2014 the size and structure of arm movement variability decreased with work pace in a standardised...

7/26/2019 2014 the Size and Structure of Arm Movement Variability Decreased With Work Pace in a Standardised Repetitive Precision Task

1/13

This article was downloaded by: [134.117.10.200]On: 29 November 2014, At: 12:32Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House37-41 Mortimer Street, London W1T 3JH, UK

ErgonomicsPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/terg20

The size and structure of arm movement variability

decreased with work pace in a standardised repetitive

precision taskDivya Srinivasan

a, Afshin Samani

b, Svend Erik Mathiassen

a& Pascal Madeleine

b

aCentre for Musculoskeletal Research, Department of Occupational and Public Health

Sciences, Faculty of Health and Occupational Studies, University of Gvle, Gvle, SwedenbLaboratory for Ergonomics and Work-related Disorders, Center for Sensory-Motor

Interaction (SMI), Department of Health Science and Technology, Aalborg University, Aalbo

East, Denmark

Published online: 12 Sep 2014.

To cite this article:Divya Srinivasan, Afshin Samani, Svend Erik Mathiassen & Pascal Madeleine (2014): The size and structu

of arm movement variability decreased with work pace in a standardised repetitive precision task, Ergonomics, DOI:

10.1080/00140139.2014.957736

To link to this article: http://dx.doi.org/10.1080/00140139.2014.957736

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of thContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon ashould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveor howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions
http://dx.doi.org/10.1080/00140139.2014.957736http://www.tandfonline.com/action/showCitFormats?doi=10.1080/00140139.2014.957736http://www.tandfonline.com/page/terms-and-conditionshttp://www.tandfonline.com/page/terms-and-conditionshttp://dx.doi.org/10.1080/00140139.2014.957736http://www.tandfonline.com/action/showCitFormats?doi=10.1080/00140139.2014.957736http://www.tandfonline.com/loi/terg20


2/13

The size and structure of arm movement variability decreased with work pace in a standardised

repetitive precision task

Divya Srinivasana*, Afshin Samanib, Svend Erik Mathiassena and Pascal Madeleineb

aCentre for Musculoskeletal Research, Department of Occupational and Public Health Sciences, Faculty of Health and OccupationalStudies, University of Gavle, Gavle, Sweden; bLaboratory for Ergonomics and Work-related Disorders, Center for Sensory-Motor

Interaction (SMI), Department of Health Science and Technology, Aalborg University, Aalborg East, Denmark

(Received 17 June 2014; accepted 15 August 2014)

Increased movement variability has been suggested to reduce the risk of developing musculoskeletal disorders caused byrepetitive work. This study investigated the effects of work pace on arm movement variability in a standardised repetitivepipetting task performed by 35 healthy women. During pipetting at slow and fast paces differing by 15%, movements of arm,hand and pipette were tracked in 3D, and used to derive shoulder and elbow joint angles. The size of cycle-to-cycle motorvariability was quantified using standard deviations of several kinematics properties, while the structure of variability wasquantified using indices of sample entropy and recurrence quantification analysis. When pace increased, both the size andstructure of motor variability in the shoulder and elbow decreased. These results suggest that motor variability drops whenrepetitive movements are performed at increased paces, which may in the long run lead to undesirable outcomes such asmuscle fatigue or overuse.

Practitioner Summary: The size and structure of motor variability are associated with important outcomes in repetitivework, such as fatigue, pain and the risk of developing musculoskeletal disorders. Motor variability decreased when the paceof repetitive work was increased, indicating higher risks of developing undesirable outcomes such as muscle fatigue oroveruse.

