Target Size and Distance: Important Factors for Designing User Interfaces for Older Notebook Users

Transcription

1 Work with Computing Systems H.M. Khalid, M.G. Helander, A.W. Yeo (Editors). Kuala Lumpur: Damai Sciences. 454 Target Size and Distance: Important Factors for Designing User Interfaces for Older Notebook Users Claudia Armbrüster, Christine Sutter, and Martina Ziefle Department of Psychology, RWTH Aachen University, Germany. Abstract. The present study addresses the design of user interfaces for older users and the usability of input devices in notebooks touchpad and trackpoint. The central question this study aims to answer is whether Fitts Law - the interaction of target size and distance affecting movement time - can predict the motor performance for notebook input devices on the one hand and older users - over 40 years - on the other hand. 28 users completed different cursor-control tasks (selection and manipulation) and the influence of task difficulty (two target sizes and distances) on performance in both input devices was examined. Keywords. Input devices, Fitts law, Older users, User interface design. 1. Introduction The ergonomic research concentrating on the usability of input devices exhibits a very long and sophisticated tradition. The topic of the relevant studies focused on very different usability aspects with respect to the quality and the ease of use of input devices, including different types of input devices (direct and indirect, e.g. Epps, Muto and Snyder, 1986), different cursorcontrol actions (object selection and object manipulation tasks, e.g. Sutter and Ziefle, 2003) and different performance measures (speed and accuracy of cursor control). As a theoretical base for usability evaluation, Fitts Law (1954) can be regarded as a standard paradigm for evaluation purposes (e.g. Card, English and Burr, 1978; Tränkle and Deutschmann, 1991; Douglas and Mithal, 1994; MacKenzie, 1992). Fitts Law, a rather robust model of human psychomotor behavior, predicts the movement time to be a loglinear function of target size and target distance. The prediction is based on rapid aiming hand movements. Mathematically, Fitts' law is stated as follows: MT = a + b log 2 (2A/W), with M = movement time a, b = regression coefficients A = distance of movement from start to target center W = width of the target The parameter log 2 (2A/W) - also labeled as index of difficulty (ID) - defines the ratio of target distance to target size. The values for the intercept a and the slope b are empirical variables determined by regression analysis. The slope of the function mirrors the increase of movement times depending on the ID, the intercept describes the minimum of movement time necessary to execute the task. According to Fitt s model, larger and closer targets are more easily to achieve, therefore faster than smaller and farther targets. In spite of the multitude of studies showing that movement times with input devices follow Fitts Law (e.g. Accot and Zhai, 2003; MacKenzie, Sellen and Buxton., 1991; Douglas et al., 1994), only few studies have concentrated on the validity of Fitts Law for notebook input devices (Batra, Dykstra, Hsu, Radle and Wiedenbeck, (1998); Sutter and Ziefle, submitted). As Fitts Law is based on the execution of real movements, with the hand, finger or a pointing device, it is of high importance to determine if Fitts Law is also valid for input devices where no movement is necessary to control the cursor. The latter is crucial especially for the trackpoint, a small isometric joystick. Here, the finger force applied to the device results in a cursor movement on the display. Therefore the consistency between finger movements and cursor movements is low. Another ergonomic concern is the usability of notebook input devices for older users. Up to now, the knowledge respecting the motor performance and the consideration of user centered ergonomic design for older users is fairly low, when using notebook input devices. Only one single study was found examining the usability of input devices in older users (Göbel, Backhaus and Krüger, 2000). Six input devices (keyboard, light pointer, trackball, graphic tablet, computer mouse and touchpad) were tested in 60 seniors (55 90 years) as well as 20 younger adults (25 40 years) using a simple selection task. The results yielded a superiority of input devices with a high movement analogy (e.g. light pointer), whereas input devices, which are primarily based on finger movements (e.g. touchpad, trackball) performed worst. Unfortunately, only a rather simple task was applied for a very short period of time (3 minutes) in which the implications of Fitts Law were not considered. But an ongoing aging process characterizes the working population in Western industrial countries, with users becoming older and older (e.g. Hayslip and Panek, 1989, Köchling and Spannhake, 1997). The decrease in sensory as well as motor performance in older adults is well known (e.g. Smith, Sharit and Czaja, 1999; Welford, 1984). With increasing age reaction time and movement time slow down as

