Interrater and test-retest reliability of two pediatric balance tests.

Maintenance of postural control is known to be a prerequisite to the performance of skilled movement. In order to complete simple or complex gross and fine motor tasks', a person must be able to maintain his or her center of gravity over the base of support. This postural control must be done by responding quickly and accurately to all internal and external environmental changes. These changes, according to Weisz, can occur under three different conditions: 1) the position of the body can change in space, 2) the position of the extremities can change in relation to the body, or 3) the supporting surface of the body can change. (1) During functional activities, these changes may occur independently or in any combination. This variability creates a variety of conditions that one must be able to sense and react to appropriately.

Children with developmental disabilities may have postural instability, which contributes to delayed or deficient motor skill development. It has been documented that children with learning disabilities, (2,5) hearing impairments, (6,7) cerebral palsy, (8-10) and mental retardation (11-13) have balance deficits.

Bobath has stated that the loss of automatic maintenance of equilibrium causes increased energy expenditure for skilled movement. (14) She postulates that the child must volitionally elicit balance reactions rather than depend on them to occur automatically; thus, coordination and efficiency are decreased. The Bobath approach to therapeutic exercise (neurodevelopmental therapy) focuses on improving postural control to allow children to develop more skilled movements. (8) Most other current therapeutic exercise methods include exercise to improve postural control. (15,17)

A review of present assessment methods of balance in standing reveals few standardized, reliable, and comprehensive measures. Common techniques used by physical therapists and occupational therapists are measurement of the number of seconds an individual can maintain a specific static posture (eg, one-foot balance), measurement of the ability to maintain balance dynamically (eg, walking on a balance beam), and subjective measurement of the maintenance of balance on moveable surfaces (eg, balancing on a tiltboard).

Various tests include standardized procedures for measurement of one-leg standing balance in children 3 to 14 years of age: the Peabody Developmental Motor Scales (PDMS), (18) the Bruininks-Oseretsky Test of Motor Proficiency (BOTP), (19) the Test of Motor Impairment (TOMI), (20) the Southern California Sensory Integration Tests (SCSIT), (21) and the Sensory Integration and Praxis Tests (SIPT). (22) Each test requires children to maintain a variety of standing postures.

These postures include one-leg standing on the floor (eyes open, eyes closed), one-leg standing on the balance beam (eyes open, eyes closed), stork balance (subject stands on one foot and places the sole of the other foot against the side of the supporting knee), and one-leg standing balance on a balance board. When maintaining these postures, the child may be asked to cross the arms, place hands on hips, place arms by the sides, or hold arms straight above the head. Children may or may not be asked to focus on a visual target. Schulmann et al found that asking subjects to visually fixate on a stationary target as compared with tracking a moving target resulted in better standing balance performance. (23) Clark and Watkins also examined availability of visual cues and three other different task demands that might affect balance in normally developing 6- to 9-year-old children. (24) They found that availability of visual cues, body position, and size of base of support significantly affected the one-leg balance performance in these children. Lastly, they found no significant difference in the amount of time children could stand on the preferred leg as opposed to the nonpreferred leg. The authors suggest that, when constructing new balance tests, test developers should consider all of these variables. (24)

Reliability gives information regarding the consistency and stability of a measurement. Systematic error in administration or scoring affects the validity of the data. (25) Every test should include a statement regarding its reliability. (26) Interrater and test-retest reliabilities, on most current standing balance measures, are either not reported or reported to be low. The PDMS and TOMI do not report interrater or test-retest scores item-by-item, so reliability cannot be determined for balance activities. (18,20) Test-retest reliability coefficients on the BOTP balance items have been reported for a group of 63 second-grade and 63 sixth-grade children. Average reliability estimates were obtained by means of Fisher's Z transformation. Coefficients of .64 and .56 were reported for the second- and sixth-grade children, respectively, on the BOTP balance subtests. No interrater reliability specific to balance activities was reported. (18) On the SCSIT, Ayres did not report interrater reliability, but did report test-retest stability. (21) Pearson product-moment correlation coefficients ranged from .55 to .72 with eyes open and .16 to .51 with eyes closed for children aged 4 to 8 years on a one-leg standing task. A total of 239 children were tested, with 36 to 54 subjects in each of the five-year age groups. (21) Ayres' new test, the SIPT, is reported to have shown higher reliability scores than the SCSIT for the standing and walking balance tests (Diana Marr, PhD; personal communication; June 15, 1988). Interrater reliability was determined on 98 children with a resulting correlation coefficient of .98. Test-retest reliability completed on 38 children with learning disabilities and 10 normally developing children yielded a correlation value of .85.

