Content area
The usability of a touch-panel interface was compared among young, middle-aged, and older adults. In addition, a performance model of a touch panel was developed so that pointing time could be predicted with higher accuracy. Moreover, the target location to which a participant could point most quickly was determined. The pointing time with a PC mouse was longer for the older adults than for the other age groups, whereas there were no significant differences in pointing time among the three age groups when a touch-panel interface was used. Pointing to the center of a square target led to the fastest pointing time among nine target locations. Based on these results, we offer some guidelines for the design of touch-panel interfaces and show implications for users of different age groups. Actual or potential applications of this research include designing touch-panel interfaces to make them accessible for older adults and predicting movement times when users operate such devices. [PUBLICATION ABSTRACT]
Full text
The usability of a touch-panel interface was compared among young, middle-aged, and older adults. In addition, a performance model of a touch panel was developed so that pointing time could be predicted with higher accuracy. Moreover, the target location to which a participant could point most quickly was determined. The pointing time with a PC mouse was longer for the older adults than for the other age groups, whereas there were no significant differences in pointing time among the three age groups when a touch-panel interface was used. Pointing to the center of a square target led to the fastest pointing time among nine target locations. Based on these results, we offer some guidelines for the design of touch-panel interfaces and show implications for users of different age groups. Actual or potential applications of this research include designing touch-panel interfaces to make them accessible for older adults and predicting movement times when users operate such devices.
INTRODUCTION
The senior citizen population has been growing and will continue to increase. At the same time, rapid progress is being made in information technology (IT), such as personal computers and the Internet, and some forms of IT are commonplace in many environments, including the workplace and the home. The opportunity for older users to access IT has been increasing (Czaja, 1997; Vanderheiden, 1997). Because it is highly likely that older adults will use some form of IT as they carry out their daily activities, the issue of aging and IT is very important. Human-computer interaction must be designed and implemented so that age-related changes in functional abilities, such as perceptual, cognitive, and motor functions, are taken into account.
Touch-panel interfaces are ubiquitous. They are used with passometers (automatic ticketvending machines) in railway stations, bookretrieval systems in libraries, and cash dispensers or automatic teller machines at banks. As people of all ages have more opportunities to use touch-panel interfaces, it is necessary to design interfaces that are easy to use. Many researchers have investigated the usability of touch panels by comparing them with other input devices, such as the computer mouse (Card, English, & Burr, 1983; Epps, 1986). Compared with a mouse, the touch panel has the advantage of simplicity that is, it requires less learning time. For older computer users, who may operate a mouse or keyboard with limited speed and accuracy, the touch-panel interface is especially attractive and can make IT more accessible.
A few studies have shown that older people are particularly interested in IT and that they can readily learn to use it (Echt, Morrell, & Park, 1998; Kelly, Morrell, Park, & Mayhorn, 1999; Morrell & Echt, 1997). Some surveys suggest that those over 65 years of age have less experience with IT than do younger people (Rogers, Cabrera, Walker, Gilbert, & Fisk, 1996; Schwarts, 1988). Those who use IT are usually well educated, wealthy, and married (Morris, 1996). The number of middle-aged and older adults who use the World Wide Web is increasing (Morrell, Mayhorn, & Bennett, 2000).
A touch-panel interface might make computers more attractive and friendly to older adults because it requires less learning time and leads to a high degree of user satisfaction (Douglas &Mithal, 1997; Galitz, 1997; Picketing, 1986; Shneiderman, 1987). Input devices used for human-computer interaction are important because the usability of an input device affects the overall effectiveness of an interactive system. Charness, Bosnian, and Elliot (1995) suggested that direct input devices, such as a light pen or a touch panel, may be especially easy for older adults to use as they are direct-access devices that eliminate the need to translate a target selection device onto a CRT display. Until now, however, few researchers have systematically investigated the usability of a touch-panel interface for older users. Although determining the target location to which users can quickly respond would be applicable to the design of a touch-panel interface, there seem to be few studies that have examined which target location led to a quick response.
In this study, two experiments were carried out among participants from three age groups: young, middle-aged, and older adults. In Experiment 1, pointing times using a mouse and a touch panel were compared among the three age groups. A performance model for a touch-panel interface that estimates pointing time was developed. In Experiment 2, for each of the three age groups, the target location that led to the faster response was clarified. Based on the results of this study, we offer guidelines for the design of touch-panel interfaces. These guidelines have implications for users of different age groups.