2 Work with Computing Systems H.M. Khalid, M.G. Helander, A.W. Yeo (Editors). Kuala Lumpur: Damai Sciences. 455 much as 200 (Stelmach, Goggin and Gracia-Colera, 1987). In addition, the speed-accuracy-trade-off becomes more distinct with age (Welford, 1976). Further to the physiological and psychological changes, older users are not accustomed to modern computer hardware and their user interfaces. Therefore, this study features the importance of an adequate design for older -user interfaces. 2. Method The two input devices were examined regarding speed and accuracy of cursor control in an object selection and an object manipulation task. An independent experimental design was used Experimental variables Independent variables were the type of input device (touchpad vs. trackpoint) and the task difficulty according to Fitts. Two target sizes (2.5x2.5 mm vs. 5x5 mm) and two distances from the starting point (25 mm vs. 50 mm) were varied. Thus, different task difficulties were created: easy (big and near), medium (big and far, small and near), difficult (small and far). Table 1 shows the indexes of difficulty for the different combinations according to Fitts Law modified by MacKenzie (1989) (MT = a + b [log 2 (2A/W + 1)]. Table 1. Indexes of difficulty (IDs) Indexes of difficulty (bit) target distance Target size (mm) (mm) 5 x x (1) 3.46 (3) (2) 4.39 (4) The combination of the two sizes and two distances results in four IDs, however two values are equal: according to Fitts Law the two medium difficulties (2 and 3) should lead to the similar motor performance. In order to separate the potentially different influences of the two factors (size and distance), the four IDs are used in later calculations. To distinguish them, the notations in brackets will be used. Dependent variables were the different measures of motor performance as specified in the following sections Measures for speed The total time was defined as the interval between the start (pressing the space bar) and the successful completion of the task. This period was fragmented in two components: the homing time - the interval between pressing the space bar until an initial contact with the input device - and the movement time - the time between the first contact and the successful completion of the task. In considerations of Fitts Law only the movement time (MT) is important and will be used for calculation purposes Measures for accuracy Due to the different difficulties of the two tasks - the object selection is quite easy, whereas the manipulation task is rather difficult as several parts have to be completed consecutively - therefore different types of errors have to be differentiated. In the selection task click errors were counted. Click errors occur in the manipulation task as well. However, three additional types of errors were registered: a highlight error, a drag error and a drop error. A click error occurred when subjects pressed the mouse button while the cursor was outside the boundaries of the square target. Highlight errors cover all errors related to the highlighting procedure (inaccurately clicking, releasing the mouse button too early or too late). A drag error was registered when the object was not grabbed correctly. Finally a drop error emerged when the object got lost during the movement or was not dropped into the target box but outside the target boundaries Experimental tasks In the object selection, the participants had to select a target object by moving the cross haired cursor from a starting position to the square and clicking inside its boundaries. If participants hit correctly, the black target changed into green providing visual feedback This task was chosen because of its analogousness to real life applications as depicted in Figure 1 the black arrow on the right hand side symbolizes the cursor way. (target) + (start) Figure 1. The selection task and its application The task started with a self-paced press of the space bar. To exclude confounding effects of different movement directions the targets appeared in eight directions around the starting point: 45º, 90º, 135º, 180º, 225º, 270º, 315º and 360º. In total, 640 single trials had to be completed by the participants. Figure 1 illustrates the task type (left) and the underlying application in the World Wide Web (right). The second task was a text manipulation task consisting of several single actions to be executed one after another: an object had to be selected, highlighted, dragged and dropped. Each trial began by pressing the space bar. A string out of geometrical forms appeared on the screen (Figure 2). The cursor had to be moved from the start to one of the grey colored fields. One of them had to be clicked on with the left mouse button in order to start the highlighting procedure. (target) + (start) Figure 2. The manipulation task and its application Then the cursor had to be moved into the other grey area where the mouse button had to be released. The

