Improving Selection Performance on Pen-
`Based Systems: A Study of Pen-Based
`Interaction for Selection Tasks
`
`XIANGSHI REN
`Kochi University of Technology
`and
`SHINJI MORIYA
`Tokyo Denki University
`
`Two experiments were conducted to compare pen-based selection strategies and their charac-
`teristics. Two state transition models were also formulated which provide new vocabulary that
`will help in investigating interactions related to target selection issues. Six strategies, which
`can be described by the state transition models, were used in the experiments. We determined
`the best strategy of the six to be the “Slide Touch” strategy, where the target is selected at the
`moment the pen-tip touches the target for the first time after landing on the screen surface.
`The six strategies were also classified into strategy groups according to their characteristics.
`We determined the best strategy group to be the “In-Out” strategy group, where the target is
`selected by contact either inside or outside the target. Analyses show that differences between
`strategies are influenced by variations in target size; however, the differences between
`strategies are not affected by the distance to the target (i.e., pen-movement-distance) or the
`direction of pen movement (i.e., pen-movement-direction). We also found “the smallest
`maximum size” of five pixels, i.e., the boundary value for the target size below which there are
`significant differences, and above which there are no significant differences between the
`strategies in error rate. Relationships between interaction states, routes, and strategy
`efficiency were also investigated.
`Categories and Subject Descriptors: D.2.1 [Software Engineering]: Requirements/Specifica-
`tions—Methodologies; D.2.2 [Software Engineering]: Design Tools and Techniques—User
`interfaces; H.1.2 [Models and Principles]: User/Machine Systems—Human factors; H.5.2
`[Information Interfaces and Presentation]: User Interfaces—Evaluation/methodology;
`Input devices and strategies; Interaction styles; Screen design (e.g., text, graphics, color);
`
`This work was done while the first author was an instructor in the Department of Information
`and Communication Engineering at Tokyo Denki University. The authors were assisted in
`conducting the experiments by Erei Miyajima, Masaya Hagihara, and Kunio Sato.
`Authors’ addresses: X. Ren, Department of Information Systems Engineering, Kochi Univer-
`sity of Technology, 185 Miyanokuchi, Tosayamada-cho, Kami-gun, Kochi, 782-8502, Japan;
`email: ren@info.kochi-tech.ac.jp; S. Moriya, Department of Information and Communication
`Engineering, Tokyo Denki University, 2-2 Kanda-Nishikicho, Chiyoda-ku, Tokyo, 101-8457,
`Japan; email: moriya@c.dendai.ac.jp.
`Permission to make digital / hard copy of part or all of this work for personal or classroom use
`is granted without fee provided that the copies are not made or distributed for profit or
`commercial advantage, the copyright notice, the title of the publication, and its date appear,
`and notice is given that copying is by permission of the ACM, Inc. To copy otherwise, to
`republish, to post on servers, or to redistribute to lists, requires prior specific permission
`and / or a fee.
`© 2000 ACM 1073-0516/00/0900 –0384 $5.00
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000, Pages 384 –416.
`
`APPLE 1006
`
`1
`
`

