ARTICLE IN PRESS

1466-853X/$ - s

doi:10.1016/j.pt

�CorrespondE-mail addr

Physical Therapy in Sport 7 (2006) 185–190

www.elsevier.com/locate/yptsp

Original research

Repeated single-limb postural stability testing elicits a practice effect

Jonathan Warrena, Anthony G. Schneidersb,�, S.John Sullivanb, Melanie L. Bellc

aKennedy Road Physiotherapy, Napier, New ZealandbSchool of Physiotherapy, University of Otago, , PO Box 56, Dunedin 9015, New Zealand

cDepartment of Preventive and Social Medicine, University of Otago, New Zealand

Received 10 August 2005; received in revised form 6 June 2006; accepted 23 June 2006

Abstract

Objective: To document the effects of repeated testing on the single-leg stance balance task.

Design: Single cohort repeated measures.

Setting: Laboratory in an educational institution.

Subjects: Thirty-two healthy males and females.

Outcome measure: The number of errors (deviations from the required posture) during each 20-s trial summed over the eight

conditions recorded on six occasions.

Results: There was a statistically significant (p ¼ .0013) decrease in the number of errors recorded over the six sessions, from 26.8

(95% CI: 23.1–30.5) to 19.7 (95% CI: 16.3–23.1). Linear regression confirmed a systematic decrease of 1.5 errors per session on

average (95% CI: 1.0–1.9; po.0001).

Conclusion: The decreased number of errors (increased performance) with repeated testing alerts clinicians to the need for care when

using this test protocol to measure rehabilitation interventions.

r 2006 Elsevier Ltd. All rights reserved.

Keywords: Balance; Practice effects; Single-leg stance; Measurement

1. Introduction

It is believed that sensorimotor control is alteredfollowing injury to articular structures of the lower limb(Bernier & Perrin, 1998; Goldie, Evans, & Bach, 1994).A range of postural stability assessments are commonlyused by sports medicine clinicians to evaluate thesedeficits including; force platforms systems generatingstability and sway parameters (Rozzi, Lephart, Sterner,& Kuligowski, 1999); joint position and kinaesthesia(Forkin, Koczur, Battle, & Newton, 1996) and tradi-tional measurements of single-leg stance sway (Hals,Sitler, & Mattacola, 2000). Each of these approaches hasits advantages and disadvantages. For example, theclinical assessment of joint position sense can bedifficult and its validity has been questioned (Beynnon,

ee front matter r 2006 Elsevier Ltd. All rights reserved.

sp.2006.06.002

ing author. Tel.: 64 3 479 7460; fax: 64 3 479 8414.

ess: tony.schneiders@otago.ac.nz (A.G. Schneiders).

Renstrom, Konradsen, Elmqvist, Gottlieb, & Dirks,2000) while the cost of sophisticated balance systemsmay limit access for many practitioners.

One common clinical approach to the measurement ofsensorimotor control of the lower limb is that of single-leg stance tests (SLST) where the patient is required tomaintain balance while standing on one leg (Forkin etal., 1996; Goldie, Evans, & Bach, 1992; O’Connell,George, & Stock, 1998; Rozzi et al., 1999). These testswere derived from the standard Romberg protocol andmodified by Freeman, Dean, and Hanham (1965) toinclude conditions of single-leg stance which weresuggested to assess proprioceptive deficits in the anklefollowing injury. These authors introduced the phrase‘‘articular deafferentation’’ to describe a sensorimotordeficit in the terminal mechanoreceptors of ankleligaments and capsule, leading to functional instabilityand impairment. The tests are usually timed andqualitatively assessed. Recently the quantification of

ARTICLE IN PRESSJ. Warren et al. / Physical Therapy in Sport 7 (2006) 185–190186

the SLST has been expanded to include movementdeviations from a specified stance position. Riemann,Guskiewicz, and Shields (1999) developed the balanceerror scoring system (BESS) as an observational methodof evaluating athletes’ postural stability without the useof complex or expensive equipment. The BESS is amovement quantification approach applied to a batteryof clinical balance tests including the SLST andperformed on different surfaces (firm or unstable). Thetest protocol requires the subject to maintain a definedposture with their eyes closed for a specific time (20 s)and predefined movement deviations from this postureare scored as errors and the total number of errorsrecorded as the subject’s score.

