A tentative efficiency index for pointing device use in computer aided design: A pilot study

Abstract

BACKGROUND:

Focusing on the efficiency aspect of computer pointing devices’ usability, this paper reports on a novel and tentative empirically derived efficiency index for 3D CAD.

OBJECTIVE:

Three commercially available computer pointing devices were compared: a standard horizontal computer mouse, a vertical device (supporting neutral pronation of the forearm) and a slanted device.

METHODS:

Pilot structured observations of 10 subjects’ activity were carried out to estimate the proportion of each unique computer mouse operation during CAD modelling with a 3D parametric software. Pointing, dragging and steering standardized tasks were implemented by software and performed by 20 users. Effectiveness and efficiency were calculated and discomfort, effort and ease of use were subjectively assessed.

RESULTS:

The mean efficiency index value was lower for the vertical device. Assessments of discomfort, effort and ease of use also supported considering preference for the horizontal and slanted devices, providing limited internal validation.

CONCLUSION:

Results suggest the tentative index may offer a valid means of ranking performance of alternative pointing devices regarding operation efficiency.

Keywords

Effectiveness usability PC mouse performance

1 Introduction

The activity of Computer Aided Design (CAD) typically involves the intensive use of a computer handheld pointing device, which is dominant over the computer keyboard [1], raising ergonomic concerns. In addition to postural and biomechanical aspects related to computer handheld pointing devices’ usage, it is also important to perform usability assessment. Hence, selection of a user-friendly pointing device involves consideration of many aspects, including usability, which is comprised of the efficiency, effectiveness and satisfaction of the person in the activity of task completion [2]. This paper focuses on the efficiency aspect of usability, reporting on the proposal of an empirically derived efficiency index based on the results of two related experimental studies. In the first study, subjective and objective measures of usability were collected while comparing three commercially available PC mice, having a major difference between them in the orientation of the devices and their shape, with additional differences in size and weight. In a second study, observations were carried out to estimate the proportion of simple computer mouse operation during CAD activity using a parametric 3D CAD software (Autodesk Inventor ^® v. 2016). Twenty subjects participated in the first study, from which average times and efficiency per operation were obtained, while ten other subjects took part in the second study. In both studies, participants were volunteering students recruited from product design programmes of study.

The study involved both the development and application of the efficiency index, albeit in a pilot approach, aiming at a first comparative evaluation, akin with the approach reported by Coelho and Dahlman [3] for the sitting comfort of alternative seat geometries. The proposed index may hence be used in selecting between alternative devices. The example followed through in the paper yields a tentative composite efficiency index value for each of the three pointing devices evaluated. Each of the devices represents an archetype, resulting in different levels of usability, and of efficiency per device.

The proposed efficiency indicator is based on previous studies that report on the assessment of usability of handheld communication devices, including research by McCauley-Bush et al. [4] and Nunes et al. [5]. The latter provides a formula for assessing appropriateness for use in emergency situations. The underlying theory that supports the current study are the seminal works in usability by Andrew Dillon [6], Jacob Nielsen [7] and Patrick Jordan [8].

1.1 Previous studies

Computer usage can be associated with the development of neck and upper extremity pain, especially hand and forearm musculoskeletal pain induced by intensive mouse use [9]. A decade ago, approximately 30% to 80% of computer work involved the mouse [10], depending on the type of work. The PC mouse has become an essential part of computer work, even today; actually, the more recent use of tablet PCs does not substitute all the types of work usually performed using a conventional PC, CAD (computer aided design) operations are part of this group. Furthermore, recent research has concluded that tablet PC users are exposed to extreme wrist postures that are less neutral than those assumed with other computing technologies [11] and may be at greater risk of developing musculoskeletal injuries, especially when these devices are intensively used for long periods of time. One important issue is that screen positioning and pointing area positioning get in conflict for best posturing. Hence, methods have been developed to relieve these problems, such as palm rejection technology, although the results of research on the use of this technology show that it generally reduces discomfort but with increased wrist extension and with no benefit to shoulder unloading [12]. Extended use of computer pointing devices is bound to endure in present and future days, because in computer tasks such as pointing, dragging and steering, continuously needed, touch screens have so far not been able to replace the PC mouse, e.g. in 3D computer aided design [13]. The complexity of certain CAD operations and the time involved to produce this kind of computer work led some companies to invest in expensive pointing devices. In this field, there are some types of pointing devices that can lead to engage both hands, one standard device for use by one of the hands and one device for use by the other hand, intended for use with certain operational functions [14]. While the proposed efficiency index is conceptually applicable to any kind of hand operated pointing device or combination thereof, in this paper only single hand operation devices are included in the experimental studies.

Computer mouse weight is thought to influence wrist motion and muscle activity of the forearm when using the device in high-speed operation, while such effect is reduced in low speed operation; moreover, a mouse with proper weight would promote improved movement efficiency and decreased muscular activity during fast operation [15]. A proper mouse weight could hence benefit the users in terms of increasing movement efficiency. Its dimensions and geometry should be based on anthropometry, hand gestures and comfortable hand postures [16]. Hand size of the subjects seems to make a difference during computer mouse usage, affecting grasp position and the level of muscle activity, suggesting that a computer mouse must be chosen according to size of the hand of the subject [17]. Moreover, previous tests performed on a standard PC mouse (model A in the present study) revealed statistically significant association between hand width and effectiveness of dragging graphical targets using the middle button of the mouse [18].