Standardized methods for objective evaluation of equilibrium when moved on a tiltboard are scarce. Fisher and Bundy have described techniques to objectively measure balance reactions of children on several tiltboard tests. (3) They have developed these tests to be functionally related and to examine the three conditions under which the environment changes as described by Weisz. (1) A determination of the quality of the response was made from photographs at the point of tilt just prior to the subject moving an arm or foot to maintain his or her balance. (27) The authors reported high interrater and test-retest reliability for the angle measurement of their tiltboard tip test (n = 37, r = 98). (3) However, the measurements taken from the photographs to document the qualitative responses to the tilt by the child were reported as not showing acceptable interrater reliability. (3) Concurrent validity has also been documented by testing boys with learning disabilities who show different performances on their tiltboard test as compared with children who are developing normally. (5) The authors found that the tiltboard tests did not correlate with the one-leg standing balance tests of the SCSIT. (5) They hypothesized that each test apparently measured a different aspect of the postural response. Therefore, they recommend using a variety of tests to assess postural stability and thus be able to make more appropriate treatment recommendations. (5) These researchers have contributed greatly to this area in regard to specific development of an objective and reliable test. (3,5,27) Their testing procedure allows for reliable analysis of the angle-of-tilt measurements, but requires photographic analysis. A quicker method of measurement, if proved reliable, could be valuable for use by clinicians.

Perham et al also described a tiltboard test used to document development of lateral equilibrium reactions. (28), A tiltboard with stops at 24 degrees of tilt was used for the study. The child's response was rated via photographic analysis and scored according to a three-point scale. These authors report establishment of interrater reliability of speed of tilt (90%) during pilot testing. Analysis of the child's reaction to the tilt was performed by both primary, investigators simultaneously with no report of discrepant scores. (28) Specific reliabilities of independent ratings of the photographs were not reported.

From this review, it is clear that clinical assessment methods for postural stability are used but not extensively researched. The purpose of this study was to examine the consistency and stability, of scores made by physical therapist and occupational therapist observers on measures of one-leg balance and tiltboard balance. Using a sample of normally developing children, the specific research questions addressed were: 1) What are the interrater reliabilities of the one-leg and tiltboard balance measures? and 2) What are the test-retest reliabilities of the one-leg and tiltboard balance measures?

Method

Subjects

Subjects were 24 children, aged 4 through 9 years, from five Headstart preschool classrooms, a private primary school, and a public elementary school. Descriptions of the study and informed consent forms approved by the University of Washington Human Subjects Committee were distributed to all classrooms having children aged 4 through 9 years. Using stratified random sampling for age and sex, subjects were chosen for the interrater reliability portion of the study so that four children (2 boys, 2 girls) were tested in each of the six age groups (4-year-olds, 5-year-olds, and so on). Twelve children--1 boy and 1 girl from each age group--participated in the test-retest reliability portion of the study.

All children except one were without physical, academic, or neurological problems, as documented by a parent questionnaire. One child had a febrile seizure in infancy, but was in a regular classroom. For the interrater reliability portion of the study, 16 of the children were white and 8 were black; for the test-retest reliability, portion of the study, 8 of the children were white and 4 were black.

Examiners

Two primary examiners, a pediatric occupational therapist (PKR) and a pediatric physical therapist (SWA) with eight and nine years' experience, respectively, each scored the children's performance on the tests. Both examiners participated in writing the procedures manual. Prior to beginning data collection for the study, both examiners studied the instruction manual and practiced scoring using a videotaped recording of a child being tested. A third person (TKC) and two occupational therapy students participated as a "secondary examiner" to "spot" the child and to indicate hands off hips or eyes open to the primary examiners during the tiltboard test. Prior to data collection, each secondary examiner's test administration performance was rated on a check list. The secondary examiner was then given feedback on her performance, and the process was repeated until she demonstrated correct procedures related to each item.

Procedure

The primary examiners alternately administered the tests and scored the children's performance. One examiner positioned and instructed the child for the one-leg balance test and positioned and instructed the child and moved the tiltboard for the tiltboard balance test, while the other primary examiner observed and scored the child's performance. Each examiner tested equal numbers of boys and girls at each age level. Silent stopwatches were used, and no cues regarding scoring were exchanged between primary examiners during the testing. One week after the first test, the examiners each retested one child out of two at each age level for whom they had been primary examiner. Again, each examiner retested equal numbers of boys and girls.

