Human Factors: The Journal of the Human Factors and Ergonomics Society http://hfs.sagepub.com/ Light Pen Use and Practice Minimize Age and Hand Performance Differences in Pointing Tasks Neil Charness, Patricia Holley, Jeffrey Feddon and Tiffany Jastrzembski Human Factors: The Journal of the Human Factors and Ergonomics Society 2004 46: 373 DOI: 10.1518/hfes.46.3.373.50396 The online version of this article can be found at: http://hfs.sagepub.com/content/46/3/373 Published by: http://www.sagepublications.com On behalf of: Human Factors and Ergonomics Society Additional services and information for Human Factors: The Journal of the Human Factors and Ergonomics Society can be found at: Email Alerts: http://hfs.sagepub.com/cgi/alerts Subscriptions: http://hfs.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://hfs.sagepub.com/content/46/3/373.refs.html >> Version of Record - Jan 1, 2004 Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 What is This? Light Pen Use and Practice Minimize Age and Hand Performance Differences in Pointing Tasks Neil Charness, Patricia Holley, Jeffrey Feddon, and Tiffany Jastrzembski, Florida State University, Tallahassee, Florida We contrasted performance with mouse and light pen input devices for younger, middle-aged, and older adults (N = 72) who were experienced mouse users. Participants used both preferred and nonpreferred hands to perform a menu target selection task. The light pen minimized age differences in performance relative to the mouse. Older adults were more lateralized on a handedness test than young adults and were less efficient using their nonpreferred hand. With practice, older adults improved their response time more than other age groups did. The mouse was rated as more acceptable and easier to use than the light pen across trials, despite the performance advantage of the light pen for all age groups. Usability ratings correlated moderately with performance. A benefit-cost analysis indicated that the more efficient light pen might cover its greater initial cost within 11 months for an older adult and within 23 months for a younger adult. Actual or potential applications of this research include advising older adults to persist with practice for new input devices, advising those who must switch to their nonpreferred hand to select a direct positioning device, and providing a methodology for determining the potential payback interval when switching to a faster, though more expensive, input device. INTRODUCTION Computer systems are ubiquitous in work environments and are becoming so in homes. Although approximately 51% of households in the United States reported having a computer system as of August 2000, older adults lagged in adopting them for personal use (Newburger, 2001). Newburger noted that fewer than 30% of households headed by someone 65 years or older reported owning a computer. In the second half of 2000 only about 15% of adults aged 65 years or older reported having Internet access, compared with about 56% for the U.S. population as a whole and 75% for those aged 18 to 29 years (Rainie & Packel, 2001). Computer system usability problems may be one reason older adults tend to lag behind their younger counterparts in computer ownership and Internet access (e.g., see Charness, 2003). Research indicates that novice older adults have difficulty using a mouse as a pointing device. Walker, Philbin, and Fisk (1997) showed that novice older adults were less accurate in using a mouse in a target acquisition task than were their younger counterparts. Adjustments they made to the interface (acceleration profiles) did minimize the differences between the two groups. In a prior study, Walker, Millians, and Worden (1996) showed that older experienced mouse users have similar problems using a mouse and that their performance also improved with adjustments to acceleration functions. Smith, Sharit, and Czaja (1999) also showed that novice older adults experienced much greater difficulty than did novice younger adults in pointing tasks, particularly for double-clicking operations. Chaparro, Bohan, Fernandez, Choi, and Kattel (1999) showed that older adults exhibited slower performance with both a mouse and a trackball and exhibited higher ratings of perceived exertion with the mouse, as compared with Address correspondence to Neil Charness, Department of Psychology, Florida State University, Tallahassee, FL 323061270; [email protected]. HUMAN FACTORS, Vol. 46, No. 3, Fall 2004, pp. 373–384. Copyright © 2004, Human Factors and Ergonomics Society. All rights reserved. Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 374 Fall 2004 – Human Factors younger adults. Charness, Kelley, Bosman, and Mottram (2001) found significantly more mouse errors during word-processing training for older than for younger novice adults, although this was not true for experienced word processors. Others have noted problems in the ability of older adults to control fine motor movements (e.g., Jagacinski, Liaou, & Fayyad, 1995; Liao, Jagacinski, & Greenberg, 1997). A few mouse use properties may pose greater difficulty for older adults than for younger adults. First, the mouse is an indirect pointing device (see Greenstein, 1997) in the sense that the user must map the movement of the device in one plane (on the mouse pad or desk surface) with movement of a cursor on a different plane (the computer screen). Second, normal mouse settings provide some gain to the device such that both velocity and acceleration in the plane of the mouse surface is augmented in the plane of the screen. There is considerable evidence that older