3 Work with Computing Systems H.M. Khalid, M.G. Helander, A.W. Yeo (Editors). Kuala Lumpur: Damai Sciences. 456 successful highlighting of the string was confirmed by a visual feedback - the string color changed to green. In the next step, the highlighted object had to be dragged by clicking onto the highlighted area and moving the object to the target box with the left mouse key pressed (drag and drop). The task was finished when the color of the target box changed to green (indicating a successful procedure) and the pressed mouse button was released. Then, the target disappeared and the next string appeared. Similar to the selection task the target box (2.5x 2.5 mm or 5x5 mm) appeared in eight directions (45º, 90º, 135º, 180º, 225º, 270º, 315º, 360º) around the starting point, excluding possible effects of movement direction. Again this task was chosen in analogy to real computing life and represented a common task in word processing (right hand side in Figure 2) Apparatus and materials The touchpad is a contact sensitive rectangular tablet, which is embedded into the wrist rest of notebooks beneath the keyboard. Its size is about 4.5 cm by 6.5 cm and the two mouse buttons are located underneath the pad. Finger movements on the touchpad are transferred to corresponding cursor movements. Thus, a high consistency between finger movements and the subsequent and collateral cursor movement is given. The trackpoint is a small isometric joystick placed between the letter keys G, H and B in the keyboard with two mouse buttons beneath the space bar in the wrist rest. The signal transmission in isometric joysticks (in contrast to analogous or digital ones) is proceeded by the finger force imposed upon the device. The resulting cursor movements are proportional to the strength and direction of the respective finger force. Here the consistency between finger movements and cursor movements is low. Two notebooks with the internal input devices described above, a Dell Inspiron 7500 with a touchpad and a Toshiba Satellite with a trackpoint were used in this study. As both input devices are technologically run in a rather different manner (motion- versus force controlled), it was a central methodological requirement that the cursor velocity was equal for both devices. Preliminary tests focused on a matching procedure. The resulting cursor velocity was successfully matched for both devices at 1500 pixels/sec ( medium speed). Finally, in order to control the visual quality of the screens in the different notebooks, both computers were connected to the same TFT screen (Philips 150x) with a screen resolution of 1024x768 pixels Procedure The factor input device was treated as a between - group factor. Two different tasks were completed by 7 participants each using the trackpoint or the touchpad. It was emphasized that the experiment assessed the usability of notebook input devices and participants should not be afraid of personal failure. Further, they were instructed to complete the tasks quickly and accurately. The completion of the 640 trials took between 45 minutes and three hours, depending on the type of input device, the task and the individual working speed. Four practice trials were given in the beginning of the experiment to become familiar with the experimental situation and the task Participants 28 persons, 11 males and 17 females, between 40 and 61 years of age participated in the experiment. Two of them were lefthanded - this was considered by adjusting the mouse buttons. In order to have a valid representation of the working force in modern society, participants were employees in different professions with a wide range of educational levels. The participation in the study was voluntary. Regarding the usage of portable computers all subjects were novices, with no experience in the use of notebook input devices. 