The concurrent validity of the BESS was investigatedby comparing the BESS scores with those obtained fromsimultaneous force-platform sway measures in a groupof non-injured subjects. The resulting modest correla-tions indicated a low to acceptable level of agreementbetween the two measurement approaches for five of thesix test conditions and the inter-tester reliability wasfound to be high (Riemann et al., 1999). The BESS hasbeen subsequently used clinically to assess posturalstability in athletes following mild head injury (MHI)(Riemann & Guskiewicz, 2000).

Riemann et al. (1999) stated that the BESS was astable measure with repeated testing sessions and thatchanges observed in BESS scores were most probablydue to resolution of the underlying impairment. Thisconclusion has been questioned by Valovich, Perrin, andGansneder (2003) who reported ‘‘practice effects’’(reduced BESS scores) when the BESS was used toassess school-aged athletes on five occasions extendingover a 30-day interval. Four of the five administrationswere concentrated over a 7-day period, where the scoresat days 5 and 7 were lower than those recorded atbaseline. The practice effect was most pronounced withtesting single-leg stance performed on an unstable foamsurface. A subsequent study (McLeod, Perrin, Guskie-wicz, Schultz, Diamond, & Gansneder, 2004) found asimilar learning effect which was also detected 60 daysfollowing the initial assessment. It was noted thatclinicians should exercise caution when using repeatedadministrations of the BESS protocol to measurepostural stability as the measurement may reflectpotential practice effects. This caution also applies inother clinical applications where a treatment may bemonitored and or measured serially. To date, noinvestigations have reported using the BESS protocolas an assessment tool following lower-limb musculoske-letal injury. This is surprising as variations of the SLSTare commonly used to measure sensorimotor deficits inathletes and to monitor recovery following clinicalinterventions (Forkin et al., 1996; Rozzi et al., 1999).It is therefore important to establish the temporalstability of the BESS scoring approach with repeated

administration in a non-injured population beforeextending its use to assess patients with musculoskeletalinjuries.

The aim of this study was to determine if the single-legstance performance, as measured by the BESS scoringprotocol, was stable over repeated testing sessions innon-injured persons.

2. Methods

2.1. Study design

A repeated measures design was used to investigatechanges in the performance on the single-leg stance taskover six identical measurement sessions during a 2-weekperiod.

2.2. Subjects

Thirty-two students and staff of a New Zealandtertiary education institution volunteered to participatein this study. Subjects were informed of the nature of theresearch and its requirements and were free to withdrawfrom the study at any time. No subjects who reportedlower limb, back, neurological, otologic conditions or ahead injury in the previous six months were admitted tothe study. The study was approved by the University ofOtago Human Ethics Committee.

2.3. Equipment

The testing protocol took place in a quiet roomand required the subjects to stand on either a solid flooror on an unstable surface consisting of a piece(46� 46� 12.5 cm thick) of medium density foam(density ¼ 55 kg/m3—manufacturer’s estimate). To en-able precise post hoc analysis of the balance taskperformance, all the trials were videotaped using adigital video camera (Sony DCR-TRV33E) mounted ona tripod at a fixed distance from the subjects.

2.4. Test protocol

2.4.1. Single-leg stance test (SLST)

The experimental task investigated was the SLST, amodification of the classic Romberg test (Freeman et al.,1965), and a task widely used in clinical practice toassess balance in patients who have sustained an injuryto a lower limb. Subjects were required to stand on oneleg with their hands placed on their iliac crests with thecontralateral knee flexed in a comfortable position, thehip in neutral, and the legs not touching. Subjects wereinstructed to remain as motionless as possible in thedesignated standing position during a 20-s scoringperiod with their eyes either open or closed. They were

ARTICLE IN PRESSJ. Warren et al. / Physical Therapy in Sport 7 (2006) 185–190 187

also advised that if they lost their balance, they shouldendeavour to return to the test position as quickly aspossible. Each subject completed four variations of theSLST test (Table 1) on both the left and right legs givinga total of eight scored tests. In all instances the subjectscompleted the test barefoot while standing on the firmsurface or on the foam with their eyes either open orclosed.

2.4.2. Scoring

Performance of the SLST was evaluated using themeasurement developed as part of the BESS (Riemannet al., 1999). Unlike the BESS, we chose to only measuresingle-leg stance but with the eyes open and closed. Thekey focus of the BESS measurement protocol is thedocumentation of any gross movement deviation fromthe original testing posture, or the opening of the eyes.The scoring elements defined in the BESS and adoptedfor the present study were as follows: lifting hands offiliac crests; opening eyes (for the eyes closed conditiononly); stepping, stumbling, or falling; moving thecontralateral hip more than 301 into flexion or abduc-tion; lifting the grounded forefoot or heel; and remain-ing out of the test position for more than 5 s.