`

`Improving Selection Performance on Pen-Based Systems
`
`•
`
`385
`
`Theory and methods; I.3.6 [Computer Graphics]: Methodology and Techniques—Interaction
`techniques
`General Terms: Design, Experimentation, Human Factors, Measurement, Theory
`Additional Key Words and Phrases: Mobile computing, pen-based systems, pen-based input
`interfaces, small targets, state-transition models, target selection strategies, classifications of
`selection strategies
`
`1. INTRODUCTION
`Pen-based systems (incorporating a small touch-sensitive screen) have
`emerged as an important access technology having carved out a large niche
`in the computer market. Pen-based input is well suited to jotting down text
`and accessing information in mobile computing situations. “Notepads”
`made pen-based systems more popular a few years ago; however, not
`enough empirical tests have been performed to determine how we can
`improve their usage and efficiency. Goldberg and Richardson [1993], Mac-
`Kenzie et al. [1994], Venolia and Neiberg [1994], and MacKenzie and Zhang
`[1999] are a few exceptions.
`In small pen-based systems, accessing information by the selection of a
`target is more often attempted than by inputting handwritten data. Com-
`mon targets are menus, data (one character of the text or graphic segment,
`etc.), ranges etc., and the selection of keys on a software keyboard dis-
`played on a screen. As the amount of information displayed on the screen is
`increasing, users have to select smaller targets. The trade-off between the
`size and accessibility of targets and the amount of information presented on
`the screen is a fundamental problem in human-computer design. This is
`especially obvious in mobile products, such as personal digital assistants
`(PDAs), personal information managers (PIMs), and other mobile pen-
`based applications.
`In order to solve the problem, some leading studies have developed a
`variety of relatively efficient selection strategies for the touch-screen
`[Potter et al. 1988; Sears and Shneiderman 1991; Sears et al. 1992], the
`mouse [Kabbash and Buxton 1995; MacKenzie et al. 1991],1 and 3D input
`systems [Zhai et al. 1996]. Potter et al. [1988] conducted an empirical
`experiment to compare three selection strategies for touch-screens; how-
`ever, only one target size was used, and finger-movement-distance and
`finger-movement-direction were not considered. Sears and Shneiderman
`[1991] tested three selection devices; touch-screen, touch-screen with stabi-
`lization, and mouse. The task was the selection of rectangular targets of 1,
`4, 16, and 32 pixels per side. Their results showed that a stabilized
`touch-screen was effective for reducing the error rates when selecting a
`target. Kabbash and Buxton [1995] developed an area cursor which is
`larger than normal in order to improve target selection. Moreover, Worden
`
`1MacKenzie et al. [1991] also used a stylus but with an indirect tablet.
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`2
`
`

`

`386
`
`•
`
`X. Ren and S. Moriya
`
`et al. [1997] have provided a study of the effectiveness of two strategies for
`target selection: “area cursors” and “sticky icons.” Zhai et al.
`[1994]
`designed and demonstrated the effectiveness of the “silk cursor” which
`provided the volume/occlusion cues for target selection.
`However, current target selection strategies for pen-based systems are
`mostly only imitations of selection techniques for mouse and touch-screen
`devices. Investigations aimed at improving selection strategies for pen-
`based input devices have been neglected. This article looks at selection
`strategies suitable for selecting small targets, and identifies and quantifies
`the influential factors that make strategies more or less efficient with a
`view to improving selection performance on pen-based systems.
`This article is organized as follows. Section 2 introduces two interaction
`models for describing and designing 2D and 3D target selection strategies.
`It also describes and evaluates six strategies and six strategy groups which
`were tested in the experiments. Section 3 presents the experiment which
`determined the best individual strategy and the best strategy group. We
`explore the effect of target size, pen-movement-distance, and pen-move-
`ment-direction on the differences between selection strategies. We also
`investigate the relationships between interaction states, routes, and strat-
`egy efficiency. Section 4 presents another experiment for determining “the
`smallest maximum size,” i.e., the boundary value of the target size below
`which the degree of difficulty was significantly affected when selecting
`targets on pen-based systems. Section 5 gives a conclusion and directions
`for future research.
`
`2. CHARACTERISTICS OF SELECTION STRATEGIES
`
`2.1 State Transition Models for Selecting a Target with a Pen
`State transition models are very useful
`for describing and designing
`pointing/selecting interactions. Buxton [1990] suggested a state transition
`model to help characterize graphical input. However models for target
`selection have not been considered in detail. Chen [1993] proposed a state
`transition diagram for describing interactions with a target, but 3D targets
`have not been reported. Our models shown in Figure 1 and Figure 2 may
`expand and refine their research on target selection using a pen.
`
`2.2.1 A State Transition Model Describing Two-Dimensional Selection
`Strategies. Figure 1 shows a simple state transition model which eluci-
`dates a number of properties for selecting two-dimensional (2D) targets.
`This model can describe target selection not only on electromagnetic type
`tablets but also on touch-sensitive type tablets (touch-screens) which are
`used in general-purpose pen-based systems. The tip of the stylus pen
`interacts with the electromagnetic tablet so that it switches on when in
`contact with the screen surface. The pen switches off when the pen-tip is
`not in contact with the screen surface.
`The state transition model (Figure 1) shows an interaction with a 2D
`target. The model shows the target and the status and position of the
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`3
`
`