3. Results To answer the question if Fitts Law applies to the notebook input devices ANOVAs and regression analyses were calculated (levels of significance: p < 5 = * and p < 1 = **). For all calculations error-free movement times were used to avoid adulteration, so 5.86 of the trials in the selection task and in the manipulation task were disregarded. To provide a single view of all data, Figure 3 shows movement times (given in milliseconds) for each of the input devices, tasks and task difficulties separately MT [msec] TOUCHPAD SELECTION TRACKPOINT TOUCHPAD TRACKPOINT MANIPULATION big-near big-far small-near small-far Figure 3. MT for both input devices and tasks regarding task difficulty A 2 x 2 x 4 ANOVA reveals the highly significant differences between the input devices (F (1,24) = 7.87; p<.01) and the tasks (F (1,24) = ; p<.01) the two betweengroup factors - as well as the difference between the task difficulties (F (1,24) = ; p<.01) the within-group factor. (Figure 3). First of all - providing a deeper insight into Fitts Law and its implications - one-way ANOVAs show that movement times (listed in Table 2) are significantly affected by target size and target distance in both tasks independently from the input device. The farther and smaller the targets are the higher the movement times and vice versa. As shown in Table 2, trials with big targets were solved 27 faster in the selection task (F (1,26) = 6.097; p<.05) and 19 faster in the manipulation task (F (1,26) =9.669; p<.05). Similar results were found for the target distance: near targets were hit 19 faster in the object selection (F (1,26) = 2.795; p= 0.107) and 13 faster in the manipulation task (F (1,26) = 4.549; p<.05). Regarding Fitts Law the results show significant effects of the task difficulty, in both tasks and with both input devices. In the selection task easy trials (big and near) were

4 Work with Computing Systems H.M. Khalid, M.G. Helander, A.W. Yeo (Editors). Kuala Lumpur: Damai Sciences. 457 solved 25 faster than medium trials (big and far, small and near) and 41 faster than difficult (small and far) trials (F (1,39) = 7.952; p<.01). In the object manipulation, easy trials were 16 faster compared to trials with a medium difficulty and 29 faster compared to difficult trials (F (1,39) = ; p<.01). Consequently, Fitts Law is valid and proves to be a good predictor, not only for the performance in different notebook input devices, but also for the motor performance of older adults. Table 2. Mean values and standard deviations in msec Movement times (msec) selection (n=14) manipulation (n=14) target size (TS) M Sd M Sd small big target distance (TD) near far A closer examination of the movement times in relation to the index of difficulty under the terms of MacKenzie (Table 1) and the performance differences between the two input devices, reveal that a difference exists between trials with the two medium difficulties. In fact these should theoretically not differ. As depicted in Table 3, the mean values (M) and the standard deviations (Sd) are listed for the total group and the two subgroups touchpad and trackpoint users independent from the task type. Table 3. Mean values and standard deviations in msec Movement times (msec) total touchpad trackpoint I (N=28) (n=14) (n=14) D M Sd M Sd M Sd The differences between ID 2 and 3 average out at 17 (t (27) = -5.3; p<.01) in the total group, 13 (t (13) = -3.4; p<.01) within the touchpad users and 21 (t (13) = -4.2; p<.01) within the trackpoint users. The specific impact of target size and target distance on movement time was analyzed separately by a stepwise regression analysis. Target size explains about 61 of variance in movement time by ID and target distance another but only 33. So 94 of the explained accounted variance can be traced back to target size and target distance with significant Pearson correlations about r (MT/TS) = ** and r (MT/TD) = 0.569**. The Pearson correlation between movement time and ID reaches r (MT/ID) = 0.993**. In addition, regression analysis of movement times and the index of difficulty for touchpad and trackpoint users and the total group were calculated. The following equations resulted from the analysis: R total = log 2 (A+W)/W R touchpad = log 2 (A+W)/W R trackpoint = log 2 (A+W)/W Standard errors (SE) belonging to the three intercepts and slopes are depicted in Table 4. Table 4. Regression coefficients: intercepts, slopes and standard errors for the three groups Regression coefficients intercept SE slope SE total (N=28) touchpad (n=14) trackpoint (n=14) On the basis of the regression equations the cursor control in older adults using notebook input devices can be exactly predicted. The graphic illustration of the equations (Figure 4) reveals that the slopes from touchpad and trackpoint differ by 310 msec. This means that the older users are dramatically slower in their motor performance when they have to work with a trackpoint and small - far away targets. movement time in msec index of difficulty in bit 4 MT trackpoint total MT MT touchpad Figure 4. Mean MT as a function