Subjects were scored one error point for eachdeviation observed during the 20-s trial. In addition torecording the number of errors for each scoring elementfor each condition, the total score (number of errors) foreach subject was recorded for the eight trials during eachof the six sessions. In keeping with the original BESSprotocol, a trial was deemed incomplete if subjects couldnot maintain the testing position without committing anerror during a minimum time period of 5 s (Riemann etal., 1999) and these trials were assigned a ‘‘maximumerror score’’, defined as one error point above themaximum score recorded during a complete trial forthat particular day. In order to facilitate the scoringprocess by providing the opportunity for detailed andretrospective analyses, each trial was videotaped andanalysed by the principle investigator (PI), an experi-enced physiotherapist, at the completion of testing.

2.5. Procedure

Subjects were briefed on the purpose of the study andintroduced to the various aspects of the task beforecommencing the initial testing session. The order oftesting was randomised for each subject for each session.In order to allow the foam to maintain its physicalintegrity it was rotated and repositioned following eachtrial. To enable the best definition and to easilydetermine if the eyes were open or closed the videocamera was orientated in portrait style. The digitalcounter on the video camera was used for timing the 20 sof each test.

The subjects were tested three times a week (Monday,Wednesday and Friday) for 2 weeks resulting in a totalof six sessions. This time frame was chosen to replicatethe approximate treatment frequency of a patientreceiving physical therapy for a lower limb musculoske-letal injury. The subjects were instructed not to practicethe SLST between testing sessions and this was verifiedand reinforced by the PI at each session.

2.6. Statistical analysis

The videotape recordings of each subject wereanalysed and errors assigned for each condition withthe total number of errors calculated. The eightindividual test scores were highly variable so the totalscore was used. There were concerns that the scoringwas not interval scaled therefore non-parametric meth-ods (Wilcoxon signed rank test) were used to confirmthe primary parametric results.

Mixed models, as implemented in the SAS procedure‘‘mixed’’ (SAS Institute, 1999), were used to take intoaccount the repeated measures nature of the studydesign. These allowed for subject-to-subject variation inthe outcome of total score versus time, as well as the factthat measurements within a subject were generally morehighly correlated than between subjects. Additionally,these models accommodated missing and unequallyspaced data. Session was considered both as a discretefixed effect (ANOVA) and a continuous fixed effect(linear regression), with both models using subject as arandom effect. The ANOVA explicitly estimated meansat each day without forcing consecutive days to haveequal changes, while the regression model estimated acommon slope. Pairwise testing of the individual scoreat each session was conducted in the ANOVA, with aTukey correction for multiple comparisons. The totalnumber of errors for each element of the scoring systemwas determined for descriptive purposes.

3. Results

The characteristics of the subjects are presented inTable 2. There was heterogeneity in the composition ofthe sample of convenience with a wide range of ageswhich were considered to be representative of thespectrum of ages generally seen by physical therapistsin patients presenting with lower-limb dysfunction.While all subjects were compliant in attending the firsttwo sessions, the number of dropouts over the course ofthe study were three, two, one and four for sessionsthree, four, five and six, respectively. The combinedreasons for this included study and work pressure,forgetfulness and adverse weather conditions; however,these subjects were included in the statistical analysis

ARTICLE IN PRESS

Table 1

Single-leg stance test (SLST) conditions

Surface

Firm surface

(stable/static)

Foam surface

(unstable/dynamic)

Eyes open SLSTa SLSTa

Eyes closed SLSTa SLSTa

aLeft and right legs.

Table 2

Subjects characteristics (n ¼ 32)

Age (years) Weight (kg) Height (cm) Sex (F/M)

Mean 28.6 71.4 170.4

SD 11.2 13.2 11.1

Range 18–57 54–94 152–191

Number 20/12

1 2 3 4 5 6

0

10

20

30

40

50

60

Session

Tot

al S

core

Fig. 1. Subjects total scores (BESS) for the eight conditions. *Group

mean (SE) in bold.

J. Warren et al. / Physical Therapy in Sport 7 (2006) 185–190188

as they contributed information to the study for thesessions they did complete.

