates a requirement that is common to all of these tasks: whenever the
screen display undergoes any significant rearrangements, it is likely that the mouse will have to be
re-centred to prevent it drifting towards an edge of the mat surface. Difficulty with this arises in
Task A where the younger children take longer to merely approach the target (not to make a fine
positioning on it, or to coordinate the clicking requirements). Difficulty with this underlies the slower
dragging in Task B, as each of the three successive targets reappears in the lower part of the screen
(the pointer having been moved towards the top on the previous trial). Finally, the larger triangle
of Task C presents a problem because the area of screen over which continuous movement has
to be made is larger-not simply because the actions of dragging and fixing are themselves more
demanding for the younger children.
Thus, one of the most significant demands in the present range of basic interface actions is the
strategic re-positioning of the mouse when dealing with a relatively dynamic screen display. How-
ever, there may be another difficulty which these children experience: it is not apparent within the
statistics reported here but it was very visible to the author across many hours spent watching these
young children at work. To trace an accurate (and thus speedy) path to a screen target, the user
must move the mouse in the direction of that target but keep the orientation of the instrument
unchanged (pointing forward, or “north”). So the mouse must not be thought to be like a toy car (or,
indeed, a toy mouse), which might naturally be turned so as to point at the target (projected onto the
mat) in order to approach it at an angle. In such an orientation, the screen pointer will not follow
the heading of the instrument and the user is likely to be confused. This was often seen to happen with
these children and despite frequent encouragement to keep the mouse straight in the early sessions, it
remained just too tempting for them to let it point towards the projected target on the mat.
Young children’s mouse control 207

So we have identified two problems that impede children in their early encounter with this
interface. Can they be avoided? The solution to the repositioning difficulty is probably a period
of exposure to an activity such as Task D where there is a particular premium on this skill: the
results presented in Fig. 3 suggest that even very young children can then acquire fluency with the
basic action. The solution to the second problem of moving the mouse at an angle might be to
develop a cordless instrument with a transmitter positioned at the front. The receiver could be a
strip of sensitive surface at the top of the mouse mat. Thus, if the mouse was angled away from
“north” the signal would be less likely to register. Perhaps the feedback arising under these
conditions would discourage children from allowing the mouse orientation to drift.
These observations suggest a real prospect in developing graphical interface software for very
young children. However, two cautionary observations should be made. Firstly, the skills examined
in these studies are the simplest we might define and there is clearly more research to be done on
the limits to children’s understanding of these interfaces. Secondly, the fact that children can
fluently control computer-based activities in this environment does not, in itself, require us to make
widespread educational provision to ensure that they do [3].

REFERENCES
1. Benzie D., Computers & Educ. 12, 209 (1988).
2. MAPE Micros and the under-fives, Micro-Scope 26, (1989).
3. Crook C. K., Early Child Beev. Cure 69, 5 (1991).
4. Wilton J. A. and McLean R. S., Computers Educ. 8, 45.5.
5. Scaife M. and Bond R., Eurly Child Dev. Care 69, 19 (1991).

You might also like