Keywords:motor control; cyclic movements; Fitts law; kinematics; linear and non-linear variability

1. Introduction

Occupations involving repetitive tasks, such as short-cycle assembly work or meat cutting, are still prevalent in working life

and commonly agreed to be associated with an increased prevalence of work-related musculoskeletal disorders (MSDs), in

particular in the hand-arm and shoulder regions (Bernard 1997; National Research Council 2001; Buckle and Devereux

2002;Mathiassen2006; Larsson, Sogaard, and Rosendal2007). Lack of variation has been suggested to be an important

cause of these MSDs, and thus cyclic tasks associated with a low cycle-to-cycle variability in joint movements, jointcoordination and/or muscle activities between successive repeats may be particularly disposed to lead to fatigue and,

eventually, pain (Mathiassen, Moller, and Forsman2003; Madeleine2010; Srinivasan and Mathiassen2012). Cycle-to-

cycle variability will, according to the motor control literature, be intrinsically present even when an individual tries to

identically repeat a particular task (Newell and Corcos 1993). Several studies have suggested that increased motor

variability may have a positive effect in mitigating fatigue development in the upper extremities (Palmerud et al.1998; Falla

and Farina2007; Farina et al. 2008; Madeleine and Farina2008; van Dieen et al. 2009). Motor variability also enables

people to utilise redundant degrees of freedom, both at the individual muscle level and at inter-muscular and inter-joint

levels (Cote et al. 2008; Madeleine and Farina 2008; Holtermann et al. 2010), which can often be a way to preserve

performance (Cote et al. 2005,2008;Selen, Beek, and van Dieen2007; Gates and Dingwell2008; Fuller, Fung, and Cote

2011). Apart from the size of cycle-to-cycle variability, the structure of variability has also been proposed to convey

important information about the motor control system (Sosnoff, Valantine, and Newell2006), and it has been suggested that

the structure of motor variability is related to pain in the neckshoulder and skill in performing repetitive work using the

upper extremities (Madeleine and Madsen2009; Madeleine2010).

In a review on motor variability published recently (Srinivasan and Mathiassen2012), the authors conclude that motor

variability, represented by both its size and structure, is a relevant issue in an occupational context, associated with several

important outcomes in repetitive work, such as fatigue and pain. They recommend further research in several key areas in

order to understand relationships between motor variability and these outcomes. One such key issue of research is to better

understand which external factors in the design of work might be used to systematically manipulate motor variability. Thus,

q 2014 Taylor & Francis

*Corresponding author. Email: [email protected]

Ergonomics, 2014

http://dx.doi.org/10.1080/00140139.2014.957736
mailto:[email protected]://dx.doi.org/10.1080/00140139.2014.957736http://dx.doi.org/10.1080/00140139.2014.957736mailto:[email protected]:[email protected]


3/13

although motor variability occurs naturally to a certain extent between cycles in repetitive tasks, it may be possible to

systematically increase or decrease it by manipulating work factors such as pace, precision demands, workstation

configuration and cognitive loads on top of the repetitive physical task. Research into this issue is necessary, both to come

up with possible interventions that can promote motor variability in work environments and, at an even more basic level, to

probe the relationships between motor variability and physiological outcomes of relevance to MSD further.

Work pace is associated with how frequently repetitive movements are performed, which is an important work factor

known to influence biomechanical exposure levels (Sundelin 1993; Mathiassen and Winkel 1996; Laursen, Jensen, and

Sjogaard1998). Thus, as a reasonable hypothesis, work pace is important to the risk of developing MSD. Work pace is also

known to affect motor planning and execution processes, since corrective movement adjustments are harder to make at a

high speed (Elliott et al. 2004). Hence, it is reasonable to also expect that work pace influences motor organisation and

motor variability and, through this route, may even have an effect on MSD risk.

There are only a few studies that look into the effects of pace on motor variability, with inconclusive evidence on the

direction or extent of the effect. An early investigation concluded that contrary to expectations from Fitts law, the size of

kinematic variability of shoulder and elbow joint movements in multi-joint arm movements did not show any speed-

accuracy trade-off, and were relatively unaffected by changes in pace (Darling and Stephenson 1993). The effect of work

pace on the size of motor variability and subsequent fatigue development was recently studied in light manual assembly

work (Bosch et al.2011). While a faster work pace led to larger variability in some kinematic variables such as wrist speed

and acceleration, it did not lead to a more variable cycle time and was accompanied by more errors in performance.