The total scores comprised of the scores from theeight test conditions, for the 32 individuals showedconsiderable variability over the six sessions. There wasalso a wide spread of baseline scores at session 1 (Fig. 1).The ANOVA identified a significant main effect ofsession (po.0001) with statistically significant differ-ences located between session 1 and sessions 3, 4, 5 and6, and between session 2 and sessions 3, 5 and 6,collectively indicating a decrease in score with repeatedtesting. This decrease in errors was confirmed with thenon-parametric Wilcoxon signed rank test between

sessions 1 and 6 (po.0001) and sessions 2 and 6(po.0001). The regression model identified a slope of�1.5 (95% CI: �1.9 �1.0) indicating an averagedecrease of 1.5 errors per session.

An analysis of the errors for each scoring element forthe 22 subjects who completed the six sessions (N ¼ 22,conditions ¼ 8 and sessions ¼ 6) revealed 1357 codedBESS errors distributed as follows; lifting hands ¼ 44(3.2%), opening eyes (in eyes closed condition) ¼ 26(2.0%), step/stumble/fall ¼ 713 (52.5%), hip move-ments ¼ 331 (24.2%), lifting forefoot or heel of thegrounded foot ¼ 233 (17.2%), and remaining out ofposition for 5 s ¼ 10 (2.0%).

4. Discussion

The primary aim of this study was to determine ifthere was a practice effect associated with the repeatedadministration of the SLST when scored using the BESSscoring protocol. The data clearly demonstrated adecrease in ‘‘BESS’’ scores (improvement) over the sixtesting sessions (2-week period). The systematic changeobserved indicates that any inferences made from repeatadministrations of the test in order to monitor recoveryor the impact of a physiotherapy intervention should betreated with caution.

These data are in keeping with those of Valovich et al.(2003) who reported a practice effect (decrease) in BESSscores at days 5 and 7 (sessions 3 and 4) in comparisonto a baseline performance for the BESS balance tests(single-leg stance, double-leg stance, and tandemstance). These authors also reported that the practiceeffect was highlighted in the single-leg stance on foamcondition, in what may be considered a more challen-ging (complex) task. A subsequent study documentedchanges at days 5, 7 and 60 in comparison to a baselinecondition (McLeod et al., 2004). Collectively, thesechanges in performance with repeated testing is contra-dictory to an earlier report by Riemann et al. (1999) whofound no evidence of a practice effect over threeadministrations of the BESS.

It is perhaps not unexpected that there might bechanges with repeat or serial administrations, as a singletrial for each of the SLST conditions is at best amomentary snapshot of the subject’s performance andmost probably not a realistic measure of their ‘true’score. The fact that the data presented here are notappearing to plateau over the time course of the study(six administrations) suggest that the subject is mostprobably still exploring potential learning strategies andhas not reached a stable level of performance, thus notallowing an accurate measurement of the performance.Learning cannot, however, be inferred from these dataas there is no way of measuring whether these changesare permanent (Schmidt, 1988).

ARTICLE IN PRESSJ. Warren et al. / Physical Therapy in Sport 7 (2006) 185–190 189

Whether or not the changes reflect a learning orpractice effect, the implications are profound. Theuse of the SLST and the accompanying BESS protocol(and perhaps its variants) to monitor recovery frominjury or to monitor a particular intervention must bequestioned as any inferences may be contaminated,as noted previously by Valovich et al. (2003). Whiletheir point that ‘‘athletic trainers should expect theirathletes to improve on subsequent test administrations’’is noted, perhaps it should be extended to cautionclinicians as to possible misinterpretation of dataresulting from the use of the BESS during balancetesting. This is particularly important with the growingadvocacy for the use of the BESS protocol in theassessment of players who may have been concussed(Guskiewicz, Bruce, Cantu, Ferrara, Kelly, McCrea, etal., 2004). Similar implications apply in situations wherebalance tests may be used to monitor an intervention.The use of assessment procedures which employ a singleor limited number of trials as the basis of theirassessment protocol must be viewed cautiously, parti-cularly if they are to be used to monitor the performanceof an individual athlete or client.

The individual SLST data reported in this study areextremely variable and may reflect both the exploring ofappropriate movement strategies and an indication ofthe heterogeneity of the subjects. The inclusion criteriawere broad so as to include a wide range of subjects(age, gender, height, etc.) to approximate the typicalclient base who may seek physical therapy treatment,thus allowing the generalisation of the results. Nearly athird of the subjects who entered the study did notcomplete the six sessions. However, there is no reason tosuspect that there was any systematic withdrawal ofsubjects which might have impacted on the results.Subjects withdrew from the study for a wide range ofreasons, none of which could be attributed to theinability to perform the balance task. The heterogeneityof the group provides a sound sample on which to assessthe potential use of this balance protocol with muscu-loskeletal clients.