adults experience significant declines in spatial abilities (e.g., Salthouse, Mitchell, & Palmon, 1989) that might support this mapping and that older adults are generally slower on translation activities, such as those found in the digit-symbol substitution test (Salthouse, 1992). It is also the case that older adults show systematic slowing in many aspects of behavior (Salthouse, 1996). Not all input devices require similar mapping operations. Touch screens and light pens offer a “where you point is where you go” (WYPIWYG) operation for cursor control. They also exhibit no gain. If it is the case that age differences in mouse use are partially accounted for by age differences in mapping operation efficiency, providing a direct positioning device should minimize these differences relative to an indirect positioning device such as a mouse. In this project we chose to compare the usability of both direct (light pen) and indirect (mouse) pointing devices for younger, middle-aged, and older adults. The first goal for this experiment was to test hypotheses positing that difficulties in using a mouse may be attributable to age-related declines in spatial ability. To do this, we compared the effectiveness of the two types of input device for younger, middle-aged, and older adults who were experienced mouse users. We selected experienced individuals for several reasons. First, we found it nearly impossible to recruit younger novice adult computer users over a 1-year period, given the ubiquity of computers in the school systems. Second, we wanted a conservative test of the potential advantage of a light pen. Finally, given that future cohorts of older adults are more likely to be experienced users of computer systems, we wanted to be able to generalize our results to the case of experienced users, contrasting this study with prior studies that have examined mouse use by novices. A second goal for this study was to contrast the use of preferred and nonpreferred hands for pointing tasks with direct and indirect positioning devices. Chronic conditions leading to impairments and disability show striking increases with age, particularly past the age of 65 (e.g., Benson & Marano, 1998). Hence, older adults in particular are more likely than younger ones to experience injury (e.g., musculoskeletal disorders) or disease (e.g., stroke, arthritis) resulting in loss of function of the preferred hand. We wanted to examine the case of adaptation to use of the nonpreferred hand, particularly for older adults. Prior research showed a complicated interaction of Hand × Device × Task for movement time in a small sample of young adults using a mouse, stylus, and trackball (Kabbash, MacKenzie, & Buxton, 1993), although hand was manipulated between subjects. One factor that may mediate the ability to switch performance to the nonpreferred hand is degree of lateralization. Those who rely a great deal on one hand, to the exclusion of use of the other, may be expected to have greater difficulty in switching. Laterality questionnaire measures often ask for frequency of use of the two hands or preference for use. There is some indication in questionnaire studies that older adults are more strongly right lateralized for handedness than are younger adults. Gilbert and Wysocki (1992) showed a decrease with age in sinistrality and mixed-handedness with a twoitem (writing, throwing) questionnaire that was returned by more than 1 million U.S. men and women. That is, there was greater concordance in hand use with increased age. Degree of handedness also tends to increase with age in nonhuman primates (e.g., Ward, Milliken, Dodson, Stafford, & Wallace, 1990). Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 LIGHT PEN VERSUS MOUSE FOR MENUS However, contrary to questionnaire reports, studies of manual performance sometimes show stronger handedness asymmetry in young adults. Francis and Spirduso (2000) selected righthanded adults in their study of age and type of motor task and showed that younger adults exhibited a greater disparity between right and left hands than did older adults, particularly for a Purdue Pegboard task and a tracing task. We asked participants to use both their preferred and nonpreferred hand to perform a pure pointing and selection (point-and-click) task and then to rate the ease of use and acceptability of the device, as well as their perceived workload across practice blocks. A third goal for our study was to trace the effects of practice on point-and-click tasks. Although we expected to show that older adults would be less efficient than younger ones in pointing tasks and that the preferred hand would be superior to the nonpreferred hand, we were particularly interested in whether practice would interact with the factors of age, hand, and device. For instance, the age-complexity hypothesis postulates greater age differences in performance with more difficult tasks (e.g., Cerella, Poon, & Williams, 1980), and skill acquisition studies have shown greater performance gains by older adults across initial trial blocks (e.g., Charness & Campbell, 1988). Data could be used to provide recommendations for the type of input device to be adopted when people must switch to the use of their nonpreferred hand. The final goal for our study was to examine the influence of individual difference factors (e.g., degree of handedness; manual dexterity; and various visual, psychomotor, and memory factors) on efficiency with direct and indirect pointing devices. METHOD Design and Participants The design formed a 3 × 2 × 2 × 5 mixedmodel factorial. Age (young, middle-aged, or old) was