METHOD
Participants
A total of 45 participants took part in two experiments. They were divided into three age groups: young (13 participants, 6 men and 7 women, 20-29 years old, mean age 23.65 ± 2.71 years); middle-aged (13 participants, 7 men and 6 women, 50-59 years old, mean age 57.14 ± 1.83 years); and older adults (19 participants, 10 men and 9 women, 65-75 years old, mean age 68.35 ± 2.68 years). They were all healthy and right-handed. None self-reported orthopedic or neurological diseases.
Apparatus
A personal computer (Dell, OPTILEX GX1501200SF) and a touch panel (Digital, FP2500T11) were used. Using C++ (Microsoft Version 5), the pointing task was programmed. The resolution of the touch panel was 640 × 480 pixels. Mouse emulation software (DMC, TSC-1310D/DD) was installed on the personal computer so that an operation similar to a mouse function could be realized for the touch-panel operation. Touching the touch-panel became equivalent to clicking the mouse.
Design
In Experiment 1, the following three conditions were selected: (a) distance to a target (four levels: 70, 100, 130, and 160 pixels); (b) target size (three levels: 30, 50, and 70 pixels); and (c) approach angle to a target (eight levels: 0°, 45°, 90°, 135°, 180°, 225°, 270°, and 315°; Figure 1). In Experiment 2, the experimental conditions were (a) target size (two levels: 30 and 50 pixels); (b) approach angle to a target (four levels: 0°, 90°, 180°, and 270°); and (c) location within a target (nine levels: nine positions in a 3 × 3 gridiron; Figure 2). Here, the distance to a target was fixed at 120 pixels.
In Experiment 1, the input devices (touch panel and mouse) and the sessions (five sessions) were within-subject factors. Age was a betweensubject factor. The data for this experiment were analyzed using analysis of variance (ANOVA) and performance modeling. When analyzing the pointing time and error rate by means of ANOVA, we arranged the data according to three factors: input device, age, and session. The mean pointing time and error rate for a total of 96 (4 × 3 × 8) trials were calculated according to the arrangement by those three factors. Thus arranged data were analyzed using a three-way (input device by session by age) ANOVA. When carrying out performance modeling of pointing time, we arranged the pointing time for each participant according to three conditions: (a) distance to the target, (b) target size, and (c) approach angle to the target. Thus arranged data were analyzed using multiple regression analysis.
In Experiment 2, the distance to the target and approach angle to the target corresponded to the distance from the initial point to the center of Location 5 (Figure 2). Whereas the effects on pointing time of the distance to the target and approach angle to the target were examined in detail in Experiment 1, the aim of Experiment 2 was to explore the target location leading to the quickest response. In Experiment 2, the numbers of different target sizes and approach angles to the target were fewer than those in Experiment 1 so as to reduce the duration of experiment. Pointing time and error rate were obtained according to the target location.
Figure 1. CRT display for the pointing experiment (Experiment 1). Experimental variables were the distance to a target (d), target size (s), and approach angle to a target (φ).Figure 2. CRT display for the pointing experiment using only a touch panel (Experiment 2). One of the locations (numbered squares) must be pointed to with the index finger (in this figure, Location 6).
Procedure
The participants were seated in front of a PC equipped with either a touch panel or a mouse for output. They were required to perform the tasks with their right hand throughout Experiments 1 and 2. The order in which they used the touch panel or the mouse was counterbalanced across participants. First, a filled circle appeared at the center of the CRT. After the participant touched the circle with his or her index finger or pointed at the circle with the PC mouse, a target appeared (Figure 1). The participant was required to touch or point to the target as quickly and accurately as possible. This activity corresponded to one trial. After the participant finished touching or pointing at this target, another filled circle appeared again at the center of the CRT. The pointing time, coordinates of the location on CRT at which participants pointed, and errors, if any, were recorded for all trials. An error trial was defined as a trial at which participants pointed outside of a target. The pointing time corresponded to the time from the appearance of the target until the participant touched or pointed to the target. In one experimental session, 96 trials (4 × 3 × 8) were conducted. The order of performance of the 96 trials was randomized across participants. All participants carried out 5 sessions for each device, resulting in a total of 480 trials per device. Between sessions, the participants were allowed to take a short break. Pointing time and errors, if any, were automatically sent to a data file. Pointing time was measured with an accuracy of 10 ms using the interval timer function of the PC.
