The effect of training on physical therapists' ability to apply specified forces of palpation.

Key Words: Force platform testing, Lumbar intervertebral joint, Palpation skills, Scale training.

The aim of this study was to evaluate whether physical therapists who were students in a 12-month postgraduate manipulative therapy diploma course could learn to apply specific forces with accuracy during palpation of an intervertebral joint. Learning to palpate intervertebral joint in order to detect the way in which it moves is an important demand placed on physical therapists, because palpation techniques are used extensively in the assessment and treatment of the spine. Currently, this skill is usually taught using methods that rely on the ability of instructors to demonstrate or describe their application of the skill to the student and the ability of the student to appreciate the intention of the instructors. Achievement of the skill has yet to be measured by an external standard common to both teacher and student, allowing substantial interpretation by the student. in this way, palpation training seems to more closely resemble an art than it does a science.

Intervertebral joint palpation is intended to give the clinician valuable information about joint behavior. Early palpation training involves learning to apply a force to a joint and to evaluate the joint's response to that force. The student is supposed to appreciate the amount of force applied to the joint, the speed and direction at which it is applied, the amount of movement produced at that joint, the way in which the joint moves or resists movement in response to that force, the pain produced by that movement, the presence of muscle activity evoked during the movement, and the comparison of this reaction to the expected normal response and to that of adjacent joints when palpated in a presumably similar way.[1] It is clearly a complex task to perform and teach.

We chose to study the ability of physical therapists to produce specified forces applied during palpation for two reasons. One reason was to determine the error made in the reproduction of the forces applied. Error in force reproduction may cause the therapist to believe that the same force of palpation was being repeated when in fact a stronger or weaker force occurred. The other reason was to investigate whether therapists could learn to produce forces that are specified numerically. Numerical scaling of force applied during the performance of a palpation technique could create a common standard between students and teachers through which perception of at least one component of the skill could be shared accurately.

Based on a review of the literature, it appeared possible to us that, with appropriate training, physical therapists should be able to learn to palpate using forces of specified magnitude.[2] Optimal training methods for the learning of a perceptual motor skill, such as recognition and reproduction of applied force, are well documented.[3] The most important of these methods for the improvement of response is accurate feedback.[4] Knowledge of performance affects both the rate of learning and the level reached by learning.[4,5] Demonstration of the required standard assists in learning,[6] but accuracy and retention of the learned response is, at least in part, a function of precision of knowledge of results.[7] Annett[2] trained students to apply a precise pressure by providing them with knowledge of results in the form of verbal or visual scale readings on completion of the response. If Annett's methods were adapted for the training of force applied during palpation techniques, it seems probable that physical therapists should be able to learn to apply forces of specific magnitudes if they are trained with accurate knowledge of results of their attempts.

We believe that methods currently used to teach passive accessory intervertebral movement (PAIVM) to therapists studying manipulation are unlikely to result in these therapists being able to reliably estimate the forces they apply during palpation. These methods include instructor demonstration, students practicing the technique on each other, performance of the technique on students by an instructor to convey the correct "feel" of the technique, and comparison of perceptions through comparison of movement diagrams. A movement diagram is a two-dimensional sketch that the therapist draws to symbolize perception of joint response to palpation. Resistance to movement is plotted against position in the joint range of movement.[1] Two training methods--students practicing with each other and comparison of movement diagrams--provide feedback following the performance of a palpation technique and can be used to improve students' perception of the amount of force applied during palpation. Of these two methods, students practicing with each other appears to be the most commonly used method of training. To conform to optimal requirements for the training of a perceptual skill, however, feedback about performance must be accurate. The methods used in the training of perception of joint response to palpation (eg, students practicing with each other and comparison of movement diagrams) do not appear likely to provide accurate knowledge of applied force to the students.