Balance is the process of maintaining the centre ofgravity within the body’s base of support (Nashner,1993) and the maintenance of balance is controlled bythe central integration of the sensory and motor systemswith the visual systems playing an important role(Diener & Dichigans, 1988; Fitzpatrick, Rodgers, &McCloskey, 1994). The BESS is based upon movementanalysis and how a subject responds to the challenges ofmaintaining their balance during a fixed period of time(20 s). Other approaches to the clinical quantification ofbalance include timing how long a prescribed balanceposture can be maintained (Emery, Cassidy, Klassen,Rossychuk, & Rowe, 2005) and computerised systems(Isakov & Mizrahi, 1997). The BESS defines a numberof postural elements (e.g. removing the hands from the

hips), which are recorded by the clinician as ‘errors’ andthen summed to provide an overall error score. Thedefined error elements also provide some insight intohow the subject responds to the ongoing challenge ofmaintaining their balance. In this study, which used avideo recording of each subject’s performance it waspossible to examine in some detail the BESS errorsdemonstrated by each individual. The most commonlyrecorded error was ‘stepping or stumbling’ followed by‘hip movements’ and then the ‘lifting the forefoot or heelof the grounded foot’ Surprisingly the subjects did notattempt to use vision by opening their eyes in the eyesclosed condition as much as might be expected. TheBESS assumes that each error element is of identicalweighting in computing the overall score. Our datasuggest that some error elements may be more likely tooccur than others and this questions the use of acomposite score approach—at least in the groupinvestigated here. However, this may not be truefor patients with musculoskeletal and/or neurologicalinjuries.

If the SLST and related BESS measurement protocolare to be used in the assessment of balance in personswith lower-limb musculoskeletal injuries, then furtherresearch with this specific patient group is clearlynecessary. Impaired balance has been documented inconjunction with lower limb ligamentous injuries andthe use of the SLST is suggested as a measure of balancein these conditions (Cornwall & Murrell, 1991; Forkin etal., 1996; Jerosch & Bischof, 1996; O’Connell et al.,1998), but clearly further research is required todetermine the ideal test protocol. For example, it isnecessary to establish how many trials are requiredbefore a stable score is reached and perhaps the numberof test conditions may need to be reduced to focus onthe possible deficits in proprioceptive input due to thenature of the musculoskeletal injury. Likewise, the errorelements may need to be revised to take into account thelikelihood of their occurrence. This information isnecessary before the clinician can make an informedchoice on the use of the SLST to monitor recovery frominjury or to assess the success of an intervention in theclient with a musculoskeletal injury.

5. Conclusion

The findings of this study indicate that single-legstance performance as measured by the BESS scoringprotocol is not stable over repeated testing sessions innon-injured persons. The SLST protocol is widely usedby clinicians as it is simple and requires minimalequipment and training. While it may provide a firstapproximation of balance performance, its use tomeasure an intervention or to track recovery must be

ARTICLE IN PRESSJ. Warren et al. / Physical Therapy in Sport 7 (2006) 185–190190

interpreted with caution and is best used in this contextin conjunction with other performance criteria.

Conflict of interest statement: The authors of thismanuscript (Jonathan Warren, Anthony G. Schneiders,S. John Sullivan and Melanie L. Bell) declare no conflicts

of interest due to any financial and personal relation-ships with other people or organisations that couldinappropriately influence (bias) this work, all within 3years of beginning the work submitted.

Acknowledgement

The authors wish to thank staff and students of theEastern Institute of Technology, Napier, New Zealand,who facilitated and participated in the study.

References

Bernier, J. N., & Perrin, D. H. (1998). Effect of coordination training

on proprioception of the functionally unstable ankle. Orthopaedic

Sports Physical Therapy, 27(4), 264–275.

Beynnon, B. D., Renstrom, P. A., Konradsen, L., Elmqvist, L.-G.,

Gottlieb, D., & Dirks, M. (2000). Validation of techniques to

measure knee proprioception. In S. C. Lephart, & F. H. Fu (Eds.),

Proprioceptive and neuromuscular control in joint stability (pp. 127–138).

Champaign, IL: Human Kinetics.