a between-subjects variable. People were selected for ages 18 to 25, 45 to 55, and 65 to 75. Hand (preferred and nonpreferred), device (mouse and light pen), and practice blocks (Blocks1–5) were within-subject variables. Within each age group, 24 participants were pre- 375 sented with each of four conditions (preferred hand/mouse; nonpreferred hand/mouse; preferred hand/light pen; and nonpreferred hand/ light pen), with the order of presentation counterbalanced across participants within age groups. All participants received five blocks of trials on each combination. The young adults (M = 21.5 years) were recruited from undergraduate psychology classes or the community and received either extra course credit or a stipend for their participation. The middle-aged adults (M = 50 years) and older adults (M = 69.5 years) were recruited from the community and were paid a stipend for their participation. All were experienced mouse users in terms of the criterion of having at least 5 hr of prior mouse use. We also gathered a large set of demographic and psychometric measures in two sessions that took place before the experiment as part of a larger project (Czaja et al., 2004). Participants were given both group-administered (Session 1) and individually administered (Session 2) tests of visual and auditory acuity, perception, memory, attention, and reaction time. In addition we administered the Grooved Pegboard (Lafayette Model 32025) test of manual dexterity to use as a predictor variable so that we could assess to what extent age-related differences in manual dexterity might mediate age-related differences in menu selection performance. Apparatus and Materials The experiment was conducted at PC workstations equipped with a FastPoint Technologies, Inc., light pen (donated by FastPoint) and a Microsoft Intellimouse™, a two-button mouse. The light pen was attached to a Universal Serial Bus (USB) hub to the right of the computer with a flexible cable and had a clickable end point. It operates a software cursor via a device driver program that interacts with the graphics card to determine when a screen pixel is activated via the left-to-right and top-to-bottom scan pattern for the CRT. Pressing the tip of the light pen against the screen registers the equivalent of a mouse click using the device’s driver. A program written in Visual C++ provided the menu-based task using standard Windows™ controls such as drop-down menus and point-and-click activation (see Figure 1). Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 376 Fall 2004 – Human Factors Figure 1. Partial screen display of the menu acquisition task. It also contained embedded subprograms to measure handedness, preference and acceptability ratings, and workload. As shown in Figure 1, the menu bar of the program contained four menu items on the left, labeled “file,” “view,” “tools,” and “help,” consecutively. (An “experimenter” menu item on the far right of the screen could be used by the experimenter to terminate the program.) Each menu box held a drop-down menu containing target boxes labeled “1,” “2,” and “3,” consecutively with a maximized window setting on a 19-inch (48-cm) Gateway VX900 monitor that was set to a resolution of 1280 × 1024 pixels. A push button labeled “start” was located in the center of the program window. (Figure 1 is cropped at the bottom and on the right and shows the label “Help 1” where “start” appears at the beginning of the trial.) Trials were initiated when a participant clicked on the start button. That label was then replaced with a target label (e.g., “Help 1”) that designated the correct target response for a trial. The target label was written to the same screen location as the start item to allow for the possible disadvantage for some pointing devices (e.g., touch screen, light pen) of obscuring the item to be selected (Greenstein, 1997). That is, after the start button is clicked on with a light pen, the target that replaces the word start may not be visible until the light pen moves away from that location. Feedback for final target acquisition was a moderately loud 75 dB SPL 1000 Hz tone when the participant clicked on the correct menu item. A message box area provided trial reaction time feedback to participants and was located just below the start/target item push button. The Edinburgh Handedness Inventory (Oldfield, 1971) was embedded in the general experimental program and was run first to allow for quick determination of a participant’s preferred hand. Two rating scales (0–10) were also embedded in the program to measure acceptability and ease of use of a given input device at the end of a trial block (e.g., following completion of a block of trials with the preferred hand using a mouse). Finally, a workload inventory, the NASA-Task Load Index (NASA-TLX) was included in the program. The NASA-TLX scale (Hart & Staveland, 1988) required the participant to rate the task on the basis of a number of dimensions encompassing workload. The entire task, consisting of two parts, was administered: paired comparison of dimensions (which was greater – mental or physical workload) and Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 LIGHT PEN VERSUS MOUSE FOR MENUS ratings of the degree of workload for each dimension. We used the combined workload rating in our analyses. Procedure Each individual experimental session began by asking participants to complete the Edinburgh Handedness Inventory. The computer scored handedness items and provided feedback (via a message box) that indicated the preferred hand. For all but 1 participant this was the right