Although students practicing with each other and comparison of movement diagrams both provide feedback, they do not provide accurate identification of forces applied during palpation. It would appear predictable that intertherapist reliability of the amount of force applied during the performance of techniques would be poor. Preliminary research into the acquisition of palpation skills has primarily examined force applied during the performance of a specific palpatory technique using a force platform[8-11] or its estimation using movement diagrams.[12,13] Poor reliability has been found in both intratherapist and intertherapist data.[14]

Recently, Lee et al[3] successfully trained physical therapy students to apply a specific force during palpation. A force platform was used for force measurement, and feedback was given via an oscilloscope that displayed both the applied force and the desired force. They found that, compared with a control group who had no oscilloscope feedback, learning did occur and was maintained at a 1-week follow-up test. The task, however, was very simple as the students had only to memorize a single force application. Physical therapists are required to palpate intervertebral joints over a wide range of applied forces. It is therefore important that accuracy of performance be obtained over the range of forces used. In this study, we trained students in our experimental group to apply a wide range of forces. In designing the training program, we chose to investigate those aspects of feedback that appear to be most significant for optimal learning of a perceptual motor skill.[2,4-7] The feedback during training was delayed, available to the student after attempting the task. Furthermore, prior to knowledge of results of each attempt, the students were asked to estimate their own accuracy. The decision to train students in this way allowed them accurate knowledge of results linked directly to the task and the opportunity to internalize the requirements for accuracy through repeated error estimation.

Earlier experiments utilized an oscilloscope linked to a force platform to provide students with feedback about palpation forces.[3] This equipment is expensive, the training procedures are time consuming for both students and staff, and the students are not free to practice in their own time. We sought easily available, inexpensive, portable equipment with which the students could learn to quantify forces applied through their hands. Calibrated bathroom scales were chosen for this purpose. We designed a program to train the experimental group to produce a broad spectrum of numerically standardized applied forces using the readout from these scales. We then assessed their ability to produce these specific forces on request.

In summary, the aim of this research was to establish whether therapists could be trained to reproduce specific forces that they applied to bathroom scales and whether this training resulted in improved ability to reproduce forces applied during the performance of posteroanterior intervertebral palpation of the lumbar spine.

Method

Subjects

Students who participated in this study were 12 physical therapists (7 female, 5 male), ranging in age from 26 to 36 years ([bar] X=29.5, SD=2.9), undertaking the 1987 Postgraduate Manipulative Therapy Diploma Course at Lincoln Institute (La Trobe University, Carlton Campus, Victoria, Australia). All therapists had between 3 and 10 years ([bar] X=6.3, SD=1.9) of clinical experience prior to beginning the course. Although 14 students were enrolled in the course, only volunteers were used in this study. During testing, therapists performed palpation techniques on each other. All therapists were fully informed of the experimental procedure and were aware that they were free to withdraw at any time.

All subjects were tested on three occasions: prior to training (pretest), on completion of the 30-day training period (posttest), and 1 month following cessation of training (retention test). At each of these three phases, subjects were tested at six levels of force application. These forces were 1, 5, 10, 15, 20, and 25 kiloponds (kp). These force magnitudes were selected as representative of the range of commonly applied forces during the performance of this posteroanterior PAIVM technique. The selection was made on the basis of previous experiments in which a force platform was used to measure forces applied during lumbar palpation.[9,11]

Instrumentation

For testing, therapists stood on a Kistler force platform.* This platform measured the force acting on it resulting from the weight of the therapist. Beside this platform, a subject (one of the other therapists) lay prone on an adjustable plinth. When the therapist applied a force to the lumbar spine of the subject, the force platform measured the new weight of the therapist, which was the initial weight minus the force applied to the lumbar spine (Fig. 1).

The force platform was connected to a computer[dagger] that was programmed to record these force differences (ie, the force applied to the subject). The error of estimating peak forces applied using this technique has been reported to be 1% to 3%.[14]

Training

A set of tested and calibrated bathroom scales of the same brand and type[double dagger] was given to each subject in the experimental group (n=6). These scales were used to provide feedback on performance of the applied forces. Training of subjects in the experimental group commenced following the pretest. They were given written practice instructions and a diary sheet. The written instructions were designed to standardize the practice technique.