A major concern reported in previous studies is related to musculoskeletal disorders. Therefore, research has been conducted by collecting data from muscle activity and motion analysis [15, 19], often the same emphasis is not given to usability, even when it comes to developing a new pointing device. Usability may be understood as an aspect of a system inherent to how easy it is to learn to use and to use. ISO standard 9241, laying out ergonomic requirements or office work with visual display terminals, defines usability as the extent a product may be used by specific users to attain specific objectives with effectiveness, efficiency and satisfaction [2]. Evaluation of pointing devices from an ergonomics and usability perspective involves the assessment of postural and biomechanical aspects as well as the efficiency, effectiveness and satisfaction of the person in the activity of task completion. This notwithstanding, standardized approaches to usability evaluation could benefit from greater specificity, which is aimed with the efficiency index proposed herewith.

Evaluation of pointing in general and its modelling started with the work of Fitts [20] and Fitts and Peterson [21], proposing Fitt’s law, which only covered the effects of target size and distance on task completion time (the errors are meant to be controlled) and was developed by MacKenzie [22], Zhai [23] and Drewes [24]. However, based on the mean actual pointing distances and their standard deviation, an effective target distance and effective target width can be computed, that allow the computation of an effective throughput that also reflects pointing accuracy [25]. Pointing accuracy evaluation tasks have been introduced by Soukoreff and MacKenzie [26]. Steering tasks have initially been modelled by Accot and Zhai [27] and by Liu and van Liere [28]. In several previous studies comparing distinct pointing device geometries, average movement time to complete specific graphic tasks was used to assess performance [29 –32].

Some studies have been carried out in the field of usability of electronic devices that propose indexes with the objective of classifying them and distinguishing usability in a specific context. Nunes et al. [5], based on the human-centred methodology developed by Jeelani [33], classified the usability of mobile devices per a specific use, presenting a formula for calculating a global score to be assigned to each device. This index is presented as an indicator of how appropriate a mobile device will be when used in emergency situations and is based on the assessment of 10 criteria that include a combination of mobile phones physical characteristics and usability attributes. McCauley et al. [4] also proposed a usability indicator considering the same typology of variables, affected by weighting coefficients, for the assessment of hand-held communication devices to determine the appropriateness of each device during emergency situations. McCauley et al. [4] suggested their methodology be altered to make it suitable in industries other than emergency management. The concept of a single indicator to predict performance is not entirely new, as it had already been considered by Card et al. [34] in their Keystroke-Level Model.

1.2 Purpose of the article and hypothesis

The fundamental assumption underlying this paper is that usability of computer pointing devices varies according to their geometric characteristics. The purpose of this article is to present the underlying principles and the experimental development of a new efficiency index for computer mice in their use with CAD software.

The index is applied to three examples of commercially available computer mice in the context of use with an exemplified parametric 3D CAD software (Autodesk Inventor ^® v.2016). The hypothesis considered for the study is that ‘the use of paradigmatically distinct PC mice geometries reaches different mean values of the new efficiency index during 3D CAD activity performed with a particular CAD software’.

2 Method

The fundamental principle followed in the index is the representativeness of the activity effectively conducted in terms of the time occupied with each operation and the type of operations carried out during the CAD activity. The development of the index followed an inductive approach, from the empirical data collection, in two phases, the first of which was experimental but naturalistic with observation of field of activity of CAD operators (study 1) and the second was also experimental and performed under controlled conditions in the laboratory (study 2).

The items for the index were chosen from the authors’ own knowledge and professional experience and CAD teaching. In this way, the items refer to the CAD operations that are performed with the PC mice in the 3D modelling processes. These operations, most of which are not exclusive to the CAD activity, were systematized in a software created for this purpose and that follows the general lines of previous work by Odell and Johnson [29] and Houwink et al. [31].

There are no comparable indexes (according to results of the literature review carried out on the usability of PC mouse), so it is a novel and original index. Given the innovative nature of the proposed index, it is hence not possible to compare the results obtained with those obtained from other indexes. Therefore, external validation is not feasible at this time. The application of the developed index was done for the example of a parametric CAD software (Autodesk Inventor ^® v.2016), and for the examples, regarding computer mice, of a horizontal device, a vertical device and a tilted device. The first study aimed to collect data on the effectiveness and efficiency of the elementary tasks (operations) for each of the devices under analysis. The second study aimed to quantify the parameters in the efficiency index equation, relative to the exemplified software.

3 Study 1

3.1 Participants

Ten subjects participated in the first study (mean age 23.9 (3.6)), 6 females and 4 males. Subjects were all right handed and had normal or corrected to normal vision. The subjects were recruited from among students of a graduate program in product design, with approval in the courses of Digital Modelling I and Digital Modelling II. All participants had at least two years of continuous experience in practicing with 3D CAD software and Autodesk Inventor in particular using a conventional pointing device geometry. Participant’s hand length varied between 164 and 198 mm and hand width varied between 76 and 96 mm.

3.2 Materials

Each participant had a similar stand-alone 3D CAD workstation running the same software and using the same computer peripherals (Fig. 1). Recording ensued through screen recording software running in each standalone workstation. During the recording time, participants modelled complex 3D designs. The participants had similar level of experience in the use of the 3D CAD software (Autodesk Inventor).