hand. Next, participants were given instructions about the menu target acquisition task. Participants were told to move the cursor over the start button and click it, which would change “start” to a target menu item (e.g., “File 1”). Then, participants were told to click on the menu item (“file”), which would initiate a dropdown menu. Finally, participants were told to click on the target item (“1”). (All clicks were single clicks.) All participants were instructed to respond as quickly as possible. They were told that the edit box located below the push button would provide feedback about their performance in hundredths of a second. At this point, participants were encouraged to ask questions. If participants had no questions they proceeded to the practice block, where they practiced three trials for each condition, starting with their preferred hand. Throughout practice, message boxes instructed participants when to change the device and/or hand they were using. After the practice block, the main experimental trials were presented. At the end of each block participants were asked to rate the acceptability and ease of use for that Hand × Device condition on a 0 to 10 rating scale. After completing five blocks, participants responded to the NASA-TLX for the Hand × Device condition that they had just completed. This process was repeated for each of the remaining three experimental conditions. Participants were then paid for their participation and debriefed. RESULTS Target Acquisition Time As is the case in normal computer menu use, if participants made errors before selecting the correct item, they continued until the correct item was acquired. We conducted analyses of 377 errors, defining an error as the case where a mouse click occurred on a nontarget item before the target item. Using analysis of variance (ANOVA), we found a significant effect of only hand, F(1, 69) = 9.4, MSE = 2.67, p < .01, with slightly more errors for the nonpreferred than for the preferred hand (0.96 vs. 1.55 errors). However, such errors were very infrequent, considering that participants completed 120 trials with each hand. ANOVA was used to evaluate the median target acquisition time. Medians were chosen because of the extended response times observed on some trials (e.g., 1 participant paused in midtrial to go to the washroom), coupled with the small number of trials per block, which made outlier trimming procedures difficult. Overall, younger adults were fastest (4042 ms), middle-aged adults were slower (4818 ms), and older adults were slowest (5875 ms) for target selection. All main effects were statistically significant (p < .05): age, F(2, 69) = 26.13, MSE = 3 887 582; hand, F(1, 69) = 106.56, MSE = 563 606; device, F(1, 69) = 148.16, MSE = 558 878; and trial block, F(4, 276) = 66.61, MSE = 56 858. Three interactions were statistically significant: Age × Device, F(2, 69) = 3.36, MSE = 563 606; Age × Hand, F(2, 69) = 6.45, MSE = 558 878; and Age × Block, F(8, 276) = 4.25, MSE = 56 858. The Age × Device interaction, displayed in Figure 2A, shows that using a light pen lessens age differences in performance as compared with a mouse even for this group of experienced mouse users. The Age × Hand interaction, shown in Figure 2B, indicates that the older adults had greater difficulty using any device in the nonpreferred hand (550-ms disadvantage) as compared with the middle-aged (364 ms) and young (311 ms) adults. The Age × Block interaction indicates that older adults had the most to gain from practice (341 ms, as compared with 236 ms for middleaged adults and 157 ms for young adults), as shown in Figure 3. The Age × Block interaction is evident only in the first and second trial blocks, in which older adults showed a greater gain in performance (249 ms) than did the middle-aged (120 ms) and younger adults (72 ms). The interaction effect disappears if the ANOVA is restricted to Blocks 2 through 5. Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 378 Fall 2004 – Human Factors A B Figure 2. A: Age × Device interaction for mean median reaction time to acquire a menu target. B: Age × Hand interaction for mean median reaction time to acquire a menu target. Standard errors are shown for each bar. Laterality Quotient and Manual Dexterity The Oldfield (1971) Laterality Index ranges from –100 (extremely left-handed) to 100 (extremely right-handed). There was only 1 strongly left-handed individual (age 51) in the sample of 72 people, somewhat below expectations (~11% of the population is left-handed or mixedhanded; Gilbert & Wysocki, 1992). If that person is excluded from the sample and only righthanded participants are considered (where righthanded is anyone with a score above 0), then there is a significant trend for older adults to be more strongly lateralized (r = .32, p < .01). The stronger lateralization for older adults probably accounts in part for the interaction of age and hand in efficiency for acquiring the target item. The Oldfield (1971) questionnaire items tap behavioral activities such as writing, brushing teeth, and striking a match. The use of the nonpreferred hand for such activities is reported here as less frequent for older adults than younger ones. A more accurate behavioral check on effectiveness with the preferred and Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 LIGHT PEN VERSUS MOUSE FOR MENUS 379 Figure 3. Age × Trial Block interaction for mean median reaction time to acquire a menu target. Standard errors are shown for each bar. nonpreferred hands is afforded by the dexterity indices from the Grooved Pegboard task. This task requires people to fill