The subjects were instructed to practice 5 to 10 minutes daily for 30 consecutive days. Scales were to be placed on a firm surface at the appropriate height for delivery of a lumbar posteroanterior PAIVM. The first force (1 kp) was attempted with eyes closed, then the therapist opened his or her eyes to see what force was actually applied. The actual force applied as registered on the scales was noted in the diary sheet. The same process was repeated for the 5-kp force, then for the 10-kp force, and so on until all force applications were attempted. This sequence was to be repeated a minimum of three times at each practice session. Entries made in the diary sheets indicated that no therapist missed more than 2 days of practice over the 30-day training period. The control group (n=6) had no practice using scales. Subjects in both groups received concurrent training in palpation techniques as part of their course work during the period over which the experiment was conducted.

Procedure

In order to determine accuracy in the reproduction of forces prior to training, all subjects were pretested using the force platform. The subjects were first asked to adjust the plinth to an appropriate height for the performance of a posteroanterior PAIVM to the lumbar spine of an asymptomatic volunteer (one of the other therapists).

Subjects then selected one lumbar vertebra and stood on the force platform in preparation for palpating the vertebra. A 30-second training period was provided, during which the computer sounded a bell when the applied force was 5 kp.

The subjects were then asked to apply a force of 1 kp. The subjects signaled when they perceived that they were applying the requested force, and the actual force being applied was recorded by the computer. This test for the force of 1 kp was repeated for the forces of 5, 10, 15, 20, and 25 kp in random sequence. These six forces were randomly tested a total of three times. At the posttest, all subjects were retested on the force platform in the same way as at the pretest, except that the 30-second familiarization with the 5-kp force was omitted. The retention test was also performed in the same way as the posttest. In addition, at the retention test, all subjects were also tested on the force platform while applying forces to the training scales rather than a lumbar spine.

In order to allocate subjects to either the control group or the experimental group, the data collected at the pretest were analyzed to determine two kinds of error: percentage systematic error (PSE) and percentage random error (PRE) (see "Data Analysis"). The PSE represents the difference between the requested force and the applied force. A low PSE score would indicate that the subject was able to apply forces that approximated the requested force. The PRE represents the difference between repeated attempts to produce the same force and is independent of the requested force. A low PRE score would indicate that the subject performed consistently but gives no indication of accuracy in the test task. We were interested in the effects of training on both of these sources of error and hence analyzed them separately. On the basis of error scores produced at the pretest, subjects were divided into two groups matched as evenly as possible for PSE and PRE scores. This was done to minimize the potential problems in random allocation of subjects with such a small sample.

Data Analysis

At each test phase, subjects attempted to apply the six forces in random sequence three times. For each force, the mean of these three attempts and its absolute difference from the requested force were calculated. This difference was expressed as a percentage of the requested force and represented the PSE score for that force. The PSE scores over the six forces were averaged to yield the mean PSE score for that subject. To obtain the PRE score for each force, the absolute differences between each attempt and the mean of the three attempts were averaged and expressed as a percentage of the requested force. The PRE scores over the six forces were also averaged to yield the mean PRE score for that subject.

Performance of subjects was assessed by calculating the mean PSE and mean PRE scores for each force at the pretest, posttest, and retention test. Two three-way analyses of variance (ANOVAs) for repeated measures on the last two factors (test phase and force) were performed on the mean PSE scores. The first analysis examined the effect of training. The factors in this analysis were group (control, experimental), test phase (pretest, posttest), and force (1, 5, 10, 15, 20, and 25 kp). The second analysis examined retention of the training effects. The factors were group (control, experimental), test phase (pretest, retention test), and force (1, 5, 10, 15, 20, and 25 kp). Two additional ANOVAs, using the same designs, were conducted to analyze the PRE scores.