`

`Improving Selection Performance on Pen-Based Systems
`
`•
`
`387
`
`Fig. 1. A state transition model describing 2D target selection with a stylus pen. The ellipses
`illustrate 2D targets. The line arrows show the transition between two states which may be in
`either direction. The short lines under the pen-tip (in b and c) show the pen-tip in contact with
`the screen (the pen is switched on by contact with the screen). State a: pen outside/above the
`2D plane, pen-tip switched off (pen not in contact with the screen); state b: outside the target,
`switched on (pen in contact with the screen); state c: inside the target, switched on (pen in
`contact with the screen). If we assume for example that state a is an initial state and c is a
`final state, the state transition route may be either a 3 b 3 c or a 3 c.
`
`pen-tip. The ellipses represent 2D targets on the screen. The line arrows
`show the transition between two states. The short lines under the pen-tip
`show that the pen-tip is in contact with the screen (the pen is switched on).
`State a shows the pen outside/above the 2D plane, pen not in contact with
`the screen (the pen-tip switched off). State b shows the pen in contact with
`the screen (and therefore switched on) but outside the target area. State c
`represents the pen in contact with the screen (therefore switched on) inside
`the target. In state a the pen is approaching the 2D screen surface from
`above, in 3D space. In states b and c the pen is in contact with the screen
`(the pen is dragged over the 2D plane). Thus there are three states: state a:
`outside/above the 2D plane, not in contact with the screen (switched off);
`state b: outside the target, in contact with the screen (switched on); state c:
`inside the target, in contact with the screen (switched on).
`2.2.2 A State Transition Model Describing Three-Dimensional Selection
`Strategies. Figure 2 shows a state transition model which elucidates a
`number of properties for selecting three-dimensional (3D) targets. We used
`an electromagnetic tablet in the experiments. This type of tablet also
`allowed us to trial 3D selection strategies, because when the pen-tip is
`above the tablet screen surface (within a height of 1 cm), the computer can
`recognize the coordinates (x, y) of the pen-tip. Thus, even though the
`bottom of a target (e.g., a menu or a button) on the screen is 2D, it can be
`highlighted or selected when the pen is above the tablet surface (within 1 cm).
`This means that the target can also be expressed as a 3D target.
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`4
`
`

`

`388
`
`•
`
`X. Ren and S. Moriya
`
`Fig. 2. A state transition model describing 3D target selection with a stylus pen. The
`cylinders with dashed lines show the body of a 3D target. The ellipses with a solid line
`illustrate the bottom of 3D targets on the tablet screen surface. The short lines under the
`pen-tip (in e and f) show the pen-tip in contact with the tablet surface. State d: the pen-tip is
`outside the 3D target, pen-tip switched off (pen not contact with the screen); state e: the
`pen-tip is outside the 3D target, switched on (the pen is in contact with the screen); state f: the
`pen-tip is inside the 3D target, switched on (pen in contact with the screen); state g: inside the
`3D target but not in contact with the screen and therefore switched off.
`
`The state transition model in Figure 2 showing an interaction with a 3D
`target consists of the target and the status and position of the pen-tip. The
`ellipses with a solid line illustrate the bottom of the 3D targets on the
`screen surface. The cylinders show the body of the 3D target. Some
`responses (e.g., highlighting) will take place when the pen is in the cylinder
`even though the pen-tip is not in contact with the screen surface. The short
`lines under the pen-tip show that the pen-tip is in contact with the screen
`surface. States d and e represent the pen outside the target. State f and
`state g represent the pen inside the target. States e and f represent the pen
`in contact with the screen surface (the pen is dragged over the 2D plane).
`States d and g represent the pen as not in contact with the screen surface.
`In this model we considered the two pen positions above and beside the 3D
`target as the same in effect. There may, however, be some design value in
`considering the implied approach paths as offering different selection
`options. Thus there are four states: state d: pen not in contact with the
`screen, outside the target (before or after entering the 3D target sensitive
`zone); state e: in contact with the screen surface, outside the target; state f:
`in contact with the screen, inside the target; and state g: approach or
`removal from the 2D plane inside the 3D target sensitive area (3D cylin-
`der).
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`5
`
`