Cornwall, M. W., & Murrell, P. (1991). Postural sway following

inversion sprain of the ankle. Journal of the American Podiatry

Association, 81(5), 243–247.

Diener, H. C., & Dichigans, J. (1988). On the role of vestibular, visual

and somatosensory information for dynamic postural control in

humans. Progress in Brain Research, 76, 253–262.

Emery, C. A., Cassidy, J. D., Klassen, T. P., Rossychuk, R. J., &

Rowe, B. H. (2005). Development of a clinical static and dynamic

standing balance measurement tool appropriate for use in

adolescents. Physical Therapy, 85(6), 502–514.

Fitzpatrick, R., Rodgers, D. K., & McCloskey, D. I. (1994). Stable

human standing with lower-limb muscle afferents providing the

only sensory input. Journal of Physiology, 480, 395–403.

Forkin, D. M., Koczur, C., Battle, R. M. S., & Newton, R. A. (1996).

Evaluation of kinesthetic deficits indicative of balance control in

gymnasts with unilateral chronic ankle sprains. Journal of

Orthopaedic and Sports Physical Therapy, 23(4), 245–250.

Freeman, M. A. R., Dean, M. R. E., & Hanham, I. W. F. (1965). The

etiology and prevention of functional instability of the foot.

Journal of Bone and Joint Surgery, 47B, 678–685.

Goldie, P. A., Evans, O. M., & Bach, T. M. (1992). Steadiness in one-

legged stance: Development of a reliable force-platform testing

procedure. Archives of Physical Medicine and Rehabilitation, 73,

348–354.

Goldie, P. A., Evans, O. M., & Bach, T. M. (1994). Postural control

following inversion injuries of the ankle. Archives of Physical

Medicine and Rehabilitation, 75, 969–975.

Guskiewicz, K. M., Bruce, S. L., Cantu, R. C., Ferrara, M. S., Kelly, J.

P., McCrea, M., et al. (2004). National athletic trainers’ association

position statement: Management of sport-related concussion.

Journal of Athletic Training, 39(3), 280–297.

Hals, T. M. V., Sitler, M. R., & Mattacola, C. G. (2000). Effect of a

semi-rigid ankle stabilizer on performance in persons with

functional ankle instability. Journal of Orthopaedic and Sports

Physical Therapy, 30(9), 552–556.

Isakov, E., & Mizrahi, J. (1997). Is balance impaired by recurrent

sprained ankle? British Journal of Sports Medicine, 31, 65–67.

Jerosch, J., & Bischof, M. (1996). Proprioceptive capabilities of the

ankle in stable and unstable joints. Sports Exercise and Injury, 2,

167–171.

McLeod, T. C. V., Perrin, D. H., Guskiewicz, K. M., Schultz, S. J.,

Diamond, R., & Gansneder, B. M. (2004). Serial administration of

clinical concussion assessments and learning effects in healthy

young athletes. Clinical Journal of Sports Medicine, 14(5), 287–295.

Nashner, L. (1993). Practical biomechanics and physiology of balance.

Handbook of balance function and testing. St. Louis: Mosby Year

Book.

O’Connell, M., George, K., & Stock, D. (1998). Postural sway and

balance testing: A comparison of normal and anterior cruciate

ligament deficient knees. Gait and Posture, 8(2), 136–142.

Riemann, B. L., & Guskiewicz, K. M. (2000). Effects of mild head

injury on postural stability as measured through clinical balance

testing. Journal of Athletic Training, 35(1), 19–25.

Riemann, B. L., Guskiewicz, K. M., & Shields, E. W. (1999).

Relationship between clinical and forceplate measures of postural

stability. Journal of Sport Rehabilitation, 8, 71–82.

Rozzi, S. L., Lephart, S. C., Sterner, R., & Kuligowski, L. (1999).

Balance training for persons with functionally unstable ankles.

Journal of Orthopaedic and Sports Physical Therapy, 29(8),

478–486.

SAS Institute. (1999). SAS/STAT V8 Software User’s Guide. Cary,

NC: SAS Institute.

Schmidt, R. A. (1988). Motor control and learning: A behavioural

emphasis (2nd ed). Champaign, IL: Human Kinetics.

Valovich, T. C., Perrin, D. H., & Gansneder, B. M. (2003). Repeat

administrations elicits a practice effect with the balance scoring

system but not with the standardized assessment of concussion in

high school athletes. Journal of Athletic Training, 38(1), 51–56.