As the force factor contained six repeated measures, the problem of violation of the assumption of homogeneity of covariance had to be considered. This problem was overcome by computing F values using the multivariate approach for all repeated measure effects with more than 1 degree of freedom. This multivariate analysis of variance is insensitive to violation of assumption of homogeneity of covariance.[15]

Results

Force Platform Data

Figure 2A presents the average PSE scores for the control and experimental groups at the pretest, posttest, and retention test. The PSE, averaged over the six forces, decreased between the pretest and the posttest for the experimental group, whereas it increased for the control group. These trends, however, were not significant. The graph further shows that this downward trend in PSE scores for the experimental group continued between the posttest and the retention test, whereas only a very slight decrease in PSE scores was suggested for the control group between the same tests. The second ANOVA confirmed that the difference in the trends exhibited by the two groups from the pretest to the retention test was significant F[1/10]=5.05, P<.05). This finding indicates that, when tested during palpation of the lumbar spine, the ability of the trained subjects to produce the requested force improved significantly more than that of the untrained subjects in the period between the pretest and the retention test.

Although the PSE scores shown in Figure 3 suggest that the improvement in the trained subjects did not occur uniformly at all forces tested (eg, the differences between groups appear to be more pronounced in the lower forces, with virtually no apparent differences at 20 and 25 kp), neither of the ANOVAs revealed this difference to be statistically significant.

Figure 2B shows the average PRE scores for the control and experimental groups for the pretest, posttest, and retention test. Little change in PRE scores appears to have occurred in either group at the different test phases, and the ANOVAs also failed to detect significant variations.

In parallel with the PSE findings, Figure 4 suggests that higher PRE values tended to occur at low forces. The analyses confirmed that for both groups significant differences in PRE score magnitude occurred as a function of the task force required F[5/5]=28.22, P<.001 and F[5/5]= 33.12, P<.001).

Performance of Force Application Using the Scales

At the time of the retention test, all subjects were further assessed in force production using the training scales instead of the lumbar spine. Errors in performance of force application on the scales were again converted to a percentage of the requested force. A two-way ANOVA for repeated measures on the second factor was performed on the data. The factors were group (control, experimental) and force (1, 5, 10, 15, 20, and 25 kp).

Figure 5A shows the PSE scores for the control and experimental groups across the six forces. The PSE values for both groups tended to diminish as the forces increased, and the ANOVA confirmed a significant force effect F[5/6]=9.16, P<.01). More importantly, the experimental group

achieved very low average PSE scores for most forces compared with the control group. The ANOVA confirmed that, on average across the six levels of force application, the experimental group had significantly lower PSE values than did the control group F[1/10] = 15.3, P<.005). In addition, the discrepancy between the two groups was more pronounced at the lower forces, and the analysis revealed significant variation across forces in the difference between the PSE scores of the two groups (F[5/6]= 6.68, P<.02).

Figure 5B shows the PRE scores for the control and experimental groups across the six forces. As with the systematic error, random error tended to diminish as a proportion of the requested force at the higher forces. The variation in PRE values across the levels of force application was significant F[5/6]=18.65, P<.001). The ANOVA revealed that the difference between groups, averaged across the six forces, was significant F[1/10] = 23.83, P<.001).

The diary sheets reflected the accuracy observed during the testing of the experimental subjects using the scales. Diaries indicated that the low errors were achieved for all subjects within 3 to 5 days of training and were maintained throughout the practice period.

Discussion

The graphs of PSE scores for the individual forces (Fig. 3) show that both groups had PSE scores of different magnitudes at the different forces. The greatest PSE scores tended to be associated with the low forces, decreasing as the forces increased. This effect was probably due to the method used to determine PSE and PRE scores. The PSE and the PRE express error as a percentage of the requested force. The graphs in Figure 3 reflect the fact that a small error in judgment of, for example, 1 kp would represent a 100% error score for the 1-kp test, but only a 10% error for the 10-kp test. The ANOVA for the pretest-posttest data indicated that this variation in PSE scores at the different forces closely approached accepted standards of statistical significance F[5/5] = 5.00, P<.051). Post hoc tests, however, were not conducted to substantiate the exact nature of these interactions because these tests are

more conservative than the ANOVA, within which the effects were only marginally significant. Furthermore, these apparent interactions in the training effect obtained at the different forces were not central to the argument, particularly given that at no force did the control group appear to improve more than the experimental group.