In Experiment 2 only a touch panel was used. Each participant carried out a pointing task similar to that used in Experiment 1 except that the shape of the target was different (Figures 1 and 2). The participant was required to touch the location (area) within a square target highlighted in a 3 ? 3 gridiron (Figure 2). A total of 72 trials (4 × 2 × 9) corresponded to one session. Only one session was carried out. The order of performance of 72 trials was randomized across participants.
RESULTS
Experiment 1
In Figure 3, pointing time is shown as a function of both age group and input device (touch panel or mouse). With the touch panel, no differences in pointing time were observed among the three age groups. With the mouse, however, pointing time increased with age. A three-way (input device by age group by session) ANOVA conducted on the pointing time revealed a significant main effect of input device, F(1, 42) = 7.6845, p < .01, and a significant input device by age group interaction, F(2, 84) = 6.213, p < .01. No significant main effects of session, age, or session by device interaction were detected. The lack of a significant main effect of age was considered to be attributable to the fewer differences in pointing time among the three age groups when using the touch-panel interface. A significant session by age by device interaction was detected, F(8, 168) = 2.135, p < .05. For each device, the statistical significance of a session by age interaction was checked. Only for the mouse condition was a significant session by age interaction detected, F(8, 168) = 3.564, p < .01. This must be attributable to the decrease in pointing time while using a mouse for the middle-aged and older groups as the session advanced. Pointing time for the touch panel was nearly constant across the five sessions for all age groups. The differences in pointing time when using the mouse and the touch panel were small in the young group but large in the older group. These results indicate that the benefit obtained by using a touch panel was greater for the older group than for the young group.
Error rates were calculated as follows. As mentioned, the number of errors was recorded for each participant. Error rates for each session were computed by dividing this number by a total of 96 trials. Error rates for all conditions were less than 10%, and there were no significant main effects or interactions as a result of a three-way ANOVA similar to that used for pointing time. However, it must be noted that the mean error rate for 30 × 30 pixels was larger (about 9%) than that for the other conditions (about 7% for 50 × 50 pixels and about 3% for 70 × 70 pixels).
Figure 4. Mean pointing time and standard deviation as a function of the distance to a target (Experiment 1). The equation in this figure was estimated using the method of least squares.
The coefficient of determination (contribution) r^sup 2^ for both models was more than .95 (.972 for Equation 2, .964 for Equation 3). The term r stands for the multiple correlation coefficient. As a result of the conventional modeling of the relationship between the index of difficulty and pointing time (Fitts, 1954; MacKenzie, 1992; Murata, 1996, 1999), we obtained r^sup 2^ as .804. In the conventional model, the effects of the approach angle to the target were not taken into account. The coefficient of determination r^sup 2^ means that 100r^sup 2^% of the variation in pointing time could be accounted for by the variability in independent variables, such as d, s, and φ, of the regression equations. The proposed Equations 2 and 3 could account for more than 95% of the variation in pointing time, whereas only 80% of the variation in pointing time could be predicted by the conventional model. These results show that the proposed model has more predictive power than the conventional model. The coefficient of determination r^sup 2^ for the conventional and proposed models was obtained for each participant to show whether or not the difference of r^sup 2^ between the conventional and proposed models is statistically significant. As a result of a one-way (model: two levels [conventional model and Equation 2 or Equation 3]) ANOVA carried out on r^sup 2^, a significant main effect of model was detected for both Equation 2, F(1, 86) = 484.604, p < .01, and Equation 3, F(1,86) = 469.318, p < .01. This result indicates that the difference in r^sup 2^ between the conventional and the proposed model (Equations 2 and 3) is significant and meaningful. It must be emphasized that the lower r^sup 2^ for the conventional Fitts' law modeling shows that the computation using Fitts' law does not account for the approach angle effect.