Fig. 1

Overview of the 3D modelling laboratory where study 1 took place (left image) and detail of the generic pointing device used in the lab (right image).

3.3 Procedures

A naturalistic observation study of the activity of 3D CAD operators was carried out. Ten participants’ individual activity was recorded simultaneously in an advanced 3D modelling course lab for 60 minutes.

Observation was made indirectly from the on-screen recordings of the activity, by a highly experienced user and tutor (with over 10 years of experience and continued practice). Observation was made for every 5th minute, counting the number of operations carried out in a total of 11 minutes. observations were categorized according to six of the operations involving a pointing device shown in Fig. 3 (pointing at large targets, pointing at medium targets, pointing at small targets, dragging with the left button, dragging with the middle button and steering; dragging with the right button was not an active function in the CAD software used). Table 1 depicts the actions within the software environment that were considered in identifying the device operations. Students recruited from product design programmes of study usually use some kind of ‘manual’ vectorization of background images during sketching, even when they use parametric CAD. Before applying some 3D features it is often necessary to start from drawings named sketches. For that reason, the authors considered these ‘manual’ operations as steering tasks. The observer who coded the observations was highly experienced and was simultaneously the trainer and monitor of all participants in study 1 with regard to 3D CAD activity using the software concerned. No other experts with a comparable high level of expertise were available for cross validation.

Fig. 3

Pointing, dragging and steering tasks (implemented by a tailor made computer software application); task sequence from top to bottom (pointing large to steering).

Table 1

Typical examples of correspondence considered between device operations and actions in the software environment exemplified

Pointing Device Operations	Pointing at large targets	Pointing at medium targets	Pointing at small targets	Dragging with left button	Dragging with middle button	Steering
Software Environment Actions (examples)	Tool Selection	Selecting tree elements	Selecting points in drawn entities	Free rotate, move plane	Panning	Recreating contours from non-vectorized background images

The number of operations of each of the categories observed within the 11 minutes analyzed was averaged over the ten participants.

4 Study 2

4.1 Participants

Ten female subjects (mean age 24.7 (3.1)) and ten male subjects (mean age 25.8 (6.0) participated in the laboratory experiments. Half of the participants were recruited from currently enrolled higher education students and another half were recruited from young graduates, within the domains of engineering, architecture and industrial design. All participants used a computer and conventional PC mouse daily, with 7 male subjects and 7 female subjects having more than two years of CAD training and practice. Subjects were all right handed and all had normal or corrected to normal vision. The length of the right hand of the 20 subjects varied between 164 and 198 mm and the hand width varied between 76 and 96 mm.

It may be argued that participants that were familiar with a particular kind of device would attain better efficiency in the use of that device in comparison with other devices participants were not that experienced with. The devices tested were all available commercially, and the size and archetype of device most used by each individual participant was not controlled in the study. Hence, the role of experience cannot be inferred from the experimental design of this study.

4.2 Materials

Figure 2 shows the devices used in study 2; model A is a Microsoft ^® standard horizontal PC mouse (reference Optical 200), while model B is an Evoluent ^® vertical PC mouse (supporting the adoption of a neutral forearm pronation posture by the person in the pointing activity; reference VM4R) and model C is an Anker ^® slanted (with mouse buttons plane angled at approximately 60 degrees with horizontal) PC mouse (reference TM137U). The standard PC mouse model A (Fig. 2) has a mass of 57 grams (taken from weighing the device on a precision scale with the cable horizontally supported; the total weight including cable and USB plug is 78 grams). Analogously, vertical PC mouse (model B) has a mass of 137 grams and the total weight including cable and USB plug is 170 grams, while the slanted PC mouse (model C) has a mass of 119 grams and the total weight including cable and USB plug is 145 grams.

Fig. 2

Handheld pointing devices studied shown with their respective brands.

4.3 Procedures

A set of tasks representative of a CAD operator’s activity were standardized and recreated by a tailor made computer software application to support the experimental studies undertaken. The standardized tasks included pointing at different sized targets, dragging with different mouse buttons, as well as steering. This set of tasks was collected and adapted from previous studies [29, 31]. All 20 subjects (10 female and 10 male) used each one of the devices performing the standard tasks in the following order: pointing at large targets (pointing large), pointing at medium targets (pointing medium) and pointing at small targets (pointing small) at first. Then, dragging targets with the left button (dragging left), dragging with the middle button (dragging middle), dragging with the right button (dragging right), and, finally, steering targets inside a representation of a tunnel.

The order of the tests with each of the three PC mice was randomized. The devices were randomly sorted and the participants performed the tests using the same device across the tasks in the sequence described above, and they then repeated the same sequence of tasks with another device after a resting period. Every single test was preceded by a practice run that was not used in the analysis of the data. A comparative overview of the graphical setup of the tasks is shown in Fig. 3. The pointing tasks consisted of alternately clicking on 18 equally distributed round targets arranged in an imaginary circle (Fig. 3). Participants clicked on the centre circle to start the task and then would move the cursor and click on the first active circle target (black-highlighted), if the click hit the target it would disappear, enabling the target on the diametrically opposite side of the circle, which when hit, would lead to the next target to randomly go active, and so on. The pointing task ran in pairs, one target was randomized and the next target stood opposite to it. The dragging tasks consisted of alternately dragging 8 equally distributed round targets arranged concentrically (Fig. 2) and participants would click and drag the circle to the diametrically opposite side matching the targets with another click. The steering task partially resembled the dragging task, it was necessary to hit the black-highlighted circle, release the mouse button, and then drive the circle to the diametrically opposite side matching the targets and trying not to get outside of the tunnel.