Although training improved the ability of subjects to produce the requested force (indicated by the significant change in PSE between the pretest and the retention test), the lack of change in the PRE scores at the different test phases indicates that subjects made errors of similar magnitude on the different test occasions when attempting to apply the same force several times. The PRE scores give no indication of the accuracy in production of the requested force but reveal the variability in response when subjects attempt to repeat what they believe to be the requested force (ie, the variations around their mean performance for a specified force).

Examination of Figure 4, however, suggests that at the 1-kp force, the trained subjects improved more than the control group when the posttest and retention test are compared with the pretest, although the ANOVAs failed to detect this effect under the multivariate F tests. It may be of interest to note that, provided homogeneity assumptions are reasonable, univariate F tests would confirm the apparent trend for both the pretest-posttest comparison (F[5/45] = 2.47, P<.05) and the pretest-retention test comparison (F[5/50] = 2.87, P<.05), suggesting that the training program may have had beneficial effects on random error at the lighter forces. Training appears to have significantly improved performance of force application when PSE scores for all forces were averaged, and this improvement was greatest in the performance of the lighter forces.

Again, the experimental group performed better at the lower forces than did the control group, particularly at 1, 5, and 10 kp. This effect was confirmed by the ANOVA, with a significant group x force interaction (F[5/61 =10.94, P<.01). Therefore, although the random error associated with palpation of the lumbar spine did not improve with training, significant training effects became apparent when the control and experimental groups were tested in the trained task (ie, palpation of the scales).

When tested on the scales, it was apparent that subjects in the experimental group had fulfilled the experimental predictions. They had learned to reproduce the forces with accuracy and demonstrated significantly lower PSE and PRE scores than did the control group. These findings indicate that, as a consequence of training, the experimental group subjects were not only more accurate than the control group subjects in the production of a requested force, but they made less random error in the repeated performance of that task. Inspection of the records kept in the diary sheets suggested that the experimental group subjects had achieved these skills quickly (ie, within 3-5 days of training) and had retained these skills 1 month following the cessation of training. These results support the experimental hypothesis that accurate feedback following the performance of a task results in learning[2,4,16] and that students are capable of learning and accurately reproducing a wide range of palpation forces.

The results further showed that, compared with the control method, this method of training therapists in the production of specific forces resulted in a greater improvement in accuracy in the production of those forces during the application of PAIVM to the lumbar spine. Although this improvement between the pretest and the posttest could not be unequivocally supported by the ANOVA, there was clear statistical confirmation of improvement in the PSE between the pretest and the retention test. The PRE scores for both groups during testing on the lumbar spine did not change significantly over the three test sessions. This finding suggests that the range of error in estimating a particular force applied to the spine was not affected by training. The ability of the trained students to approximate the requested force (PSE), however, was significantly improved with training. The trained subjects had developed an appreciation of the external scale associated with their applied forces.

The improvement in lumbar palpation performance of the trained subjects at the retention test may be due to any of several factors. The posttest was conducted at the end of a week of theory examinations, and the students were fatigued. At this test, most subjects in the experimental group reported that they were not confident of their performance of the force applications, as the feel of the lumbar spine was very different from that of the scales. One explanation for this lack of confidence during palpation of the lumbar spine might be that the subjects had been trained using scales that have a firm surface but were tested during the palpation of the lumbar spine, which is pliable. In addition, several therapists commented that the movement of the lumbar spine associated with the breathing of the subject being palpated made it difficult for them to be sure of the force they were applying, as they had to concentrate on following the movement of the lumbar spine as well as the force that they applied. During training using the scales, the therapists did not have this distraction. The improved performance of subjects in the experimental group at the retention test may have been due, in part, to learning of the "new" task (ie, force estimation during palpation of the lumbar spine) that took place at the posttest. Some ongoing learning may also have taken place in the absence of scales practice, because testing using scales at the time of the retention test showed high accuracy in performance of the force applications. Hence, with the acquired, internalized scale of applied forces, palpation practice in the period between the posttest and the retention test could also account for the continued improvement.