The one-leg balance and tiltboard balance tests were administered in a random order as part of a comprehensive standing balance assessment protocol, which includes the Clinical Test of Sensory Interaction on Balance designed by the authors. This measure is based on Shumway-Cook and Horak's research. (29) The children were also briefly evaluated to ensure that their muscle strength and joint range of motion were within normal limits.

One-leg standing balance. One-leg balance was tested in both eyes-open and eyes-closed conditions. The subject was barefoot and was positioned 2 ft* from a wall with hands on hips. A visual target was placed on the wall at eve level. The subject was instructed to lift one foot and look at the target. For the condition of standing balance with eyes open, timing was started as soon as the subject lifted his or her foot. Timing was stopped when the subject touched the free foot to the floor, removed hands from hips, or moved the supporting foot from the original position. Timing also was stopped if, after two warnings, the subject continued to hook the free leg behind the support leg, drop the free leg below 45 degrees of knee flexion, or look away from the visual target. For the eyesclosed one-leg standing balance test, the subject was instructed to lift one foot and, once balance was established, to close the eyes. Timing was started as soon as the subject closed his or her eyes. In addition to the criteria for the eyes-open condition, the criteria for the eyes-closed condition included terminating a trial when a subject opened his or her eyes. After the initial trial on the foot of the child's choice, balance on the opposite foot was scored. This procedure was then repeated so that for both the eyes-open and eyes-closed conditions subjects were scored two times on each foot. The best performance on each foot and in each condition was recorded. There were two reasons why this scoring procedure was followed rather than averaging the two trials. First, a child sometimes has one poor trial, and if the two trials are averaged, the score does not reflect optimal performance. Second, in the clinical setting it is easier to take the highest score rather than to average two scores. In addition to recording the balance time, the examiner recorded the quality of the balance reaction by noting whether the subject used a hip or ankle strategy to maintain balance. (30) An ankle strategy was defined as movement to maintain balance that was initiated from the ankle. A hip strategy was defined as movement occurring first at the hip to maintain balance. A child was recorded as using a hip strategy if one was seen at any time during a trial. if the subject maintained balance for three seconds or less, the strategy used could not be determined and was recorded as such on the data form.

Tiltboard balance. Tiltboard balance testing was also administered in the eyes-open and eyes-closed conditions. The subject was positioned with feet together, medial malleoli opposed, on footprints marked in the center of an 18- x 18-in + tiltboard. The tiltboard had angle markers extending 4 in from the board to the wall, where a chart with degree lines from 0 to 60 degrees was located. The subject was instructed to stand with hands on hips and maintain balance as long as possible while the primary examiner alternately tipped the tiltboard slowly to the right or left. A second examiner guarded against falls and watched for any postural adjustments, especially upper extremity movement (ie, hands off hips), that would result in termination of the trial. Other movements that signaled the end of the trial included raising a foot, stepping, or beginning to fall so as to require support from the secondary examiner. In the eyes-closed condition, the trial also was terminated if the subject opened his or her eyes. Two trials were administered to each side, alternating right and left sides, for both the eyes-open and eyes-closed conditions. The best performance to each side wis recorded. The degree line (in increments of 5 [degrees]) to which the angle marker was closest at the point where the subject made a postural adjustment was recorded, up to the maximum of 60 degrees.

Results

One-Leg Standing Balance Summary statistics, reliability coefficients, and magnitudes of differences between scores for duration of one-leg standing balance are shown in Tables 1 and 2 [omitted] for interrater reliability and Tables 2 and 3 [omitted] for test-retest reliability. Examination of descriptive data and scatter plots revealed skewed score distributions for most variables. Therefore, Spearman rank-order correlation coefficients were used as indexes of reliability. To supplement these correlations, the percentages of time that the interrater and test-retest scores were within a certain magnitude of difference were calculated.

For interrater reliability, coefficients were high for one-leg standing balance in the eyes-open and eyes-closed conditions, both for right and left feet separately and right and left feet combined. Wilcoxon matchedpairs signed-rank tests were used to determine whether one rater consistently scored higher than the other. (31) No significant differences were found (p >.05). The magnitudes of difference between scores of the two raters were low. One hundred percent of the time, the raters' scores were within three seconds of each other.