The purpose-built software collected several parameters of the trials including time to complete tasks and errors undergone, enabling calculation of effectiveness and efficiency usability parameters. The effectiveness (efa) for pointing and dragging tasks was calculated from Equation (1) whereas for the steering task Equation (2) was used. Efficiency (efi) was calculated from Equation (3). These equations are based on error rate (failures in hitting targets divided by total number of targets in the task) and time to complete tasks. Note that in Equation (1) the number of failed targets is always less or equal than the total number of targets. The steering task represents an exception, in that there are no targets identified, but the deviation from the shortest path is considered, akin to a grade of relative failure. ${efa}_{(point \land drag)} = 1 - \frac{No . FailedTargets}{No . TotalTargets}$ (1) ${efa}_{(Steering)} = \frac{minimum mean deviation}{mean deviation (subject)}$ (2)

$\begin{matrix} {efi}_{(point \land drag \land steering)} = \\ efa \times \frac{minimum mean completion TIME}{mean completion TIME (subject)} \end{matrix}$ (3)

efa_{(point∧drag)} – effectiveness of pointing and effectiveness of dragging

efa_(steering) – effectiveness of steering

efi_{(point∧drag∧steering)} – efficiency of pointing (dragging and steering)

No.FailedTargets – number of failed targets by the subject for the particular task

No.TotalTargets – total number of targets to be hit for the particular task

minimum mean deviation – lowest mean deviation across the whole set of replications of subject-device combinations

mean deviation(subject) – mean deviation achieved in the steering task for the subject-device combination

minimum mean completion TIME – lowest mean completion time across the whole set of replications of participant-device combinations (for the particular task)

mean completion TIME (subject) – mean time to complete the particular task for the participant-device combination

Time to complete tasks involves setting a gold standard for the task, which consists of the minimum time achieved in task completion within the experiments; this is then used to gauge the subject’s time for task completion in the particular task. In equation 2, the value of the ‘minimum mean deviation’ is obtained by extracting the minimum mean deviation across the whole set of replications of participant-device combinations. Likewise for the value of the ‘minimum mean completion TIME’ in equation 3. Hence the minimum mean in both equations refers to the combination of subject and pointing device that lead in the focused task to the lowest mean in the whole set of experiments carried out. Hence, mean times to perform single operations within each task were also obtained from dividing the total task completion time by the number of targets in the task and then averaging over all twenty subjects participating in study 2. For the purpose of calculating the mean time per operation, results obtained concern model A (standard) and the non-conventional models B (vertical) and C (slanted), considered separately at first, and then average times are obtained across the three models.

The elapsed time between targets and required to complete each task was counted for a fixed number of targets, starting with the mouse click on a central target (circle) and ending when the fixed number of targets was completed. The software also records the intermediate times between successfully completed targets. All tasks were performed in two cycles. The pointing task included the random activation of 12 targets per cycle (of 18 possible), totalling 24 targets. The tasks of dragging and steering relied on the random activation of 4 targets per cycle (of the possible 8), making a total of eight targets.

Pointing, dragging and steering standardized tasks were performed with software support, errors and times to complete tasks were also accounted for and recorded by the same purpose-built software. A failure was recorded in the case of pointing tasks whenever the mouse cursor was not positioned inside the target (circle) at the click of a button; analytically the value of the radius of the circle was used as reference. Dragging tasks were counted as a failure when the centre of the transported (dragged) circle was not dropped near the target circle centre, whose threshold is defined by the circle radius value. Steering task deviations were measured through an iterative process; each time the centre of the led circle deviated from the straight path, the position was measured and if the object was delivered successfully on target the average deviation occurred during transport to the target was recorded; otherwise the task was reinitiated automatically.

Participants also assessed their discomfort and effort subjectively in the completion of the tasks using each one of the pointing devices, as well as rating the ease of use of each device in the course of the activity within the performance of the standardized tasks. Subjects were given 3 scales: discomfort (3 items), ease of use (7 items) and effort (5 items). Both subjective and objective evaluation parameters are compared across the sample between the three devices under focus. Table 2 summarizes the comparative study performed. Subjects were given 3 scales (discomfort, ease of use and effort), each one composed of several items. Ratings were provided in 6-point Likert scales. Statistical analysis was carried out using IBM SPSS version 23.

Table 2

Overview of the tasks and evaluation parameters in the comparative study

Device	Tasks	Usability Evaluation
		objective		subjective
		measures	calculations	(subjects ratings)
Model A – Microsoft;	pointing (different size targets);	time (to completion);	efficiency;	discomfort; effort;
Model B – Evoluent;	dragging (different mouse buttons);	errors (undergone)	effectiveness	ease of use
Model C – Anker	steering

Each session in study 2 lasted between 10 and 12 minutes per device. Two other non-commercial alternative pointing devices were evaluated in the same experiment, and the order of evaluation was randomized for each subject across the several devices evaluated. The total time of use of the five devices tested was 50 to 60 min per participant. This paper focuses only on the 3 commercially available devices tested.