`

`Improving Selection Performance on Pen-Based Systems
`
`•
`
`389
`
`It should be noted that although the illustrations show ellipses in Figure
`1 and cylindrical targets in Figure 2, the shape of the target has no
`definitive bearing on this discussion. Our focus here is on the description of
`selection strategies using the state transition models. For example, we
`shall describe 2D target selection strategies using the state transition
`model in Figure 1. Here assume state a is an initial state, and state c is a
`final state. The states and transitions used to select a target can be
`expressed as a 3 b 3 c or a 3 c. In other words, new strategies may
`be designed using these state transition models. The state transition
`models representing the manipulation of a pen from an arbitrary initial
`state to an arbitrary final state can become a strategy. For example, in the
`“Slide Touch” strategy the initial state is a, and the final state is c (see
`Section 2.2); in the “Direct Off” strategy, the initial state is a, and the final
`state is a (see Section 2.2). Theoretically, an infinite range of selection
`strategies exists.
`We consider that these states are an adequate basis for 2D (a, b, and c in
`Figure 1) and 3D target design (d, e, f, and g in Figure 2) because they
`include all the normal conditions for these types of pen-based systems (pen
`in contact with the screen, pen not in contact with the screen; pen switched
`on, pen switched off; pen inside the target area, and pen outside the target
`area). Furthermore, the models may be modified to include other conditions
`such as pen side switches.
`
`2.2 Six Strategies Used in Two Experiments
`The six strategies for selecting a target in the two experiments (see
`Sections 3 and 4) are illustrated in Figure 3. The arrows show the direction
`of pen-tip movement. The dashed lines indicate that the pen-tip is not in
`contact with the screen surface (either before or after contact), and the
`solid lines (in Slide Touch, Direct Off, and Slide Off) show that the pen-tip
`is in contact with the screen surface. The pen-tip is automatically switched
`on by contact with the screen surface. The dark points show where target
`selection is affected in the strategy process. Here, we explain the six
`strategies and show how these strategies fit into the state transition
`models (Figures 1 and 2). Assume I is a collection of initial states, described
`as I ⫽ 兵其; F is a collection described as F ⫽ 兵其; M is a collection of middle
`states, described as M ⫽ 兵其; R is a collection of routes, described as R ⫽ 兵其.
`An arrow “3” means that a state changes to another state. “7” means the
`changes between two states may be in either direction. Table I shows initial
`states 共I 兲, middle states 共M 兲, final states 共F 兲, and routes 共R兲 for the six
`strategies.
`
`—Direct On strategy: the pen approaches from above. The target is selected
`only momentarily at the time the pen makes contact with the screen in
`the target area. Here, I 共Initial state兲 ⫽ 兵a其, F 共Final state兲 ⫽ 兵c其, R
`共Route兲 ⫽ 兵a 3 c其, and there is no middle state 共M 兲 (see Figures 1 and 3).
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`6
`
`

`

`390
`
`•
`
`X. Ren and S. Moriya
`
`Fig. 3. This figure represents the six strategies (Direct On,, Slide Touch, Direct Off, Slide
`Off, Space On, and Space Touch described according to the sate transition models used in the
`two experiments. The figure also shows the strategies (On, Off, 2D, 3D, In, and In-Out) as they
`grouped according to their characteristics (see Section 2.3). The In strategies (on the left) are
`duplicated (center column) to indicate that they are functional possibilities within the In-Out
`strategies to which they correspond and with which they constitute a group (2D On or Off or
`3D On). The figure shows only the simplest representation of each route and does not include
`possible repeated steps. Im many routes the initial and/or middle steps may be repeated any
`number of times before selection is affected, e.g., in the Space On strategy the figure shows d
`3 g 3 f, but this could be represented as d 7 g 3 f (e.g., d 3 g 3 d 3 g 3 f )
`because the repeated step does not affect the selection of the target though it may affect the
`highlighting function.
`
`—Slide Touch strategy is an extension of the Direct On strategy. Here also
`the target is selected when the pen touches it for the first time, but in
`this case the pen initially lands outside the target area before moving
`into it. Here, I ⫽ 兵a其, F ⫽ 兵c其, M ⫽ 兵b其, R ⫽ 兵a 3 c, a 3 b 3 c其.
`—Direct Off strategy: the target is highlighted only while the pen is
`touching it. The selection is made at the moment the pen is taken off the
`target. Here, I ⫽ 兵a其, F ⫽ 兵a其, M ⫽ 兵b, c其, and R ⫽ 兵a 3 c 3 a, a 3 b
`7 c 3 a其.
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`7
`
`