of movement difficulty (including the performance in both tasks) Now the different tasks, the object selection and the object manipulation, are considered separately depending on the two input devices. The regression equations for the two tasks selection and manipulation and input devices touchpad and trackpoint are as follows: R selection = log 2 (A+W)/W R manipulation = log 2 (A+W)/W R selection-touchpad = log 2 (A+W)/W R selection-trackpoint = log 2 (A+W)/W = log 2 (A+W)/W R manipulation-touchpad R manipulation-trackpoint = log 2 (A+W)/W Figure 5 reveals that there are indeed performance differences between different tasks, task IDs and input devices. Comparing the slopes with each other, the differences in the

5 Work with Computing Systems H.M. Khalid, M.G. Helander, A.W. Yeo (Editors). Kuala Lumpur: Damai Sciences. 458 simple selection task are rather low. But enormous differences appear in the manipulation task especially when using the trackpoint. movement time in msec MT-manipulation trackpoint MT-manipulation total MT-manipulation touchpad MT-selection trackpoint MT-selection total MT-selection touchpad index of difficulty in bit Figure 5. Mean MT as a function of movement difficulty (including the performance in both tasks) In Table 5 the predicted movement times (according to the modified equation from MacKenzie) are confronted with the real mean values out of the sample of this study. In all conditions the predicted values do not match with the observed ones. The columns called show the difference in percent. Overestimations occur in the selection task to 26 in the easiest trials (ID 1) when using the trackpoint, and also the motor performance with the touchpad is overestimated (23 ). Table 5. Movement times: comparison of predicted and observed values in both task with touchpad and trackpoint Movement times (msec) I D selection-touchpad r (MT/ID) =0.964* selection-trackpoint r (MT/ID) =0.998** I D manipulation-touchpad r (MT/ID) =0.963* manipulation-trackpoint r (MT/ID) =0.998** Only when the targets are small and far away (ID 4), the predictions are correct in the selection task. In the more complex object manipulation, the findings are rather different. Fitts Law predictions are valid when rather easy motor tasks had to be completed (ID 1, big and near). Even in more difficult trials (ID 2), when big sized objects had to be moved and dropped in far away target boxes, Fitts assumptions match the motor performance accurately. Whenever the object is small, independently of input device and target distance (ID 3 and 4), movement times increase disproportionately to Fitts predictions (underestimation by 15 ). Apparently Fitts Law shows strengths and weaknesses in movement time predictions. The question is whether target size and target distance possibly contribute to motor performance in a different degree? In the following section the impact of both factors is calculated in stepwise-regression analyses. Explained variances in MT by size and distance is given in Table 6, arranged according to the level of inspection from a more general point of view (total group) to a very detailed insight depending on input device and task type. Table 6. Explained variances in MT by of size and distance Explained variances () size distance total (N=28) touchpad (n=14) trackpoint (n=14) selection (n=14) manipulation (n=14) touchpad-selection trackpoint-selection touchpad-manipulation trackpoint-manipulation As can be depicted from Table 6 the target size consistently accounted the higher amount of variance. When considering the performance of the total group the ratio is 2 : 1 (size : distance). Comparing the input devices, the ratio for the touchpad is nearly 1:1 but for the trackpoint the target size has a stronger influence (3.2 : 1). With respect to the type of task (selection versus manipulation), target size is again more important for the motor performance, with a slightly higher impact in the selection task (2.6 : 1) compared to the manipulation task (2.2 : 1). Evidently, the object size has a distinctly bigger influence on motor performance than the object distance, especially when using a trackpoint. 4. Discussion and Conclusions The impact of the present findings for user interface design meeting the demands of a graying society is presented in three sections. The first section is concerned with the usability of the two common notebook input devices touchpad and trackpoint. The second one focuses on the validity of Fitts Law (the modified version from MacKenzie, 1992) with respect to users age and type of internal input device. Finally, on the basis of the present results recommendations are posed regarding the needs of older notebook users for a user-friendly interface design.