4.4 Analysis

The non-parametric Mann-Whitney U test [35] was applied to the distributions of the four subjective evaluation variables (hand discomfort, forearm discomfort, overall ease of use and overall effort) across the three models under study. Subjective usability variables were correlated (Spearman rank order correlation, per the approach described in Coelho et al. [36]) with objective variables. Mean efficiency of tasks and mean efficiency were plotted for the three PC mouse models.

4.5 Comparative evaluation of usability results

Participants ranged in age from 20 to 38 years old (mean = 25 years, SD = 4.8 years) and all of them were right handed. Hand width (hand breath) and hand length were measured using a retractable steel tape measure, resulting, respectively on female hand width with a mean of 79.9 mm (SD = 4.06 mm), female hand length with a mean of 177.3 mm (SD = 5.73 mm), male hand width with a mean of 88.8 mm (SD = 4.02 mm) and male hand length with a mean of 191.7 mm (SD = 4.67 mm).

The non-parametric Mann-Whitney U test [35] was applied to the distributions of the four subjective evaluation variables (shown in Figs. 4 and 5 as stacked histogram bars) across the three models under study. As a result, the null hypothesis stating that ‘the distributions are the same across the three categories of pointing devices’ was not rejected with statistical significance over any of the four variables of interest. This notwithstanding, medians differed for model A in comparison with B and C in what concerns hand and forearm discomfort, which were one level more intense for model A (6 point Likert scales were used).

Fig. 4

Stacked histogram of hand discomfort and forearm discomfort per PC mouse model (All rated from ‘1’ to ‘6’; Discomfort: from ‘1’ – extreme discomfort to ‘6’ – no discomfort); model A (standard), model B (vertical), model C (slanted); statistical tests did not reveal significant differences across devices.

Fig. 5

Stacked histogram of overall ease of use and overall effort per PC mouse model (All rated from ‘1’ to ‘6’; Ease of Use: from ‘1’ – very difficult to ‘6’ – very easy; Effort: from ‘1’ – extreme effort to ‘6’ – no effort); model A (standard) – Microsoft, model B (vertical) – Evoluent, model C (slanted) – Anker; statistical tests did not reveal significant differences across devices.

Figure 6 depicts mean effectiveness of task completion using PC mouse models A, B and C and from these results it is observed, globally, that model A and model C seem to be both more effective than model B. This applies to almost all the tasks performed by the subjects (except for dragging with the left button).

Likewise, mean efficiency of task completion is given in Fig. 7. This shows that the mean efficiency of tasks completion is comparably lower in model B. Model C achieves the highest efficiencies across the board, with model B scoring the lowest efficiencies across the board, while model A stands in between the other two models. The variables under focus were analyzed to statistically prove or disprove the differences among subgroups, such as those exemplified in Figs. 4–6 giving good support relatively to objective evaluation parameters of usability.

Fig. 6

Mean effectiveness of tasks plotted against the three PC mouse models considered in the study (n = 20); model A (standard) – Microsoft, model B (vertical) – Evoluent, model C (slanted) - Anker.

Fig. 7

Mean efficiency (per device focused and per task); standard deviation shown as error bar (n = 20); model A – standard, model B – vertical, model C - slanted

Additionally, the subjective usability variables depicted in Figs. 4 and 5 were correlated (Spearman rank order correlation, per the approach described in Coelho et al. [36]) with the objective variables depicted in Figs. 6 and 7, across the three categories of pointing devices included in the study. Forearm discomfort did not correlate with any of the objective usability variables considered. Hand discomfort correlated with effectiveness of pointing at large targets (rho = 0.267, p = 0.039). In what concerns overall effort, correlation was found with effectiveness of the pointing at large and medium targets (rho = 0.312, p = 0.015; rho = 0.287, p = 0.026) and with efficiency of dragging with the middle button (rho = 0.263, p = 0.042). Finally, the subjective variable of overall ease of use was positively associated to the following six objective usability indicators: effectiveness of pointing at large, medium and small targets (rho = 0.356, p = 0.005; rho = 0.292, p = 0.023; rho = 0.284, p = 0.028), efficiency of pointing at medium and small targets (rho = 0.290, p = 0.025; rho = 0.308, p = 0.017), and with the efficiency of dragging with the middle button (rho = 0.285, p = 0.027). These results indicate the very expressive importance of the pointing operations in formulating the subjective impression of overall ease of use.

Mean time per single operation is shown in Fig. 8, for the seven tasks included in study 2. Results shown derive from software automatically recorded time data and are specific to each pointing device. Moreover, the results are obtained by averaging across the participants’ individual time for task completion divided by the number of targets in the task. All mean times concerning both models B (vertical) and C (slanted) are higher than the ones for model A (standard), hinting at the participants familiarity with model A, and possibly, the increased difficulty to control model B (as the operating fingers are bound to become out of view of the operator obstructed by the vertically positioned body of the device). The mean times for model C (slanted) are positioned in between the times for models A (standard) and B (vertical) in all but one case (dragging with the left button, where mean times for model C are higher than the times for the other two models).

Fig. 8

Mean time [s] for single operation completion (per device focused and per task); standard deviation shown as error bar (n = 20); model A – standard, model B – vertical, model C – slanted.