Spearman coefficients for test-retest reliability for one-leg standing balance varied from a low of .59 for the eyes-closed-left-foot condition to a high of 1.00 for the eyes-open-right-foot condition. Because of the lower reliability for the eyes-closed-left-foot condition, scores for right and left feet were combined to lengthen the test. With the scores for left and right feet combined in both eyes-open and eyes-closed conditions, reliability coefficients were moderate to high. To determine whether test scores were consistently higher or lower than retest scores, Wilcoxon matched-pairs signed-rank tests were used. No significant differences were found (p >.05). One hundred percent of the time the magnitudes of difference between test and retest scores were three seconds or less for right and left feet in both eyes-open and eyes-closed conditions. However, for left and right feet combined in both eyes-open and eyes-closed conditions, although a small difference was observed between test and retest scores for most subjects, a five- or six-second difference was observed for a small percentage of subjects. The reliability of quality measurements made during the one-leg standing balance tests was computed using response-by-response percentages of agreement and Cohen's Kappa statistics. (32) The sample size for the four conditions varied because quality was evaluated only when a child's duration of standing balance exceeded three seconds. Results are presented in Table 4 [omitted] for interrater agreement and in Table 5 [omitted] for test-retest agreement.

Tiltboard Balance

Interrater and test-retest results for the tiltboard tests (degrees of tilt to right and left combined) are reported in Tables 6 to 8 [omitted]. Again, Spearman rank-order correlation coefficients were used as indexes of reliability! because examination of the data again revealed skewed score distributions. The scores from tilting to the right and left directions were analyzed separately and then were combined to increase the length of the test so as to improve reliahility. For both the tiltboard test eyes-open and eyes-closed conditions, interrater reliabilities were high, and in the majority of cases the two examiners' ratings were exactly the same. Again, when the Wilcoxon matched-pairs signed-rank tests were used to determine whether one rater consistently rated higher than the other, no significant differences were found (p >.05).

In contrast to the interrater analysis, the test-retest correlations and the magnitude of differences for the tiltboard balance test (degrees for right and left directions combined) were poor for both the eyes-open and the eyes-closed conditions. When the Wilcoxon matched-pairs signed-rank tests were used to check for differences between test and retest, no significant differences were found for the tiltboard test in the eyes-open condition (P >.05). However, in the eyes-closed condition, children were able to achieve an average of 29.6 degrees more tilt in the retest situation than in the test situation. This difference was significant (z = 2.70,p = .007, two-tailed).

Discussion

One-Leg Standing Balance This study suggests that two independent raters can time a one-leg balance test in normally developing children with a high degree of consistency. This high degree of interrater reliability was probably attributable to the clear delineation of instructions for positioning the child and determining when to start and stop timing. These results are comparable to the 1988 SIPT interrater reliability coefficients reported by Western Psychological Corporation (Diana Marr, PhD-1 personal communication; June 15, 1988).

Similarly, stability of scores (test-retest reliability) on this measure was very high (.91-1.00) for the eyes-open condition. However, for the eyes-closed condition, scores were lower (.59-.77). Even though less than ideal, these results are substantially higher than those reported by Ayres for standing balance tests, which are part of the SCSIT. (21) Also, in comparing our results with those of Ayres, it should be pointed out that the statistical test used in our study was nonparametric and not as powerful as the Pearson product-moment correlation coefficient used by Ayres. (31) The reliability results are similar to the coefficients given in personal communication regarding the new SIPT (Diana Marr, PhD; personal communication; June 15, 1988).

Even though our results showed higher reliability than the SCSIT, there was concern about the lower reliability for the eyes-closed-left-foot test condition. Therefore, reliability was examined by combining right- and left-foot scores for each condition as a means of lengthening the test, a procedure known to increase reliability if all other conditions are held constant. (26) AS shown in Table 3, this procedure resulted in a higher level of reliability for the eyes-closed condition (r[.sub.s] = .77). Therefore, for clinical use, it is recommended that right- and left-foot scores be combined.

It should also be noted that although children who are normally developing are able to balance on one leg with eyes open an average of 24 to 25 seconds, they as a group can balance on one leg with eyes closed an average of only 6 to 7 seconds. This clustering of the subjects' scores at the low end of the scale by normally developing children may indicate that the eyes-closed test will not be sensitive to differences between normally and abnormally developing children. Until further testing with children who have balance difficulties is completed, clinicians should be guarded about interpreting eyes-closed-one-leg balance testing results.