Figure 5. Mean pointing time and standard deviation as a function of target size (Experiment 1). The equation in this figure was estimated using the method of least squares.Figure 6. Mean pointing time and standard deviation as a function of approach angle (Experiment 1).
Performance modeling was also conducted separately for the three age groups as follows. The unit for pt is expressed in milliseconds. The results are summarized in Table 1. The coefficient of determination for separate models was also more than .95. However, the separate models were not significantly or meaningfully different from the overall model from the viewpoint of predictive power.
Experiment 2
The purpose of this experiment was to investigate the difference in pointing time among nine target locations and to clarify which target location (1-9) was most desirable when pointing to a target. A two-way (age by target location) ANOVA conducted on pointing time revealed a main effect of only target location, F(8, 336) = 3.638, p < .01. There were no significant differences in pointing time with the use of the touch panel among the three age groups, so data from each group were pooled.
The relationship between the location (1-9) within a square target and the pointing time is shown in Figure 7. The relationship was obtained by pooling the s and φ conditions. Pointing time was affected by the locations within a target. Location 5 led to the fastest pointing time; pointing times for Locations 3 and 9 were slower. Similar results were obtained for each of the eight combinations of s and φ.
Error rates for each target location were calculated by pooling the conditions of d and φ. Then the mean error rate for all participants was calculated for each target location. The error rates were less than 15% across the nine locations.
DISCUSSION
Experiment 1
Pointing time when using a PC mouse was longer for the older group than for the young and middle-aged groups. With the touch panel, however, there were no significant differences in pointing time among the three age groups (Figure 3). Although the error rate for the touch panel was higher than that for the PC mouse for all age groups, there were no statistically significant differences in error rates when using the touch panel or the PC mouse. The slightly higher error rate might be overcome by slightly increasing the correct pointing area. As mentioned, the pointing time when using the touch panel was nearly constant across the five sessions for the three age groups. The results also indicate that a touch panel requires less learning time. As Charness et al. (1995) predicted, it has been empirically verified that the touch-panel interface is a promising input device that compensates for the diminished perceptual, cognitive, and motor functions of typical senior citizens. The pointing time data indicate that a direct input device, such as a touch panel, is superior to an indirect input device, such as a PC mouse, and would be especially useful for older people.
Figure 7. Pointing time compared among the nine target locations (Experiment 2). Target locations (labeled 1-9) are specified in Figure 2. Values within parentheses represent the standard deviations in pointing time.
For a menu target acquisition task using a mouse and a light pen, Charness, Holley, Feddon, and Jastrzembski (2004) compared performance among young, middle-aged, and older adults and found that a direct input device (light pen) reduced age-related differences. They suggested that direct input devices may be recommended for older adults. Although the type of direct input device and the experimental task in Charness et al. (2004) were different from those in this study, the result agrees with that in this study. Rogers, Fisk, McLaughlin, and Pak (2005) showed that the superiority of a direct input device (touch panel) over an indirect device (rotary encoder) changed according to the task requirements and participant age. They showed, for both younger and older adults, that a touch panel was faster than a rotary encoder when participants were operating a drop-down list box that did not require scrolling, although the type of indirect input device and the task were different from those in this study. The result was also similar to the finding in this study.
As IT becomes more common, older users are likely to use it to conduct routine or daily activities. Therefore, the issue of aging and IT is very important in human-computer interaction. Designers of human-computer interactions will need to consider how age-related changes in perceptual, cognitive, and motor functions affect older users. The touch-panel interface is a promising means of bridging the divide between older adults and IT.
As there were no significant differences in pointing time among the three age groups when using a touch panel, performance modeling was carried out by pooling data from each age group to obtain the overall model. The modeling was also done separately for the three age groups. The contribution (r^sup 2^) for each model was higher than .95, and highly predictive performance models were obtained. The predictive power did not differ between the overall model and the separate model in Table 1. In other words, the separate models were not significantly or meaningfully different from the overall model. In spite of this, the parameters γ and δ related to the approach angle effect were meaningfully different among the age group groups (Table 1). At present, it is not clear what is indicated by the difference in parameters γ and δ among age groups from the viewpoint of human movement science. This problem should be a topic of future research.
When it is necessary to predict pointing time and discuss the effects of moving distance, target size, and approach angle to design a system or software that frequently requires pointing operations, the proposed Equations 2 and 3 can be used effectively. The separate models shown in Table 1 may be used in addition to Equations 2 and 3.