Table 3

Mean observed operation counts (standard deviation shown within parentheses) and computed coefficients

Pointing Device Operations	Pointing at large targets	Pointing at medium targets	Pointing at small targets	Dragging with left button	Dragging with middle button	Steering
Mean operation count (n = 10)	30.9 (10.6)	63.1 (24.0)	37.4 (15.3)	24.4 (8.7)	11.9 (5.1)	8.9 (5.8)
Fractional time of use	a = 10.1%	b = 21.8%	c = 18.1%	d = 23.1%	e = 11.2%	f = 15.7%

5 Efficiency index

Considering the mean single operation times obtained from study 1, coefficients representing the relative fraction of time each operation category was undertaken were then calculated for each of the six operations. Considering the efficiency values obtained from study 2 for each of the tasks and computer pointing devices, combined with the aforementioned coefficients, enabled computing a weighted efficiency (obtained from Equation (4)) for each of the pointing devices, which is specific to the particular software and context of the studies.

$\begin{matrix} {efi}_{w} = \\ \frac{a . {efi}_{{poi}_{l}} + b . {efi}_{{poi}_{m}} + c . {efi}_{{poi}_{s}} + d . {efi}_{{drag}_{l}} + e . {efi}_{{drag}_{m}} + f . {efi}_{st}}{a + b + c + d + e + f} \end{matrix}$ (4)

efi_w – weighted efficiency

efi_{poi
_l}- efficiency of pointing large

efi_{poi
_m}- efficiency of pointing medium

efi_{poi
_s}- efficiency of pointing small

efi_{drag
_l} - efficiency of dragging left

efi_{drag
_m} - efficiency of dragging middle

efi_st- efficiency of steering

a, b, c, d, e and f - fractional time of use of each type of operation

5.1 Analysis

The statistical T-test for paired comparisons was applied to compare the efficiency index across the three PCs models under study. The Spearman rank order correlation was computed on the efficiency index with respect to subjective usability variables.

5.2 Observation of 3D CAD actions and efficiency index development results

Participants’ (6 males and 4 females) mean age was 23.9 years (standard deviation of 3.6 years). Observed operation counts, averaged over the 10 participants in study 21 are presented in Table 3. In addition, coefficients a, b, c, d, e and f, from Equation (4) are also included, translating into fraction of time of engagement in each kind of task. The operation of dragging with the right button is not included in calculating the coefficients, as this action is not used in the CAD software focused in this study.

Finally, combining the results from studies 1 and 2 with those shown in Table 3 enables computing Equation (4) for the three devices focused in the study. Results are shown in Table 4. The statistical T-test for paired comparisons yielded significance (p < 0.01) for the differences in the efficiency index across two pairs of devices (models A and B, models B and C). The efficiency index was correlated with the subjective usability assessment variables depicted in Fig. 4 yielding a moderate significant correlation with global ease of use (rho = 0.301; p = 0.019).

Table 4
Efficiency index mean and standard deviation per device, for the focused CAD software (n = 20)

Model Efficiency Index Mean (SD)

A (standard) 0.600 (0.063)

B (vertical) 0.503 (0.053)

C (slanted) 0.637 (0.082)

Model	Efficiency Index Mean (SD)
A (standard)	0.600 (0.063)
B (vertical)	0.503 (0.053)
C (slanted)	0.637 (0.082)

6 Discussion and conclusion

An experimental set up with a total of 30 participants was the basis on which to perform usability evaluation of three handheld devices (PC mice) geometrically and paradigmatically quite distinct. The first one is a standard, classic, horizontal and symmetric PC mouse and the second device is an alternative vertical PC mouse (supporting the adoption of a neutral forearm pronation posture by the user in the pointing activity), while the third one was a slanted device (with an inclination around 60 degrees). The study included both subjective and objective evaluation parameters of usability.

The differences reached in efficiency among the three models for the tasks under interest, is statistically supported, in spite of the small sample size and short session time that may have benefited the classic device (model A - standard), showing clearly better performance results for model C (slanted). The tasks pointing at medium size targets, pointing at small size targets and dragging with middle button of the PC mouse play a key role in several computer aided design software tools, hence the present study may help users to better choose their PC mice.

Association between subjective and objective variables suggests the prominent role of pointing tasks in the subjective formulation of the concept of overall ease of use. This notwithstanding, discomfort subjective variables were not significantly associated to any of the objective usability parameters considered. The results suggest that the envisaged health benefits in what concerns a lowered risk of musculoskeletal disorders of the hand, wrist and forearm proposed in the adoption of the vertical PC mice are opposed by reduced efficiency and may increase effort and discomfort (hand and forearm) in the short term leading to the perception of lower ease of use. The results of the comparison reported in this article are in line with those of Hedge et al. [37] and of Odell and Johnson [38] suggesting the use of slanted configurations of handheld pointing devices, to enable achieving a compromise between the expected long term effects on health and the objective and subjective task completion usability parameters.

6.1 Limitations and future studies

The study reported in this paper is not without limitations, with the small number of participants limiting generalizability of the results. This notwithstanding, the potential for further development in the lines of the conceptual model for an efficiency index that is presented is reinforced by the association found between subjective ease of use of the devices and the efficiency index obtained from the use of the device in task performance within a CAD environment.

A tentative efficiency index was reported in this paper, based on pilot experimental studies. Future studies are required to ascertain in greater depth the validity of the proposed efficiency index. Future studies also include the expansion of the tasks included in the efficiency index, with consideration of zooming operations.