Results from the examination of interrater reliability of quality of one-leg standing balance were inconclusive. Although the response-by-response percentages of agreement between two raters for three of the measurements (right foot-eyes open, left foot-eyes open, right foot-eyes closed) were moderate to high, these results should be viewed with caution because a large majority of children were rated as using an ankle strategy. In the fourth condition, where the ratings were distributed more evenly between the two categories, the response-by-response percentage of agreement failed to reach acceptable levels. Kappa, a "chance corrected per cent agreement measure with a statistical base," (32(p310)) also was used to assess agreement between raters. In all cases, these values were low, indicating that more research is necessary prior to use of the quality rating. Definition of a hip strategy is fairly simple, but quick, subtle hip strategies can be observed by one examiner and easily missed by the other. Videotape analysis of the quality may be necessary for determinations of subtle changes.

Tiltboard Balance

For the tiltboard balance measure, interrater reliabilities (correlation coefficients and magnitudes of difference between scores) were very good for both the eyes-open and eyes-closed conditions. Again, this high reliability is likely attributable to the explicit test administration instructions and the simplicity of the measurement system. Also, because of the use of a secondary examiner to spot the child and to observe and report upper body reasons for terminating the trial, the primary examiner can focus directly on observing degrees of tilt. In addition, the use of the secondary examiner to stabilize the child, if necessary, contributes to the safety of the test. The disadvantage of using two examiners for one client in a clinical setting is obvious. However, the test is very short (approximately five minutes) and overall may be less time-consuming than photographic analysis. These results are comparable to those of Fisher and Bundy for analysis of tiltboard tip angles. (3)

Although the interrater reliability results for the tiltboard measure were high, the test-retest results for both eyes-open and eyes-closed conditions were low. For the eyes-open test, there was inconsistent movement of the scores up and down. What is of particular interest is the highly significant improvement (29.6 [degrees] ) between the mean test and retest scores for the eyes-closed condition. This improvement could he due to a practice or learning effect and the administration of too few trials to achieve a maximum score. We believe that the eyes-closed test is a novel task for most children. Therefore, children may initially he apprehensive about failing. Then, as the child builds trust in the examiners, performance improves. These effects may not have been controlled by giving just two trials. Further research incorporating more trials must be completed before therapists can have confidence in the stability of children's scores on this measure.

Another factor that may have affected these results is that speed of tilt was not monitored for consistency. The protocol used instructed the examiner to "tilt slowly" but did not provide for a standard speed of tilt. Weisz (1) and Keshner and Cohen (33) stated that the speed of tilt may influence the sensory and motor response to the tilt. Therefore, future test protocols should incorporate a standard speed of tilt. This potential inconsistency in speed of tilt may have affected both eyes-open and eyes-closed testing results. It could be one reason for the variability of the eyes-open test-retest results, but we believe that the large difference in the eyes-closed condition cannot be fully explained just by this factor.

This reliability study, completed during development of a batters, of tests for examining postural stability in children, involved the testing of normally developing children. This testing was necessary to define normal balance behaviors so that in the future comparisons can be made with performances by children with balance dysfunction. Determinations of test reliability are important and necessary in order to understand the confidence therapists can have in the test results for initial assessment and for monitoring change with treatment programs. The next step is to examine the reliability of these procedures when used on children with potential balance difficulties. Reliability also should be examined using therapists who are not involved in the test development but who instead have learned to administer the test by reviewing the manual.

Conclusions

Within the development of more comprehensive and reliable testing procedures for evaluation of postural stability, 24 normally developing children were given two tests of balance: a one-leg standing balance test (eyes open and eyes closed) and a tiltboard balance test (eyes open and eyes closed). interrater and test-retest reliability were calculated and found to be moderate to high for the timing of one-leg standing but low for the quality assessment. Interrater reliability, was high for the tiltboard test, but test-retest reliability was low. As a result of these findings, clinicians should recognize that they can have little confidence in the stability of quality scores and tiltboard degrees of tilt scores. Further study is needed to better define quality measures and to determine whether controlling for speed of tilt and increasing the number of trials will improve the tiltboard test-retest reliability. Research also should be directed at determining test reliabilities for populations of children with balance difficulties and for examiners who only read the manual prior to testing.

Footnotes

* 1ft = 0.3048 m.

+ 1 in = 2.54 cm.