`

`Improving Selection Performance on Pen-Based Systems
`
`•
`
`391
`
`Table I.
`
`Initial States, Middle States, Final States, and Routes of the Six Strategies
`
`Strategies
`
`Direct On
`
`Slide Touch
`
`Direct Off
`
`Slide Off
`
`Space On
`
`Space Touch
`
`Initial
`States
`
`Middle
`States
`
`Final
`States
`
`Routes
`
`Numbers of
`Routes
`
`a
`
`a
`
`a
`
`a
`
`d
`
`d
`
`(no)
`
`b
`
`b, c
`
`b, c
`
`g
`
`d, g
`
`c
`
`c
`
`a
`
`a
`
`f
`
`a 3 c
`
`a 3 c
`a 3 b 3 c
`
`a 3 c 3 a
`a 3 b 7 c 3 a
`
`a 3 c 3 a
`a 3 b 7 c 3 a
`a 3 c 7 b 3 a
`a 3 b 7 c 7 b 3 a
`
`d 7 g 3 f
`
`e, f
`
`d 7 g 3 f
`d 7 g 3 d 3 e
`
`1
`
`2
`
`2
`
`4
`
`1
`
`2
`
`—Slide Off strategy is an extension of the Direct Off strategy. The target is
`highlighted only while the pen is in contact with it; however, the
`selection is made when the pen is removed from any point on the screen
`either inside or outside the target area. Here, I ⫽ 兵a其, F ⫽ 兵a其, M ⫽
`兵b, c其, R ⫽ 兵a 3 c 3 a, a 3 b 7 c 3 a, a 3 c 7 b 3 a, a 3 b 7
`c 7 b 3 a其.
`
`—Space On strategy: the pen approaches from above. The target is high-
`lighted while the pen is within the 1 cm high cylinder above the target.
`Selection is made at the moment the pen makes contact with the bottom
`of the target area (i.e., inside the bottom circle). Here, I ⫽ 兵d其, F ⫽ 兵f 其,
`M ⫽ 兵g其, R ⫽ 兵d 7 g 3 f 其 (see Figures 2 and 3).
`
`—Space Touch strategy is an extension of the Space On strategy. The
`target is highlighted while the pen is within the 1 cm high cylinder above
`the target. After highlighting, the selection is made when the pen makes
`contact with any point on the screen either inside or outside the target area.
`Here, I ⫽ 兵d其, F ⫽ 兵e, f 其, M ⫽ 兵d, g其, R ⫽ 兵d 7 g 3 f, d 7 g 3 d 3 e其.
`These strategies may be considered to be “strategy types” rather than
`fixed strategies. This means that the name of the strategy specifically
`indicates the point of actual selection and sometimes includes information
`about the route.
`The Direct On and Direct Off strategies are already in common use. The
`Slide Touch strategy corresponds to the “first-contact” strategy [Potter et
`al. 1988]. The Slide Off, Space On, and Space Touch strategies were new
`strategies designed by Ren and Moriya [1997a].
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`8
`
`

`

`392
`
`•
`
`X. Ren and S. Moriya
`
`2.3 Classification of the Six Strategies
`Although there are a few studies on selection strategies, none have paid
`attention to particular strategy characteristics until now. We classified
`(grouped) the six strategies so that we could evaluate particular character-
`istics which pertain to selection strategies in general. Figure 3, therefore,
`shows the classifications of the six strategies according to their character-
`istics, as well as the six individual strategies. Notice that strategies may
`appear in more than one group depending on the various combinations of
`characteristics pertaining to them. The classifications were formulated
`after consideration of the six conditions created by the pen manipulations
`[Ren and Moriya 1995]. These conditions are: contact with the screen,
`removal from the screen, contact inside the target, contact outside the
`target, target highlighted, and target not highlighted.
`—2D and 3D strategy groups: Targets exist both as planes (2D) and as solid
`bodies (3D). Here, the 2D strategies are the Direct On, Slide Touch,
`Direct Off, and Slide Off strategies. The 3D strategies are the Space On
`and Space Touch strategies.
`—On and Off strategy groups: Contact and removal of the pen from the
`screen were considered as movements between the 2D plane and 3D
`space. Pen contact involves a movement from 3D space to the 2D plane,
`while removal involves a movement from the 2D plane to 3D space. These
`interactions were considered to be suitable conditions for the subject to
`recognize and confirm the moment of target selection. The strategies in
`which selection was made by contact with the screen (Direct On, Slide
`Touch, Space On, and Space Touch strategies) were named On strategies.
`The strategies in which selection was made by removal from the screen
`(Direct Off and Slide Off strategies) were named Off strategies. Where
`the target existed on the 2D plane, both the On and Off strategies were
`deployed. Where the target existed in 3D space, considering that the
`pen-tip is approaching the body of the 3D target from above in general
`when selecting a 3D target, we here only discuss the On strategies (Space
`On and Space Touch strategies).
`—In and In-Out strategy groups: We considered the movement of the pen
`into and out of the target from the perspective of the user’s eyes. When
`the pen moved into or out of the target, users could confirm whether or
`not the target was highlighted. Those strategies in which selection was
`made by contact within the target area were named In strategies (the
`Direct On, Direct Off, and Space On strategies). On the other hand, those
`strategies in which selection was made by contact either inside or outside
`the target were named In-Out strategies (Slide Touch, Slide Off, and
`Space Touch).
`
`3. EXPERIMENT ONE
`This section presents a comparison of the six strategies individually and
`the strategies grouped. We seek to determine the best individual strategy
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`9
`
`