6.2 Concluding remarks

The proposed efficiency index, applied to three distinct pointing device models in a specific software context, enabled compiling efficiency across a spectrum of tasks in one single indicator. Moreover, it was moderately associated with the subjective variable of global ease of use, supporting its conceptual validity. The proposed efficiency index, embeds a methodology that can be applied to any handheld pointing device in different CAD software environments; its application to other CAD software packages and other alternative PC mice will be explored in future studies. This will assist decision makers and consumers alike in selecting handheld pointing devices according to the particular configuration of the software environment where these are to be used and to the benefit of added productivity and human-systems integration at the computer pointing interface.

The efficiency index reported is presented as tentative and conceptual, with room for future expansion, application and validation. Its conceptual originality is emphasized over its validity limitations given the pilot nature of the study.

Conflict of interest

None to report.

Footnotes

Acknowledgments

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The authors express their thanks to Rui A. Pitarma, Ph.D. for supportive action and to Noel Lopes, Ph.D. and Isabel L. Nunes, Ph.D. for methods development supportive contribuitons. The authors’ role was funded in part by Fundação para a Ciência e a Tecnologia project UID/EMS/00151/2013 C-MAST. A subset of the results of the study presented in this paper was previously presented in a conference [].

References

Cail

, Aptel

. Biomechanical stresses in computer-aided design and in data entry. International Journal of Occupational Safety and Ergonomics. 2003;9(3):235–55.

International Organization for Standardization. ISO 924111-1998 - Ergonomic requirements for office work with visual display terminals (VDTs): Guidance on usability. 1998.

Coelho

, Dahlman

. Articulation at shoulder level-A pilot experimental study on car seat comfort. Applied Ergonomics. 2012;43(1):27–37.

Bush

, Gaines

, Watlington

, Jeelani

, Curling

, Armbrister

. The development of a device selection model for wireless computing devices in high consequence emergency management. In: Rebelo

, Soares

, editors. Advances in Usability Evaluation, Part I. CRC Press; 2012, pp. 486–99.

Nunes

, Patriarca

, Matos

. Usability-based mobile phone selection for communications in emergency situations. In: Rebelo

, Soares

, editors. Advances in Human Factors and Ergonomics 2014. Advances in Ergonomics in Design, Usability & Special Populations: Part II. Proceedings of the 5th AHFE Conference; 19-23 July 2014; Krakow, Poland. USA: AHFE Conference; 2014, pp. 498–509.

Dillon

. User acceptance of information technology. In: Karwowski

, editor. Encyclopedia of Human Factors and Ergonomics. London: Taylor and Francis; 2001.

Nielsen

. Usability engineering. San Francisco: Elsevier Science & Technology; 1994, p. 362.

Jordan

. An introduction to usability. London: Taylor & Francis; 1998, p. 136.

Conlon

, Krause

, Rempel

. A randomized controlled trial evaluating an alternative mouse or forearm support on change in median and ulnar nerve motor latency at the wrist. American Journal of Industrial Medicine. 2009;52(4):304–10.

10.

Dennerlein

, Johnson

. Different computer tasks affect the exposure of the upper extremity to biomechanical risk factors. Ergonomics. 2006;49(1):45–61.

11.

Young

, Trudeau

, Odell

, Marinelli

, Dennerlein

. Wrist and shoulder posture and muscle activity during touch-screen tablet use: Effects of usage and configuration, tablet type, and interacting hand. Work. 2013;45(1):59–71.

12.

Camilleri

, Malige

, Fujimoto

, Rempel

. Touch displays: The effects of palm rejection technology on productivity, comfort, biomechanics and positioning. Ergonomics. 2013;56(12):1850–62.

13.

Lourenco

, Pitarma

, Coelho

. Ergonomic development of a computer pointing device - a departure from the conventional PC mouse towards CAD users. In: Proceedings of the 19th Triennial Congress of the International Ergonomics Association; 2015; Melbourne (AU); 2015 Available from: http://ergonomics.uq.edu.au/iea/proceedings/Index_files/papers/1200.pdf

14.

Fiorentino

, Uva

, Fabiano

, Monno

. Improving bi-manual 3D input in CAD modelling by part rotation optimization. Computer-Aided Design. Advanced and Emerging Virtual and Augmented Reality Technologies in Product Design. 2010;42(5):462–70.

15.

Chen

, Lee

, Cheng

. The weight of computer mouse affects the wrist motion and forearm muscle activity during fast operation speed task. European Journal of Applied Physiology. 2011;112(6):2205–12.

16.

Rempel

, Camilleri

, Lee

. The design of hand gestures for human-computer interaction: Lessons from sign language interpreters. International Journal of HumanComputer Studies. 2014;72(10-11):728–35.

17.

Agarabi

, Bonato

, De Luca

. A sEMG-based method for assessing the design of computer mice. In: Proceedings of the 26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society; 2004; San Francisco, USA: IEEE Engineering in Medicine and Biology Society; 2004, pp. 2450–53.

18.

Lourenco

, Pitarma

, Coelho

. Association of hand size with usability assessment parameters of a standard handheld computer pointing device. In: Arezes

, Baptista

, Barroso

, Carneiro

, Cordeiro

, Costa

, et al., editors. Occupational Safety and Hygiene IV. London: Taylor & Francis Group; 2016, pp. 339–43.

19.