References

1 Weisz S: Studies in equilibrium reaction. Nerv Ment Dis 88:150-162, 1938

2 Ayres AJ: Sensory Integration and Learning Disorders. Los Angeles, CA, Western Psychological Services, 1972

3 Fisher AG, Bundy AC: Equilibrium reactions in normal children and in boys with sensory integrative dysfunction. The Occupational Therapy Journal of Research 2(3):171-183, 1982

4 Polatajko HJ, Sullivan T: Postural-sway responses in learning-disabled children: Pilot data. The occupational Therapy Journal of Research 7(l):37-51, 1987

5 Bundy AC, Fisher AG, Freeman M, et al: Concurrent validity of equilibrium tests in boys with learning disabilities with and without vestibular dysfunction. Am J Occup Ther 41:28-34, 1987

6 Effgen SK: Effect of an exercise program on the static balance of deaf children. Phys Ther 61:873-877, 1981

7 Horak FB, Shumway-Cook A, Crowe TY, et al: Vestibular function and motor proficiency of children with impaired hearing or with learning disability and motor impairments. Dev Med Child Neurol 30:64-79, 1988

8 Bobath K, Bobath B: The facilitation of normal postural reactions and movements in the treatment of cerebral palsy. Physiotherapy 50:246-262, 1964

9 Shambes GM: Static postural control in children. Am J Phys Med 55:221-252, 1976

10 Nashner LM, Shumway-Cook A, Marin O: Stance posture control in select groups of children with cerebral palsy: Deficit in sensory organization and muscular coordination. Exp Brain Res 49:393-409, 1983

11 Shumway-Cook A, Woollacott MH: Dynamics of postural control in the child with Down syndrome. Phys Ther 65:1315-1322, 1985

12 Connolly BH, Michael BT: Performance of retarded children, with and without Down syndrome, on the Bruininks-Oseretsky Test of Motor Proficiency. Phys Ther 66:344-348, 1986

13 Rider RA, Mahler TJ, Ishee J: Comparison of static balance in trainable mentally handicapped and nonhandicapped children. Percept Mot Skills 56:311-314, 1983

14 Bobath B: Postural Reflex Activity Caused by Brain Lesions. London, England, William Heinemann Medical Books Ltd, 1971

15 Reed KL: Models of Practice in Occupational Therapy. Baltimore, MD, Williams & Wilkins, 1984, pp 309-329

16 Clark F, Mailloux Z, Parham D: Sensory integration in children with learning disabilities. Clark PN, Alles AS (eds): Occupational Therapy for Children. St Louis, MO, C V Mosby Co, 1985, pp 359-405

17 Sullivan PE, Markes PD, Minor MD: An integrated Approach to Therapeutic Exercise, Theory and Clinical Application. Reston, VA, Reston Publishing Co Inc, 1982

18 Folio MR, Fewell R: Peabody Development Motor Scales Manual. Allen, TX, DLM Teaching Resource, 1983

19 Bruininks RH: Bruininks-Oseretsky Test of Motor Proficiency: Examiner's Manual. Circle Pines, MN, American Guidance Service, 1978

20 Stott DH, Moyes SA, Henderson SE: Test of Motor Impairment Manual. Guelph, Canada, Brook Editors and Publishers Ltd, 1984

21 Ayres AJ: Southern California Sensory Integration Test Manual. Los Angeles, CA, Western Psychological Services, 1980

22 Ayres AJ: Sensory Integration and Praxis Test Manual. Los Angeles, CA, Western Psychological Services, 1988

23 Schulmann DL, Godfrey B, Fisher AG: Effect of eye movements on dynamic equilibrium. Phys Ther 67:1054-1057, 1987

24 Clark JE, Watkins DL: Static balance in young children. Child Dev 55:854-857,1984

25 Krebs DE: Measurement theory. Phys Ther 67:1834-1839, 1987

26 Anastasi A: Psychological Testing, ed 6. New York, NY, Macmillan Publishing Co, 1988

27 Izraelevitz TA, Fisher AG, Bundy AC: Equilibrium reactions in preschoolers. The Occupational Therapy Journal of Research 5(3):154-169,1985

28 Perham H, Smick JE, Hallum A, et al: Development of the lateral equilibrium reaction in stance. Dev Med Child Neurol 29:758-765, 1987

29 Shumway-Cook A, Horak FB: Assessing the influence of sensory interaction on balance: Suggestion from the field. Phys Ther 66:1548-1550, 1986

30 Horak FB: Clinical measurement of postural control in adults. Phys Ther 67:1881-1885, 1987

31 Siegel S: Nonparametric Statistics for the Behavioral Sciences. New York, NY, McGraw-Hill Book Co, 1956

32 Bartko JJ, Carpenter WT: on the methods and theory of reliability. J Nerv Ment Dis 163:307-317, 1976

33 Keshner EA, Cohen H: Current concepts of the vestibular system reviewed: 1. The role of the vestibulospinal system in postural control. Am J Occup Ther 43:320-330, 1989