In another study of butchers performing repetitive meat cutting work (Christensen et al. 2000), workers were classified as

being either slow or fast depending on their work cycle times. Representative samples of electromyography (EMG) fromthe forearm muscles indicated that there was a trend for the fast group to have higher variability than the slower group, but

the difference was not statistically significant. In parallel, repetitive low-load work performed at higher pace has been found

to be characterised by decreased size and structure of variability for arm and trunk movement (Madeleine 2010).

Thus, while a few studies suggest that work pace is a potential determinant of motor variability during repetitive arm

movements, the particular nature of this dependency is still unsettled. Our hypothesis is that since it is harder to make

continuous adjustments to movements at faster paces, movements would be planned with lower flexibility (reduced

utilisation of the abundance of degrees of freedom) at higher paces. Hence, higher pace would result in lower size and

structure of movement variability in order to preserve performance. The aim of this study was to pursue this hypothesis in

standardised low-load repetitive movements requiring high precision.

2. Methods

In this controlled laboratory experiment, pipetting was used to model repetitive precision work using the upper extremities(Park and Buchholz 2013). Considering the role of gender in motor variability (Svendsen and Madeleine 2010; Cote

2012), only females were recruited in this study. Thirty-five healthy university students in the nursing and health

education programmes were recruited through advertisements placed on university notice boards and the intranet. The

participants were aged 25 (5.8) years (mean (SD)), were 164 (8.8) cm tall and had average body mass of 60.4 (7.3) kg. All

participants were right-handed, and were free from any shoulder pain or injury at the time of the study, and had at least

one year of experience in pipetting (28 (7) months). All subjects signed an informed consent prior to inclusion. The

experiment was approved by the Ethical Review Board in Uppsala, Sweden, and conducted in accordance with the

Helsinki Declaration.

2.1. Experimental set-up

Each pipetting cycle consisted of aspirating coloured water from a pick-up tube (diameter 20 mm), transferring it to one of

eight target tubes in a 10 10 array of identical tubes (diameter 6 mm) and returning to the pick-up tube for the next cycle(Figure 1). Thus, the centre of the target tube array was at a fixed distance of 32 cm from the pick-up tube.

The participants sat in a rigid chair, and trunk movements were restricted by fixing their torso to the back of the chair by

belts. An unrestricted torso might have made the task slightly more realistic from an occupational standpoint, but allowing

for torso movements would have led to different use of the torso by different participants and thus to a less controlled

condition with respect to the primary focus of the study, i.e. motor variability in the armhand segment, i.e. the primary

movers in seated, upper-extremity precision tasks.

According to standard ergonomics guidelines, the table surface was height-adjusted to be at each participants elbow

height when they sat upright, and the chair was positioned such that the participants midline was aligned with the central

axis of the target tube array (Figure 1). The pick-up tube was at an arms length away from the participant when the upper

D. Srinivasanet al.2


4/13

and lower arms were held out horizontally. Thus, the task was standardised according to individual anthropometry in terms

of the height and distance of components, while the distance that had to be covered during the pipetting task was equal for

all subjects (i.e. 32 cm).

One pipetting session consisted of transferring liquid to each of the eight target tubes 20 times, i.e. in total 160 cycles, in

a randomised order. The target tube to be selected in each cycle was indicated using lights mounted below each tube and

controlled by a custom-built computer program, which also indicated the required pace by a beep sound, thus functioning as

a metronome. The participants were forced to follow the pacing of the task strictly since the computer program would

change the light to a different target tube at the end of every cycle. A standard amount of liquid was to be transferred to the

target tubes in each pipetting cycle, as preset on the pipette, but the participants were informed that this exact amount wasless important, since the primary requirement of the experimental task was to reach the correct target tubes at the set pace.