Compared with the conventional model (r^sup 2^ = .804), the proposed model, which considers the effects of the approach angle to a target, can attain more predictive accuracy. These models enable one to predict the pointing time for a touch-panel interface and to use the predicted pointing time to design a display that is desirable for all age groups. The lower r^sup 2^ for the conventional Fitts' law computation does not account for the approach angle effect. The proposed model, which takes into account the approach angle effect, enhances the predictive power.
It must be noted that the results are applicable only to simple pointing tasks (requiring participants only to point to a target) but not to tasks requiring much precision, such as dragand-drop tasks performed when selecting a submenu from a pull-down menu. The benefit of the touch panel for middle-aged or older adults may be restricted to reasonably sized targets (larger than 50 × 50 pixels) because the target size of 30 × 30 pixels led to higher error rates, as stated in the Results section.
Experiment 2
As shown in Figure 7, pointing time was the shortest at Location 5 (center of the square). Similar relationships were obtained for each combination of s and φ conditions. The result suggests that the participant should point to the center, regardless of s and φ. Locations 3 and 9 led to slower pointing times. These locations are not recommended when the users are required to point to a square target.
Rogers et al. (2005), using unstacked and stacked button tasks, examined the effects of button size, movement direction, and movement distance on movement time for a touch-panel interface. They showed, for both younger and older adults, that movement was fastest to the buttons in the center column, corresponding to Locations 2, 5, and 8 in this study (see Figure 7), although the size and arrangement of targets were different from those in this study. This agrees with the result in this study. They also observed a tendency, similar to that in this study, for Locations 3 and 9 to lead to slower movement. Rogers et al. (2005) further tried to model the movement time using Fitts' law. They did not incorporate movement direction in the model, as this study did. For both stacked and unstacked buttons, the modeling results were worse than those in this study. The modeling method in this study would help to improve the predictive power of Rogers et al. (2005).
TABLE 2: Guidelines for the Design of Touch-Panel Interfaces, With Implications for Different Age Groups
General Discussion
The results of this study are applicable to predicting pointing time when designing a touchpanel interface for older users. The results also indicate that the touch-panel interface should be aggressively pursued to make IT accessible to older adults. Now that the target locations that lead to faster responses have been identified, the results should be incorporated into the design guidelines for touch-panel interfaces. A touchpanel interface should be recommended instead of a mouse, particularly for middle-aged and older adults. The design guidelines based on the two experimental studies including the three age groups are summarized in Table 2. The target size and the moving distance are expressed using the exact size (in millimeters).
Rogers et al. (2005) suggested that one input device is not always the best one. Although a pointing task was used in this study, one should consider the task requirements suggested by Rogers et al. (2005) and, taking age into account, establish what task requirements are facilitated by touch-panel interfaces. Future research will examine the effects of a touch-panel slope on work efficiency and localized muscular fatigue of the upper body. Usually, the slope of a touchpanel interface is fixed at 90°, with the touch panel installed vertically. An investigation of the optimal slope for a touch panel that ensures higher efficiency and lower workload for the upper body is necessary not only for promoting the usage of IT or computers for older adults but also for enhancing the usability of touch-panel interfaces common in many places, such as at passometers in railway stations and automatic teller machines at banks.
ACKNOWLEDGMENTS
This study was partially supported by the New Media Development Association, Japan. This study was conducted to develop a database for older computer users as a project of the Research Institute of Human Engineering for Quality of Life (HQL), Japan.
REFERENCES
Card, S. K., English, W. K., & Burr, B. J. (1983). Evaluation of mouse, rate-controlled isometric joysticks, step keys, and text keys for text selection on a CRT. Ergonomics, 21, 601-613.
Charness, N., Bosman, E. A., & Elliot, R. G. (1995). Senior friendly input devices: Is the pen mightier than the mouse? In Proceedings of the 103rd Annual Convention of the American Psychological Association (pp. 495-551). Washington, DC: American Psychological Association.
Charness, N., Holley, P., Feddon, J., & Jastrzembski, T. (2004). Light pen use and practice minimize age and hand performance differences in pointing tasks. Human Factors, 46, 373-384.