a board with 25 small pegs as quickly as possible. Both time to complete the task and noticed errors are recorded. Error rates were very low (M = .32) and did not vary by age, hand, or their interaction (all p values > .15). For time to complete filling the pegboard, there were main effects of age, F(2, 69) = 4.1, MSE = 895, p < .01, and of hand F(1, 69) = 20.2, MSE = 68.9, p < .01. Time to complete the task increased with age (82.4, 84.7, and 98.6 s, respectively, for young, middle-aged, and older adults). Time to complete the task was greater for the left hand than for the right hand (91.7 vs. 85.4 s). There was no interaction between age and time to complete the task, F(2, 69) = 1.5, p > .2. When manual dexterity in the nonpreferred hand is entered as a covariate in the analysis of menu target acquisition performance, the interaction between age and hand for mean median time is no longer significant. That is, manual dexterity mediates, in part, the differential slowing seen in using an input device in the nonpreferred hand for older adults. Individual Differences and Performance To what extent are the age-related differences in performance with the light pen and mouse mediated by individual differences in basic abilities? We regressed a mean manual dexterity measure (averaging times across hands), together with two spatial ability measures (paper folding test and cube comparison test), and age on four measures of performance: mean light pen response time across blocks for the preferred and nonpreferred hands and mean mouse response time averaged across blocks for the preferred and nonpreferred hands. Results are presented in Table 1. Age was the only significant predictor for the preferred-hand performance measures, although there was an apparent trend for mean manual dexterity (p = .061, p = .082, respectively). However, manual dexterity contributed independently of age for nonpreferred-hand performance with both mouse and light pen. Consistent with the earlier ANOVA, age was a less-strong predictor of performance for the light pen than for the mouse. Also, manual dexterity was a stronger predictor of response time for the light pen than for the mouse, perhaps because people were experienced mouse users and the extent of physical movement was greater for the light pen. This analysis is inconsistent with the view that translation and mapping operations, considered to require spatial ability, are responsible Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 380 Fall 2004 – Human Factors TABLE 1: Individual Difference Predictors of Performance Beta SE β p (Constant) Age Cube comparison Paper folding Dexterity 1613.6 15.1 –10.9 –15.9 4.9 354.79 3.22 9.12 16.56 2.57 — .496 –.1340 –.1000 .177 .000 .000 .237 .339 .061 Preferred hand/light pen F(4, 67) = 10.6 R2 = .389 (Constant) Age Cube comparison Paper folding Dexterity 1746.77 8.16 –8.88 –19.94 3.77 294.82 2.68 7.58 13.76 2.14 — .355 –.1440 –.1650 .181 .000 .003 .245 .152 .082 Nonpreferred hand/mouse F(4, 67) = 18.4 R2 = .524 (Constant) Age Cube comparison Paper folding Dexterity 1755.75 17.23 –13.88 –12.52 9.21 415.09 3.77 10.67 19.38 3.01 — .470 –.1410 –.0650 .277 .000 .000 .198 .520 .003 Nonpreferred hand/light pen F(4, 67) = 10.1 R2 = .376 (Constant) Age Cube comparison Paper folding Dexterity 945.09 9.43 –17.95 6.12 13.94 487.52 4.43 12.53 22.76 3.53 — .251 –.1780 .031 .408 .057 .037 .157 .789 .000 Dependent Variable Predictors Preferred hand/mouse F(4, 67) = 16.3 R2 = .494 for older adults’ difficulty with using a mouse efficiently. Ease of Use and Acceptability Ratings Ease of use and acceptability judgments made on a 0–10 rating scale were essentially equivalent, with the two variables (aggregated across conditions) correlated at r = .94. ANOVA was used to evaluate the aggregated (summed) mean rating. The aggregate produced significant main effects of hand and block. The main effect of hand, F(1, 69) = 98.08, MSE = 15.37, p < .0001, indicated that participants considered the preferred hand to be easier and more acceptable to use than the nonpreferred hand. The main effect of block, F(4, 276) = 8.55, MSE = 0.56, p < .0001, indicated that ratings were more favorable with practice over blocks, such that the combined ratings were lowest following Block 1 and highest following Block 5. Significant interactions were apparent for both Hand × Device and Hand × Age. The significant interaction of Hand × Device, F(1, 69) = 9.32, MSE = 67.2, p < .003, showed that these experienced mouse users rated the mouse as easier to use in their preferred hand; however, as seen in Figure 4A, they rated the light pen as easier to use in their nonpreferred hand. Be- cause the light pen (a direct pointing device) does not require the user to deal with translation and mapping operations, it appears to be easier to use in a novel situation, such as using the nonpreferred hand for a point-and-click task. The Hand × Age interaction, F(2, 69) = 3.37, MSE = 9.58, p < .04, shown in Figure 4B, indicated that all three groups rated the preferred hand similarly; however, both the younger and middle-aged groups rated the nonpreferred hand more highly than did the older group. This result is consistent with the earlier results on menu performance and on general lateralization showing that older adults are less effective with their nondominant hand. When a correlation is computed within subjects by using 20 pairs of scores (for the 5 blocks × 2 devices × 2 hands), the mean correlation with ease of use in the sample is r = –.43, and an ANOVA shows that the mean correlation is significantly different from zero, F(1, 69) = 