Each participant thus performed one pipetting session at a standardised work pace of 2.8 seconds/cycle (i.e. slow paced)

and another pipetting session at a work pace of 2.4 seconds/cycle (i.e. fast paced). Data were also collected in four other

sessions on the same day in which other work factors such as precision, movement direction and cognitive loads were

manipulated, but only the two pacing conditions will be analysed in this paper. The slow and fast paces of pipetting were

chosen to roughly correspond to 100 and 120 MTM paces for the given set-up (MTM: Methods-Time Measurement, a

predetermined motion time system used in industrial settings to set the standard time for completion of a task; Maynard,

Stegemerten, and Schwab1948). Whether the participants could keep up the pace or if they missed any tube in the sequence

was checked during the data analysis stage, using the described performance metrics. Each pipetting session could be

performed without any significant localised muscle fatigue in the arm, as confirmed by the amplitude and stable frequency

content of electromyographic signals from the upper trapezius and forearm extensor muscles (not reported) throughout each

session (Hagg, Luttmann, and Jager2000; Cifrek et al.2009). The order of the different pipetting sessions was randomised

with 10 minutes of rest in between two successive sessions. In order to get familiarised with the experiment, all participants

performed a training session of 100 pipetting cycles in the slow-paced pipetting condition before actual data collection.

2.2. Data collection and processing

Kinematic data were recorded by means of two synchronised electromagnetic tracking systems (Fastrak, Polhemus, USA),

at a sampling rate of 30 Hz. Thumb forces exerted on the pipette were recorded using a thin-film finger-tip force sensor

(A201, Tekscan Inc, USA) mounted on the pipettes push button. A three-segment rigid-body model of the upper arm,

forearm and hand segments (described in Domkin et al. 2005) was used for estimating the shoulder, elbow and wrist joint

angles as defined by the International Society of Biomechanics conventions (Wu et al.2005). The data were filtered using a

fourth-order, low-pass Butterworth filter with a cut-off frequency of 3 Hz. For each cycle, the start point was defined as the

time instant when the pipette tip was at the pick-up tube with minimum velocity; and the end point was defined as the instant

when the force on the pipettes push button was at its maximum (to dispel liquid into the target tube). The time instantswhen the velocity of the pipette tip increased above and decreased below 5% of peak velocity were used as cut-off points to

further trim each cycle. During data processing, cycle times were examined for any major deviations from the set cycle

time, as a measure of task performance. Twenty processed cycles of shoulder elevation and elbow flexion joint angles of

pipetting movements to each target were used for further analysis of motor variability.

The following variables were computed for each cycle to describe the kinematic properties of shoulder elevation and

elbow flexion movements: range of motion (ROM), peak velocity (Peak Vel), average velocity (Avg Vel), time to peak

velocity (Time PV) and area under the movement curve (Area). The cycle-to-cycle standard deviations of these variables

were used as linear motor variability metrics (ROMSD, Peak VelSD, Avg VelSD, Time PVSD, AreaSD). These linear

metrics reflect the size of variability of shoulder elevation and elbow flexion movements (Madeleine, Mathiassen, and

Figure 1. Experimental set-up.

Ergonomics 3


5/13

Arendt-Nielsen2008; Madeleine, Voigt, and Mathiassen2008).For each subject, the cycle-to-cycle standard deviation of

each variable was first computed across 20 repetitions to each target, and the median value across all 8 targets was used as

an estimate of the cycle-to-cycle standard deviation of each variable. In addition to these variables, time-normalised angles

were calculated for each cycle by time normalising the angle recordings and expressing them in terms of 11 equidistant

samples, such that each normalised data point represented ,10% movement time (cf. Chau, Young, and Redekop2005).

Similarly, time-normalised phase was also calculated and expressed in terms of 11 samples/cycle, where phase at any time

instant was defined as (cf. Hamill et al.1999):

f tan21unorm

vnorm;

where unorm is joint angle normalised to maximum joint angle in the cycle and vnorm is joint velocity normalised to

maximum velocity in the cycle

The standard deviation across all cycles of time-normalised angle and time-normalised phase were calculated for each

target, and their median values across all eight targets were computed as Time norm AngleSD and Time norm PhaseSD.