`

`Improving Selection Performance on Pen-Based Systems
`
`•
`
`393
`
`and the best strategy group. We also evaluate the effects variables have on
`the differences between the strategies. If the significance (or insignificance)
`of the differences between strategies is maintained when variables such as
`direction, distance, or target size are changed, we may consider that the
`particular variable has no significant influence on the efficiency of the
`strategies in general. Conversely if the significance (or insignificance) of
`the differences between strategies is not maintained when variables such
`as direction, distance, or target size are changed, we may consider that the
`particular variable has a significant influence on the efficiency of the
`strategies in general. For this to be conclusive it was obviously necessary to
`test the variables in a comprehensive and balanced way.
`
`3.1 Method
`
`3.1.1 Participants. Twenty-one volunteer participants (17 male, 4 fe-
`male; all right-handed, university students) were tested for the experiment.
`Ten had had previous experience with pen-input systems, while the others
`had no experience.
`
`3.1.2 Apparatus. The experiment was run on an NEC 9801DA PC and a
`Wacom tablet-cum-display with a stylus pen. The liquid crystal display
`resolution was 640 ⫻ 400 pixels. One pixel was about 0.36 mm. The
`pen/screen contact area was 1.40 mm in diameter.
`
`3.1.3 Procedure. First the experiment was explained to each subject.
`Each subject had 20 practice trials immediately before the experiment
`started. The message “Select a target as quickly and accurately as possible
`using the strategy” was displayed on the screen of the experimental tool
`when the experiment started.
`The steps for selecting a target were as follows (Figure 4):
`
`(a) The initial position was displayed at the center of the screen. The
`initial position was the place where the pen was pointed immediately
`before beginning the selection procedure. The subject had been told
`which strategy to use and how many trials were to be done.
`
`(b) The subject touched the initial position with the pen.
`
`(c) A target was displayed with size and position changed at random by the
`computer. Targets of a particular size were never displayed in the same
`position twice. The distances between the initial position and the target
`were 39, 131, or 160 pixels, randomly selected by the computer.
`
`(d) The subject attempted to select the target and then received a message
`on the screen to indicate whether or not he or she had made a
`successful selection.
`
`(e) The subject then repeated (a) to (d) above.
`
`(f) A message indicating the end of the test was displayed when the
`subject had completed the task.
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`10
`
`

`

`394
`
`•
`
`X. Ren and S. Moriya
`
`Fig. 4. The display of a target. The small, centered dot is the initial position. The large circle
`shows one of the 24 possible positions for the display of a target. The circular dotted lines
`show the three pen-movement-distances from the initial position to the target. The solid lines
`indicate the eight pen-movement-directions to the target from the initial position.
`
`After they finished testing each strategy, the subjects were asked to fill
`in a questionnaire. The first question was: “For the strategy tested just
`now, when selecting T, how do you rate Q? Please answer on a 1-to-5 scale
`(1 2 3 4 5).” Here, 1 ⫽ lowest preference, and 5 ⫽ highest preference. “T”
`means large or small targets as tested in the particular trial. “Q” consisted
`of the six subquestions regarding selection accuracy, selection speed, selec-
`tion ease, learning ease, satisfaction, and desire to use. The second ques-
`tion was: “Which positions (i.e., direction and distances) were most comfort-
`able for selecting the targets in the strategy?” The subject marked his/her
`preferences on Figure 4.
`The strategies were not mixed. In a given trial each subject used only one
`strategy. The data for each strategy were recorded automatically as follows:
`(1) Presence or absence of error when a target was selected. One selection
`was a continuous operation from the moment the pen touched the
`initial position until removal of the pen from the screen surface.
`Feedback to the subject indicated whether the selection was successful
`or not. In either case, the subject could not cancel the selection.
`(2) Position and size of the target displayed.
`(3) The time lapsed between display of the target and the moment when
`the pen contacted the screen.
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`11
`
`