Czaja, S. J. (1997). Computer technology and the older adult. In T. K. Landauer & P. Prabhu (Eds.), Handbook of human-computer interaction (2nd ed., pp. 797-824). Amsterdam: NorthHolland.
Douglas, S. A., & Mithal, A. K. (1997). The ergonomics of computer pointing devices. London: Springer.
Echt, K. V, Morrell, R. W., & Park, D. C. (1998). The effects of age and training formats on basic computer skill acquisition in older adults. Educational Gerontology, 24, 3-25.
Epps, B. W. (1986). Comparison of six cursor control devices based on Fitts' law models. In Proceedings of the Human Factors Society 30th Annual Meeting (pp. 327-331). Santa Monica: Human Factors and Ergonomics Society.
Fitts, P. M. (1954). The information capacity of the human motor system in controlling the amplitude of movement. Journal of Experimental Psychology, 47, 381-391.
Galitz, W. O. (1997). The essential guide to user interface design An introduction to GUI design principles and techniques. New York: Wiley.
Kelly, C. L., Morrell, R. W., Park, D. C., & Mayhorn, C. B. (1999). Predictors of electronic bulletin board system use in older adults. Educational Gerontology, 25, 19-35.
MacKenzie, I. S. (1992). Fitts' law as a research and design tool in human-computer interaction. Human-Computer Interaction, 7, 91-139.
Morrell, R. W., & Echt, K. V. (1997). Designing instructions for computer use by older adults. In A. D. Fisk & W. A. Rogers (Eds.), Handbook of human factors and the older adult (pp. 335-361). New York: Academic Press.
Morrell, R. W., Mayhorn, C. B., & Bennett, J. (2000). A survey of World Wide Web use in middle-aged and older adults. Human Factors, 42, 175-182.
Morris, D. C. (1996). Ownership, use, and non-use of personal computers by older adults. Journal of Educational Technology Systems, 25, 5-12.
Murata, A. (1996). Empirical evaluation of performance models of pointing accuracy and speed with a PC mouse. International journal of Human-Computer Interaction, 8, 457-469.
Murata, A. (1999). Extending effective target width in Fitts' law to a two-dimensional pointing task. International Journal of HumanComputer Interaction, 11, 137-152.
Pickering, J. A. (1986). Touch-sensitive screens: Technologies and their application. International Journal of Man-Machine Studies, 25, 249-269.
Rogers, W. A., Cabrera, E. F., Walker, N., Gilbert, D. K., & Fisk, A. D. (1996). A survey of automatic teller machine usage across the adult life span. Human Factors, 38, 156-166.
Rogers, W. A., Fisk, A. D., McLaughlin, A. C., & Pak, R. (2005). Touch a screen or turn a knob: Choosing the best device for the job. Human Factors, 47, 271-288.
Schwarts, J. (1988). The computer market. American Demographics, 10, 38-41.
Shneiderman, B. (1987). Designing the user interface: Strategy for effective human-computer interaction. Reading, MA: AddisonWesley.
Vanderheiden, G. C. (1997). Design for people with functional limitations resulting from disability, aging, or circumstance. In G. Salvendy (Ed.), Handbook of human factors and ergonomics (pp. 2012-2052). New York: Wiley.
Atsuo Murata is a professor in the Department of Computer and Media Technologies at Hiroshima City University. He received his Ph.D. in engineering in 1987 from Osaka Prefecture University.
Hirokazu Iwase is a lecturer at Tokyo Seitoku University. He received his Ph.D. in information sciences in 2002 from Hiroshima City University.
Date received: June 6, 2003
Date accepted: February 7, 2005
Atsuo Murata, Hiroshima City University, Hiroshima, Japan, and Hirokazu Iwase, Kanagawa University, Kanagawa Prefecture, Japan
Address correspondence to Atsuo Murata, Hiroshima City University, Department of Computer and Media Technologies, 3-4-1, Ozukahigashi, Asaminami-ku, Hiroshima 731-3194, Japan; [email protected]. HUMAN FACTORS, Vol. 47, No. 4, Winter 2005, pp. 767-776. Copyright © 2005, Human Factors and Ergonomics Society. All rights reserved.
Copyright Human Factors and Ergonomics Society Winter 2005