91, MSE = 0.146, p < .01, and that there are no differences among age groups, F(2, 69) = 0.13. A similar result is evident for the mean correlation between acceptability and median response time, with the sample correlation being r = –.30 and ANOVA showing this to be significantly different from zero, F(1, 69) = 34.8, MSE = 0.189, Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 LIGHT PEN VERSUS MOUSE FOR MENUS 381 A B Figure 4. A: Device × Hand interaction for mean aggregated ease of use and acceptability ratings. B: Age × Hand interaction for mean aggregated ease of use and acceptability ratings. Standard errors are shown for each bar. p < .01, and again, no age group differences are observed, F(2, 69) = 0.90. The moderate correlation between performance and ease of use (or acceptability) implies that participants may have been monitoring their own performance and using it in part to assign usability ratings. NASA-TLX Workload Analysis Following each condition, participants rated the amount of perceived workload in each of the four Hand × Device conditions using the NASA- TLX Workload Rating Scale. ANOVA revealed significant main effects of age, F(2, 69) = 4.664, MSE = 1116.24, p < .01; hand, F(1, 69) = 17.88, MSE = 118.47, p < .0001; and device, F(1, 69) = 4.493, MSE = 177.364, p < .04. The older group perceived the task as having a greater workload (M = 60) than did the other two groups, which were not significantly different from each other as tested by LSD contrasts (M = 45 for the middleaged and M = 49 for the young participants). Not surprisingly, the nonpreferred hand was rated as Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 382 Fall 2004 – Human Factors creating a greater workload (M = 54) than the preferred hand (M = 48). Finally, the mouse was perceived as creating less workload (M = 49) than the light pen (M = 53). There was also a significant interaction of Hand × Device, F(1, 69) = 6.361, MSE = 84.91, p < .01. The mouse was rated as creating the least amount of workload in the preferred hand (M = 45), and all other conditions were not significantly different (light pen/preferred hand M = 51; light pen/nonpreferred hand M = 54; mouse/nonpreferred hand M = 53). Benefit-Cost Analysis At the time of this writing (the year 2003), a mouse cost about $10 U.S. A light pen (or touch screen-equipped monitor) cost about $500 U.S. Although there were significant differences in performance favoring the light pen, particularly for older adults, it is useful to try to estimate the practical impact. That is, can we quantify the Gray and Boehm-Davis (2000) view that “milliseconds matter”? We attempted to answer the question, How long would it take a firm to recover the initial cost difference between the two devices for young, middle-aged, and older adults? As noted earlier, the practice block effect lines for devices became parallel for all age groups by the second trial block. Therefore the mean performance time differences for the devices for the different age groups can be used to estimate the efficiency of a device. We needed estimates of the average industrial wage and of the number of mouse clicks that users perform in a day in order to estimate a payback period. The average hourly private earnings in the United States in February 2003 was approximately $15 U.S. The sole study we were able to locate concerning mouse use was that by Liu, Conlon, Lee, Wang, and Chu (2001). They estimated a mean of 1045 mouse clicks per day with a standard deviation of 940 for male and female engineers, suggesting a very skewed distribution. That sample is not likely to be representative of general computer users and may not be particularly representative of older adult samples, in comparison with the group we tested. However, given the very broad range of mouse use implied by those data, we may adequately bound more typical populations. More research is needed on this topic. As shown in Table 2, using that mean and standard deviation, payback comes within a broad range. Assuming the mean number of mouse clicks, the cost differential for a light pen or touch screen over a mouse would be eliminated after about 23 months for young adults, 18 months for middle-aged adults, and10 months for older adults. However, the high standard deviation for the estimate of mouse clicks indicates that in the worst case, 105 clicks/day, it might take 19 years for young adults, 15 years for middle-aged adults, and 8.5 years for older adults to recover the cost of using a light pen. In the latter case, it would not be cost effective to buy the more expensive input device. In the two more favorable conditions (assuming 1985 and 1045 clicks/day) it would be appropriate. A number of factors could affect the sensitivity of this analysis. First, we lack solid information on the distribution of the number of clicks per day for a broad range of computer tasks (and what type of click: single vs. double). Second, older workers are generally paid more than younger workers (hence shortening their recovery period). Finally, we do not know the likelihood of substitution of workers for equipment TABLE 2: Time to Recover the Difference in Cost Between a Light Pen and a Mouse Light pen target acquisition time (ms) Mouse target acquisition time (ms) Gain/target for the light pen, given task had 2 targets (s) Gain/target using the average industrial wage ($U.S.) Targets required to equal $500 Days to equal $500 if 1045 clicks/day Days to equal $500 if 105 clicks/day Days to equal $500 if 1985 clicks/day Young Middle-Aged Older 1852 2184 0.166 0.0007 722 892 692 6885 364 2176 2601 0.2125 0.0009 564 706 540 5378 284 2534 3268 0.367 0.0015 326 975 313 3114 165 Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 LIGHT PEN VERSUS MOUSE FOR MENUS in larger firms (possibly lengthening the payback period, as compared with an assumption of a constant-sized workforce). DISCUSSION Practical Implications The observed main effects together with the three interactions for menu task performance can be translated into practical guidelines. 