Structural variability was quantified using sample entropy and recurrence map analysis. Both methods of computing the

structure of motor variability have been used in previous studies of properties of motor control during occupationally

relevant tasks (Madeleine and Madsen 2009). All pipetting cycles to each target were divided into three sets with

approximately equal number of cycles concatenated together in each set. The sets were offset-corrected by subtracting their

mean value, and normalised to their own standard deviation. Sample entropy (SaEn) (Richman and Moorman 2000),

percentage of determinism in the recurrence map (DET) and Shannon entropy of the sequential recurrent map (RMEn)

(Webber and Zbilut1994,2005) were calculated, first to each target in each set and then averaged across the sets. Median

values of the indices across all eight targets were computed and used as non-linear metrics of motor variability in shoulder

elevation and elbow flexion movements.

In addition to these motor variability metrics, liquid transfer times, i.e. the time to pick up liquid from the pick-up tube

and deliver it into the designated target tube, were computed as a measure of temporal motor behaviour. Liquid-transfer-

time variability (cycle-to-cycle standard deviations of transfer time) and task performance were computed and compared

between the two pacing conditions. Task performance was quantified by counting the total number of mistakes made by

each participant in a complete 160-cycle session, i.e. the sum of the number of tubes missed in the pipetting sequence and

the number of times liquid was pipetted to a wrong tube.

2.3. Statistical analysis

The statistical analysis was performed in MATLABw (release 2012, The MathWorks, Inc., Natick, MA, USA). A total of 10

dependent variables representing different kinematic properties were used to describe shoulder elevation and elbow flexion

variability (cycle-to-cycle SD of range of motion, peak velocity, average velocity, time to peak velocity, time-normalised

joint angle, time-normalised phase of angle vs. angular velocity, sample entropy, percentage of determinism and Shannon

entropy of the recurrence map). Since most of the variables showed low-moderate correlations between each other, we ran a

multivariate analysis including all 10 dependent measures.

Neither the linear metrics nor the non-linear metrics of motor variability were normally distributed. Hence, a

multivariate generalised spatial signed-rank test (Mottonen and Oja 1995; Oja and Randles 2004; Mottonen, Oja, and

Serfling2005) was applied to test for the effect of pace, based on paired comparisons of all 10 variables together in the slow-

paced versus the fast-paced condition (p , 0.05 was considered statistically significant). Post-hoc analysis was performed

for each variable using univariate Wilcoxon signed-rank tests for the 10 variables, with Holms sequentially rejective

procedure to adjust for multiple tests (Holm1979). Holms adjustment implies that the result for any particular variable is

considered significant only if the corresponding p-value is less thanp/n, wherepis the univariate test result and nis the rankof that particular test among all univariate tests according to p. For instance, the variable with the smallest p-value of all 10

tested variables needs to have a p , 0.05/10 0.005 to be considered statistically significant.

Since Cohens standardised effect sizes are not justified due to the non-parametric approach for data documentation and

analysis, we computed an analogous metric, to express changes in median values normalised by the dispersion among

subjects, expressed as interquartile range:

ESmedian value in fast paced condition2median value in slow paced condition

IQR ; 1

where IQR difference between the 75th and 25th percentile values of the variable in the slow-paced condition.



6/13

Apart from the motor variability metrics, univariate Wilcoxon signed-rank tests were run separately on liquid-transfer-

time variability and task performance to test for the effect of pacing condition on these variables ( p , 0.05 was considered

statistically significant).

3. Results

Averaged over all cycles, targets and subjects, the group mean liquid transfer time (SD) was 1.2 (0.2) seconds in the slow-

paced condition and 1.0 (0.2) seconds in the fast-paced condition, confirming the controlled experimental change in average

work pace. Liquid-transfer-time variabilities, i.e. cycle-to-cycle standard deviations of transfer times, expressed as median

(interquartile range) were 0.25 (0.08) in the slow-paced condition and 0.17 (0.07) in the fast-paced condition. Wilcoxon

signed-rank test indicated that this trend of liquid-transfer-time variability decreasing with increased pace was indeed

statistically significant (z 4.6, p , 0.0001), and effect size calculated according toEquation (1)was 0.9.