`

`Improving Selection Performance on Pen-Based Systems
`
`•
`
`395
`
`Fig. 5. Mean selection times (with standard error bars) for each individual strategy in
`Experiment One.
`
`(4) The time lapsed between contact with the target and removal from the
`screen.
`(5) The time lapsed between contact with the screen and contact with the
`target.
`These times were measured to an accuracy of 10 ms. Data as defined in
`item (3) were recorded for the Direct On, Space On, and Space Touch
`strategies. Data as defined in item (5) were recorded for the Slide Touch
`strategy. Data as defined in item (4) were recorded for the Direct Off and
`Slide Off strategies.
`3.1.4 Design. The experiment used a mixed factorial design.
`—Size of the target: To examine the relationship between target size and
`strategy, three target sizes of 3, 5, and 9 pixels (1.1 mm, 1.8 mm, and 3.2
`mm diameter circles) were used in all trials. All the targets for the
`experiment were circular. Circular targets were used so that the distance
`between the initial position and the edge of all targets on each radius
`remained constant in all directions.
`—Pen-movement-distance: The distance to the target was the radius of a
`circle in which the center point was the initial position (Figure 4). To
`examine the relationship between distance and strategy efficiency, the
`
`ACM Transactions on Computer-Human Interaction, Vol. 7, No. 3, September 2000.
`
`12
`
`

`

`396
`
`•
`
`X. Ren and S. Moriya
`
`distances of 39, 131, and 160 pixels (14.0, 47.2, and 57.6 mm) were
`determined by a preliminary experiment. (Distances of 39 pixels and 131
`pixels were the average values used by 10 subjects in a preliminary
`experiment. When the wrist was in a fixed condition, 39 pixels was the
`radius of the arc which could be drawn by the subjects; 131 pixels was
`the radius of the circular arc which was the maximum finger-movement-
`distance. The outside circle radius of 160 pixels was determined accord-
`ing to the size limitations (height) of the screen. It was also a distance by
`which the wrist could be moved.).
`
`—Pen-movement-direction: Eight directions were used. They were at 0, 45,
`90, 135, 180, 225, 270, and 315 degrees from the initial position (Figure
`4).
`Each subject had a total of 92 trials for each strategy. These consisted of
`20 practice trials and 72 test trials (⫽ 3 target sizes ⫻ 3 distances ⫻ 8
`directions). A break was taken at the end of each strategy trial. Whenever
`the subject felt tired he or she was allowed to take a rest. Each subject
`completed 432 test trials (⫽ 6 strategies ⫻ 72). In each strategy 1512 test
`trials (⫽ 21 subjects ⫻ 72) were completed. The order for the six strategies
`was different for each of the 21 subjects.
`
`3.2 Results
`An ANOVA (analysis of variance) with repeated measures was used to
`analyze performances in terms of selection times, error rates, and subjec-
`tive preferences. Post hoc analysis was performed with Tukey’s honestly
`significant difference (HSD) test.
`
`3.2.1 Comparison of Selection Times for the Individual Strategies.
`There was a significant interaction between the six individual strategies in
`selection time, F共5,120兲 ⫽ 10.8, p ⬍ 0.0001. From this we could con-
`clude that the selection time was influenced by the particular strategy, i.e.,
`the selection times changed according to the strategy being applied. Figure
`5 shows the average selection times for each of the six strategies. The Slide
`Touch strategy was the fastest among the six strategies 共mean ⫽ 0.98s兲.
`The post hoc Tukey HSD test showed that the Slide Touch strategy was
`faster than the Direct Off, Slide Off, Space On, Space Touch strategies 共p
`⬍ 0.05兲. There was no significant difference in selecti