Lin

, Young

, Dennerlein

. Evaluating the effect of four different pointing device designs on upper extremity posture and muscle activity during mousing tasks. Applied Ergonomics. 2015;47:259–64.

20.

Fitts

. The information capacity of the human motor system in controlling the amplitude of movement. Journal of Experimental Psychology. 1954;47(6):381–91.

21.

Fitts

, Peterson

. Information capacity of discrete motor responses. Journal of Experimental Psychology. 1964;67(2):103–12.

22.

Mackenzie

. Fitts' law as a research and design tool in human-computer interaction. Human-Computer Interaction. 1992;7(1):91–139.

23.

Zhai

. Characterizing computer input with Fitts' law parameters–the information and non-information aspects of pointing. International Journal of Human-Computer Studies. 2004;61(6):791–809.

24.

Drewes

. Only one Fitts' law formula please!. In: CHI '10 Extended Abstracts on Human Factors in Computing Systems. Proceedings of 28th annual conference on Human Factors in Computing Systems. Atlanta, USA, New York, USA: ACM Press; 2010. pp. 2813–22.

25.

Mackenzie

, Isokoski

Fitts throughput and the speed-accuracy tradeoff. In: Proceedings of the 26th annual CHI conference on Human Factors in Computing Systems - CHI 08. Florence, Italy, New York, USA: ACM Press; 2008, pp. 1633–36.

26.

Soukoreff

, Mackenzie

. Towards a standard for poi ing device evaluation, perspectives on 27 years of Fitts research in HCI. International Journal of Human-Compute Studies. 2004;61(6):751–89.

27.

Accot

, Zhai

. Beyond Fitts law. In: CHI '97 Extended Abstracts on Human Factors in Computing Systems. Proceedings of the ACM SIGCHI conference on Human Factors in Computing Systems. Atlanta, USA, New York, USA: ACM Press; 1997, pp. 295–302.

28.

Liu

, Liere

. The effect of varying path properties in path steering tasks. In: Proceedings of Joint Virtual Reality Conference of EuroVR - EGVE - VEC 2010. Stuttgart, Germany. Aire-la-Ville, Switzerland: Eurographics Association; 2010, pp. 9–16.

29.

Odell

, Johnson

. Evaluation of a mouse designed to improve posture and comfort. In: Proceedings of the 2007 work with Computing Systems Conference. Stockholm, Sweden, 2007.

30.

Lee

, Fleisher

, Mcloone

, Kotani

, Dennerlein

. Alternative computer mouse design and testing to reduce finger extensor muscle activity during mouse use. Human Factors: The Journal of the Human Factors and Ergonomics Society. 2007;49(4):573–84.

31.

Houwink

, Hengel

, Odell

, Dennerlein

. Providing training enhances the biomechanical improvements of an alternative computer mouse design. Human Factors: The Journal of the Human Factors and Ergonomics Society. 2009;51(1):46–55.

32.

Lai

, Yang

. An ergonomics-oriented product development and evaluation process: A case study of children's pointing device. JCIT: Journal of Convergence Information Technology. 2012;7(6):323–31.

33.

Jeelani

. The development of a human centered methodology for the identification of communication needs and the assessment of hand-held communication devices used to support communication flow in high consequence emergency management [thesis]. 2011.

34.

Card

, Moran

, Newell

. The keystroke-level model for user performance time with interactive systems. CACM Communications of the ACM. 1980;23(7):396–410.

35.

Siegel

, Castellan

. Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill Book Company; 1988.

36.

Coelho

, Harris-Adamson

, Lima

, Janowitz

, Rempel

. Correlation between different hand force assessment methods from an epidemiological study. Human Factors and Ergonomics in Manufacturing & Service Industries. 2011;23(2):128–39.

37.

Hedge

, Feathers

, Rollings

. Ergonomic comparison of slanted and vertical computer mouse designs. Human Factors and Ergonomics Society. 2010;54(6):561–5.

38.

Odell

, Johnson

. Evaluation of flat, angled, and vertical computer mice and their effects on wrist posture, pointing performance, and preference. Work. 2015;52(2):245–53.

39.

Lourenco

, Coelho

. Performance evaluation of PC mice. In: Arezes

, Baptista

, Barroso

, Carneiro

, Cordeiro

, Costa

, et al., editors. Occupational Safety and Hygiene V. London: Taylor & Francis Group, 2017, pp. 157–61.

A tentative efficiency index for pointing device use in computer aided design: A pilot study

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSION:

Keywords

1 Introduction

1.1 Previous studies

1.2 Purpose of the article and hypothesis

2 Method

3 Study 1

3.1 Participants

3.2 Materials

4.1 Participants

4.2 Materials

4.5 Comparative evaluation of usability results

5.2 Observation of 3D CAD actions and efficiency index development results

Table 4 Efficiency index mean and standard deviation per device, for the focused CAD software (n = 20) Model Efficiency Index Mean (SD) A (standard) 0.600 (0.063) B (vertical) 0.503 (0.053) C (slanted) 0.637 (0.082)

6.1 Limitations and future studies

6.2 Concluding remarks

Conflict of interest

Footnotes

Acknowledgments

References

Table 4
Efficiency index mean and standard deviation per device, for the focused CAD software (n = 20)

Model Efficiency Index Mean (SD)

A (standard) 0.600 (0.063)

B (vertical) 0.503 (0.053)

C (slanted) 0.637 (0.082)