1. Older adults particularly would benefit from using a direct positioning device such as a light pen on computer tasks that require pointing as the main operation. Even experienced younger mouse users were faster when using the light pen. 2. The Age × Device interaction suggests that older adults using a mouse may experience translation problems from gain, which are not present with the light pen. Evidence for this interpretation is indirect. The light pen, which exhibits no gain, minimized age differences in performance. We rule out general spatial ability as a contributing factor because spatial ability (as measured by paper folding and cube comparison) does not seem to mediate age-related performance differences. However, manual dexterity, independent of age, was a significant individual difference predictor of menu task performance. To the extent that older adults are less able to control fine movement, as others have shown in the analysis of micromovements involved in general movement control, such difficulties may be exacerbated when learning to control a device with gain. However, more direct tests of the gain hypothesis are needed. Nonetheless, our results support the guideline that when designing for older user populations, one should be cautious about using devices with gain. 3. The Age × Hand interaction indicates that older adults, who may need to switch to their nonpreferred hand for controlling an input device (e.g., because of injury or arthritis), will initially have much more difficulty adjusting than will younger adults. 4. The Age × Trial Blocks interaction indicates that older adults have the most to gain from practice with an input device. Hence, older adults should be encouraged (e.g., via in- 383 structions accompanying the product) to persist with practice to ensure optimal performance. Caveats. It may also be the case that other input devices would yield a differential benefit for older adults. As has long been recognized in the human-computer interaction literature, the relative advantage of a device depends on the type of task being performed (e.g., Card, Moran, & Newell, 1983). For instance, when young adults performed a tracking task, a trackball led to lower tracking error and lower workload ratings than a mouse did (Hancock, 1996). For some types of interface controls (sliders, up/ down buttons, scrolling), a rotary encoder device was found to be superior to a touch screen (Rogers, Fisk, McLaughlin, & Pak, in press), particularly for older adults. For drop-down list boxes, the touch screen generally yielded faster performance. For data entry tasks, automated speech recognition is superior to keyboard and mouse only when the user types fewer than 45 wpm with text entry, and it is not superior for numerical data entry (Mitchard & Winkles, 2002). Thus an important caveat for these generalizations is that the task we investigated was a pure pointing task using drop-down menus. Many computer tasks involve interleaved use of two input devices: a pointing device and a keyboard. Interleaving creates problems for skilled touch typists because one hand must leave the home row key position to reach for a pointing device and then return to that position for further typing: the homing problem. We might expect a different pattern of results with mixed tasks because a light pen requires more extensive movements to access the screen and to move the cursor position than does a mouse. Further, we did not investigate clickand-drag activities (used to move objects to new positions on the screen) with the pointing devices. It may be more efficient and less effortful to click and drag with a mouse than with a light pen. ACKNOWLEDGMENTS This research was supported in part by the National Institute on Aging via Grant NIA1 PO1 AG17211-04, Project CREATE. We thank the Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014 384 Fall 2004 – Human Factors many departmental independent study undergraduates for helping with the project. REFERENCES Benson, V., & Marano, M. A. (1998). Current estimates from the National Health Interview Survey, 1995. In Vital and health statistics, Series 10: Data from the National Health Survey (No. 199, pp. 79–80). Washington, DC: National Center for Health Statistics. Retrieved January 21, 2003, from http:// www.cdc.gov/nchs/data/series/sr_10/sr10_199acc.pdf Card, S. K., Moran, T. P., & Newell, A. (1983). The psychology of human-computer interaction. Hillsdale, NJ: Erlbaum. Cerella, J., Poon, L. W., & Williams, D. M. (1980). Age and the complexity hypothesis. In L. W. Poon (Ed.), Aging in the 1980s: Psychological issues (pp. 332–340). Washington, DC: American Psychological Association. Chaparro, A., Bohan, M., Fernandez, J. E., Choi, S. D., & Kattel, B. (1999). The impact of age on computer input device use: Psychophysical and physiological measures. International Journal of Industrial Ergonomics, 24, 503–513. Charness, N. (2003). Commentary: Access, motivation, ability, design, and training: Necessary conditions for older adult success with technology. In N. Charness & K. W. Schaie (Eds.) Impact of technology on successful aging (pp. 28–41). New York: Springer. Charness, N., & Campbell, J. I. D. (1988). Acquiring skill at mental calculation in adulthood: A task decomposition. Journal of Experimental Psychology: General, 117, 115–129. Charness, N., Kelley, C. L., Bosman, E. A., & Mottram, M. (2001). Word processing training and retraining: Effects of adult age, experience, and interface. Psychology and Aging, 16, 110–127. Czaja, S. J., Charness, N., Fisk, A. D., Hertzog, C., Nair, S., Rogers, W. A., et al. (2004). Factors predicting the use of technology: Findings from the Center for Research and Education on Aging and Technology Enhancement (CREATE). Manuscript submitted for publication. Francis, K. L., & Spirduso, W. W. (2000). Age differences in the expression of manual asymmetry. Experimental Aging Research, 26, 169–180. Gilbert, A. N., & Wysocki, C. J. (1992). Hand preference and age in the United States. Neuropsychologia, 30, 601–608. Gray, W. D., & Boehm-Davis, D. A. (2000). Milliseconds matter: An introduction to microstrategies and to their use in describing and predicting interactive behavior. Journal of Experimental Psychology: Applied, 6, 322–335. Greenstein, J. L. (1997). Pointing devices. In M. V. Helander, T. K. Landauer, & P. V. Prabhu (Eds.), Handbook of human-computer interaction (2nd ed., pp. 1317–1348). Amsterdam: Elsevier. Hancock, P. A. (1996). Effects of control order, augmented feedback, input device and practice on tracking performance and perceived workload. Ergonomics, 39, 1146–1162. Hart, S. G., & Staveland, L. E. (1988). Development of NASATLX (Task Load Index): Results of experimental and theoretical research. In P. A. Hancock & N. Meshkati (Eds.) Human mental workload (pp. 139–183). Amsterdam: North Holland. Jagacinski, R. J., Liao, M.-J., & Fayyad, E. A. (1995). Generalized slowing in sinusoidal tracking by older adults. Psychology and Aging, 10, 8–19. Kabbash, P., MacKenzie, I. S., & Buxton, W. (1993). Human performance using computer input devices in the preferred and non-preferred hands. In Proceedings of the INTERCHI ’93 Conference on Human Factors in Computing Systems (pp. 474–481). New York: Association for Computing Machinery. Liao, M., Jagacinski, R. J., & Greenberg, N. (1997). Quantifying the performance limitations of older and younger adults in a target acquisition task. Journal of Experimental Psychology: Human Perception and Performance, 23, 1644–1664. Liu, W., Conlon, C., Lee, S., Wang, J., & Chu, M. (2001, June). VDT keyboard and mouse usage by male and female engineers. Presented at the 2001 American Industrial Hygiene Conference and Exposition, New Orleans, LA. Mitchard, H., & Winkles, J. (2002). Experimental comparisons of data entry by automated speech recognition, keyboard, and mouse. Human Factors, 44, 198–209. Newburger, E. (2001). Home computers and Internet use in the United States: August 2000. Washington, DC: U.S. Census Bureau. Retrieved January 22, 2003, from http://www.census. gov/prod/2001pubs/p23-207.pdf Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh inventory. Neuropsychologia, 9, 97–113. Rainie, L., & Packel, D. (2001). More online, doing more: 16 million newcomers gain Internet access in the last half of 2000 as women, minorities, and families with modest incomes continue to surge online. Washington, DC: Pew Internet & American Life Project. Retrieved September 25, 2001, from http://www. pewinternet.org/pdfs/PIP_Changing_Population.pdf Rogers, W. A., Fisk, A. D., McLaughlin, A. C., & Pak, R. (in press). Touch a screen or turn a knob: Choosing the best device for the job. Human Factors. Salthouse, T. A. (1992). What do adult age differences in the digit symbol substitution test reflect? Journal of Gerontology: Psychological Sciences, 47, P121–128. Salthouse, T. A. (1996). The processing-speed theory of adult age differences in cognition. Psychological Review, 103, 403–428. Salthouse, T. A., Mitchell, D. R. D., & Palmon, R. (1989). Memory and age differences in spatial manipulation ability. Psychology and Aging, 4, 480–486. Smith, M. W., Sharit, J., & Czaja, S. J. (1999). Aging, motor control, and the performance of computer mouse tasks. Human Factors, 41, 389–396. Walker, N., Millians, J., & Worden, A. (1996). Mouse accelerations and performance of older computer users. In Proceedings of the Human Factors and Ergonomics Society 40th Annual Meeting (pp. 151–154). Santa Monica, CA: Human Factors and Ergonomics Society. Walker, N., Philbin, D. A., & Fisk, A. D. (1997). Age-related differences in movement control: Adjusting submovement structure to optimize performance. Journal of Gerontology: Psychological Sciences, 52B, P40–P52. Ward, J. P., Milliken, G. W., Dodson, D. L., Stafford, D., & Wallace, M. K. (1990). Handedness as a function of sex and age in a large population of lemur. Journal of Comparative Psychology, 104, 167–173. Neil Charness is professor of psychology and an associate of the Pepper Institute on Aging and Public Policy at Florida State University. He obtained his Ph.D. in psychology from Carnegie Mellon University in 1974. Patricia Holley is a postdoctoral fellow at Florida State University. She received her Ph.D. in psychology from the University of South Florida in 1999. Jeffrey Feddon is a software support engineer at Aculab USA Inc., Panama City, Florida. He received his M.A. in psychology at the University of West Florida in 1996. Tiffany Jastrzembski is a doctoral student in the Psychology Department at Florida State University. She received her M.A. in psychology in 2003. Date received: April 30, 2003 Date accepted: March 8, 2004 Downloaded from hfs.sagepub.com at UNIV OF CALIFORNIA SANTA CRUZ on November 26, 2014