Performance results indicated that the number of mistakes made by the participants, expressed as median (interquartile

range), were 0(1) in both the slow-paced and the fast-paced conditions, and that there were no significant differences in

performance due to pace. Of all cycles, 9.7% deviated by more than 10% from the set cycle time, confirming a very

consistent performance according to task requirements.

While the focus of this study is motor variability, the mean values of kinematic parameters are shown in Figure 2for

descriptive purposes, in order to understand the nature of the pipetting task and the average kinematics properties of

movements performed in each pacing condition. From the mean values for shoulder ( Figure 2a) and elbow (Figure 2b) joint

angles, it can be seen that when pace was increased, the average range of motion decreased, peak and average velocities andthe time to peak velocity increased, and area under the movement curve decreased. Paired t-tests of each variable between

the slow- and fast paced-conditions indicated that only the decrease in the range of motion and area under the movement

curve were significant for shoulder elevation, whereas most similar changes with pace were significant for elbow flexion

movements, except for the time to peak velocity.

Box plots of all motor variability metrics at both paces are shown in Figures 3and 4. With a few exceptions such as

cycle-to-cycle standard deviation of time to peak velocity of shoulder elevation and sample entropy of elbow flexion, the

general trend in these figures showed a decrease in the size and structure of motor variability when pace was increased

during pipetting. The order of pacing conditions in the experiment had no significant effect on these results (p 0.13).

The multivariate signed-rank test showed that the effect of pace was significant for shoulder elevation variability

(p 0.0004). Post-hoc tests were performed using univariate signed-rank tests with Holms adjustments to see which of the

outcome variables were significantly different between the two conditions. The results of the post-hoc analyses, as well as

effect sizes calculated according toEquation (1)are presented inTable 1.It can be seen that the cycle-to-cycle standard

deviation of area under the movement curve (AreaSD), the time-normalised angle and phase variabilities, sample entropy,DET and RMEn were significantly smaller in the fast-paced condition when compared to the slow-paced condition, with

effect sizes mostly in the moderate range, except for RMEn, which was found to be high.

Similarly, the multivariate signed-rank test for elbow flexion indicated that the effect of pace was also significant on

elbow flexion variability (p 0.0005). Post-hoc tests with univariate signed-rank tests using Holms adjustments were then

performed and the results are shown inTable 1. Cycle-to-cycle standard deviations of time to peak velocity (Time PVSD),

area under the movement curve (AreaSD), time-normalised angle and phase variabilities, and DET and RMEn were

significantly lower in the fast-paced condition than in the baseline condition. Effect sizes showed large effects for Area SD

and RMEn, and the other significantly affected variables showed moderate effect sizes.

Thus, even though subjects differed considerably in the extent of motor variability, as illustrated by the wide ranges for

most metrics inFigures 3and4,most of the variables were sensitive to pace, as suggested by the significant tests for a pace

effect and the correspondingly moderate-high effect sizes.

4. Discussion

Pipetting work is a repetitive task requiring low force levels and high precision (Park and Buchholz 2013). We used

pipetting as an experimental model in this study not just to investigate the kinematics attributes of that taskper se, but as a

generalised representation of short-cycle repetitive work occurring in occupational life, which is suspected to be associated

with an increased risk for disorders in the upper extremities, shoulders and neck (Bjorksten, Almby, and Jansson1994;

David and Buckle1997). Short-cycle manual precision tasks may, for instance, also be found in several branches in industry

(Fallentin et al.2001; Hansson et al.2006; Nordander et al.2008), in food processing (Ohlsson et al.1994; Juul-Kristensen

et al.2002) and in letter sorting (Thomsen et al. 2002). Even though the explicit numerical results presented in this paper

may be valid strictly only for pipetting work performed under the current experimental conditions, we believe that the

Ergonomics 5
http://-/?-http://-/?-http://-/?-http://-/?-


7/13

general effects of work pace on motor variability found in this study will apply even to other kinds of short-cycle repetitive

manual handling tasks.