6 Work with Computing Systems H.M. Khalid, M.G. Helander, A.W. Yeo (Editors). Kuala Lumpur: Damai Sciences Touchpad versus trackpoint The results clearly indicate that the performance with the trackpoint is by far worse than the one with the touchpad. Even in the easy trials, the participants working with the trackpoint were 800 msec (34 ) slower in the selection and 900 msec (12 ) slower in the manipulation compared to the touchpad. The performance differences between both input devices can possibly be referred to the higher consistency between finger and cursor movement in the touchpad. As the participants reported after the experiment, older users had severe difficulties in acclimating to the trackpoint usage and to control the finger force. It is important to take into account that the touchpad is the superior input device. A fast, time saving and intuitively accessible cursor control can be achieved Fitts law: strengths and weaknesses Generally speaking, Fitts Law proved to be valid for both internal notebook input devices touchpad and trackpoint reflecting the motor performance even in older users. Even though predictions by Fitts Law did not match observed movement times perfectly, neither due to the task type (selection and manipulation) nor to the different types of input devices. The most obvious discrepancy between predicted and observed performance could be traced back to the fact that differential contributions of size and distance occurred. Theoretically both should have an equal influence on the movement times and difficulty of movement, respectively. Clearly this was not the case. As shown, the target size is much more important than the target distance. 61 of the explained variance can be attributed to the size but only 33 to the distance. Further research has to reveal if valuable refinements of Fitts Law (e.g. Accot and Zhai, 2003) may predict the discrepancies identified with respect to target size and distance in a more precise way. In their work, Accot and Zhai (2003) concentrated on the effect of target shape (including heights) on motor performance Recommendations for an user-friendly interface design It is recommended that novice users especially those over 40 years of age - should preferably choose a touchpad for an easy, fast and accurate working. The handling of small target sizes (2.5 x 2.5 mm) is quite difficult for older users particularly when combined with long distances note that the longest distance examined here was 5 cm and bear in mind that in most software applications clearly longer distances have to be bridged. 5. Acknowledgement For Alex who showed us how short but wonderful life can be. 6. References Accot, J., & Zhai, S. (2003). Refining fitts law models for bivariate pointing. Proc. CHI 03 (pp ). New York: ACM. Batra, S., Dykstra, D., Hsu, P., Radle, K.A., & Wiedenbeck, S. (1998). Pointing devise performance for laptop computers. Proc. of the Human Factors and Ergonomics Society 42 nd Annual Meeting 1998 (pp ). Santa Monica: Human Factors and Ergonomics Society. Card, S.K., English, W.K., & Burr, B.J. (1978). Evaluation of mouse, rate-controlled isometric joystick, steps keys, and text keys for text selection on a CRT. In Baecker, R. & Buxton W. (Eds.). Readings in human-computers interaction (pp ). San Mateo: Morgan Kaufmann. Douglas, S.A., & Mithal, A.K. (1994). The effect of reducing homing time on the speed of a finger-controlled isometric pointing device. Proc. CHI 94 (pp ). Epps, B.W., Muto, W.H., & Snyder, H.L. (1986). Comparison of six cursor devices on a target acquisition task. SID International Symposium Digest of Technical Papers (pp ). Los Angeles, CA: Society for Information Display. Fitts, P.M. (1954). The information capacity of the human motor system in controlling the amplitude of movement. Journal of Experimental Psychology, 47(6), Göbel, M., Backhaus, C., & Krüger, M. (2002). Eigenschaften mobiler eingabegeräte für senioren. Retrieved August 14th, 2002, from Veroeffentlichungen/downloads/2002_GfA_Goebel_Seni oreneingabegeraete. Hayslip, B., & Panek, E.P. (1989). Psychomotor performance. Adult development and aging (pp ). Harper & Row, Publishers, New York. Köchling A., & Spannhake B. (1997). Demographische veränderungen und ihre auswirkungen auf die arbeit. In H. Luczak & E. Volpert (Eds.), Handbuch arbeitswissenschaft (pp ). Stuttgart: Schäffer- Poeschel. MacKenzie, I.S. (1992). Fitts law as a research and design tool in human-computer interaction. HCI, 7, MacKenzie, I.S., Sellen, A., & Buxton W. A. (1991). Comparison of input devices in element pointing and dragging tasks. Proc. CHI 91 (pp ). New York: ACM. Smith, M.W., Sharit, J., & Czaja, S.J. (1999). Aging motor control, and the performance of computer mouse tasks. Human Factors, 41, Stelmach, G.E., Goggin, N.L., & Gracia-Colera, A. (1987). Movement specification time with age. Experimental Aging Research, 13(1). Sutter, C., & Ziefle, M. (2004, submitted). Psychomotor efficiency in users of notebook input devices: confirmation and restrictions of Fitts Law as an evaluative tool for user friendly design. Annual meeting of the Human Factors and Ergonomic society. Tränkle, U., & Deutschmann, D. (1991). Factors influencing speed and precision of cursor positioning using a mouse. Ergonomics, 34(2), Welford, A.T. (1976). Motivation, capacity, learning and age. International Journal of Aging, 7, Welford, A.T. (1984). Psychomotor performance. Annual Review of Gerontology and Geriatrics, 4,