Thirty-five healthy female subjects, who acted as their own controls, were tested to study the effect of pace on motor

variability. The subjects were experienced in pipetting, and further performed a practice session of 100 pipetting cycles

prior to data collection, so that learning effects would not mask the effects of interest in the study. It was verified that all

subjects were sufficiently experienced and proficient in the task by seeing that the median number of mistakes was zero. The

Figure 2. (a) Mean values in slow- and fast-paced pipetting of kinematics properties of shoulder elevation movements, averaged over allcycles, targets and subjects, with error bars representing between-subjects standard deviations; significant p-values are shown for

results of paired t-tests of slow-paced versus fast-paced conditions at a 0.05. (b) Corresponding properties of elbow flexionmovements. Note that the numerical scale on they-axes applies to different variables, as indicated with the identification of the bars.



8/13

fact that performance was almost perfect in both the slow-paced and fast-paced conditions suggested that speed-accuracy

trade-offs were not an issue in this experimental task.

As pace increased, liquid-transfer-time time decreased, and the peak and average velocities of shoulder and elbow

movements increased, as a direct consequence of the reduction in time available to perform the task. The average range of

Figure 3. Motor variability of shoulder elevation movements in slow-paced and fast-paced conditions for each of the 10 metrics, asshown on thex-axis. They-axes show the size of cycle-to-cycle standard deviation in appropriate units for the linear metrics and the sizeof variability in the non-linear metrics. Box plots indicate median, 25th and 75th percentiles, and minimum and maximum values.*Significant difference (p , 0.05) between the two paces in univariate post-hoc tests (see text for further explanation and Table 1forexactp-values).

Figure 4. Motor variability of elbow flexion movements in slow-paced and fast-paced conditions for each of the 10 metrics, as shownon the x-axis. The y-axes show the size of cycle-to-cycle standard deviation in appropriate units for the linear metrics and the size ofvariability in thenon-linear metrics. Box plots indicate median, 25th and75th percentiles,and minimum andmaximum values. *Significantdifference (p , 0.05) between the two paces in univariate post-hoc tests (see text for further explanation and Table 1for exactp-values).

Ergonomics 7


9/13

Table1.

Medianvaluesofshoulderelevationandelbowflexionvariabilitymetricsintheslow-andfast-pacedconditions,resultsofWilcoxonsigned-ranktestsfortheeffectofpace

oneachvariable(z-statisticand

correspondingp-valuesfromtestsshown),aswellaseffectssizesforsignificantef

fects(cf.Equation1).

Shoulderelevation

Elbowflexion

Median(IQR):

slow-paced

Median(IQR):

fast-paced

z

p

Effectsize

(ES)

Median(I

QR):

slow-paced

Median(IQR):

fast-paced

z

p

Effectsize

(ES)

ROMSD

(8)

4.3(2.6)

3.4(1.9)

1.91

0.06

3.7(1.4)

3.5(1.8)

2.04

0.04

PeakVelSD

(8/s)

13.4(5.9)

11.9(6.4)

1.11

0.26

15.2(4.6)

14.4(3.8)

2

0.20

0.84

AvgVelSD

(8/s)

6.4(2.5)

5.6(3.4)

0.47

0.63

8.6(2.1)

7.6(3.2)

0.11

0.91

TimePVSD

(%)

11.2(2.9)

12.1(3.8)

2

1.70

0.10

5.8(2.0)

5.0(1.6)

3.04

0.0

02

0.4

3

AreaSD

(8s)

7.7(4.1)

5.8(3.4)

3.78

2014 the size and structure of arm movement variability decreased with work pace in a standardised...

Documents