The results of our experiment indicate that therapists can be trained to produce numerically specified forces during lumbar posteroanterior PAIVM. In addition, they appear capable of learning to produce, upon request, forces selected from a wide spectrum of forces. Despite the apparent complexity of familiarization with this broad spectrum of forces (1-25 kp), the trained task appears to have been learned quickly (within 2-3 days). There was excellent performance of the trained task and improved performance of the test task at 1 month following the cessation of training. These findings indicate that the ability to apply specified forces was well maintained. Moreover, the training program achieved statistically significant improvements despite the small number of subjects.

Lee et al[3] trained subjects in force production using simultaneous oscilloscope feedback. Our subjects and those trained by Lee et al performed significantly better than the control group at the retention test. We gained additional insight into this skill retention, as our retention test took place 1 month after the cessation of training as opposed to the subjects in the experiment by Lee et al, who were retested 1 week following cessation of training.

More importantly, our study also demonstrates that this skill is achievable not only as imitation of a set force, but as identification of applied forces selected from a wide spectrum of possible forces. Furthermore, our trained subjects learned to label their applied force using a standardized numerical tag. Therefore, communication among therapists about the forces that they are applying appears to be possible.

The forces that therapists in this experiment were trained to produce should not be confused with "grades of movement" as defined by Maitland.[1] Grades of movement are established in relation to the therapist's perception of the point in the range of movement at which the joint resists a palpation technique. The force required to produce a specific grade of movement, therefore, will vary with the joint being palpated and the palpation technique used. Force can be applied to a joint without appreciation of the effect this may be having on the structures under palpation and is not a substitute for the evaluation of joint behavior required when grades of movement are used to document a treatment technique. The value of knowing how much force is used to produce a grade of movement (for a particular joint at a point in time) lies in the option to communicate this to other therapists who are teaching or learning each others' perception of grades of movement.

As discussed earlier, in the training of palpation skills, instructors must communicate their perception of joint responses to palpation to their students. No external criteria exist by which to teach or gauge the acquisition of this skill. By quantifying forces applied to a joint, at least one aspect of this skill can be communicated. The ability to quantify applied forces offers teachers and students a common external standard against which other perceptual events can be matched and communicated. This scaling of applied forces offers an opportunity for therapists to share what is currently an independent interpretation of a perceptual skill.

In clinical practice, physical therapists who use manipulative therapy techniques routinely require the ability to accurately repeat a palpation technique. Accuracy in the reproduction of technique may be enhanced if the random error associated with the therapists' estimation of applied force could be minimized. Although the PRE calculated for the performance of the lumbar palpation did not improve with our training, the random error associated with the performance of the trained task (ie, the application of force to the scales) was significantly less for the trained group when compared with the control group. Given the cost effectiveness of this training method, it is highly desirable for additional studies to further explore improvements in transfer of the trained task to the test task. As transfer of training between tasks improves with similarity between tasks,[4,16] the most practical approach would be to explore modification of the bathroom scales such that the surface of the scales more closely approximates the feel of the lumbar spine. Alternatively, scales could be placed on a pliable surface for practice, rather than on a rigid surface as in this study. Another possibility would be to pursue training as outlined in this experiment but to intersperse the training period with brief training sessions palpating the lumbar spine with feedback from the force platform.

A small number of specialized therapists involved in training of manipulative therapy skills took part in this training process; thus, generalization regarding the success or applicability of these methods must be made with caution. Nevertheless, the results of this pilot study are encouraging and represent a method by which some quantification of a difficult perceptual skill may begin to evolve.

Conclusions

Palpation skills are complex and difficult to teach and to learn. This research dealt with only one aspect of the acquisition of these skills--the amount of force applied by the therapist. This method of training production of forces resulted in high levels of accuracy in the trained task and improved accuracy in the test task. Our results therefore indicate that therapists may be able to be trained to recognize and communicate the amount of force used during palpation. Further study is required to verify that alteration to the method of training will result in improved performance of the test task.

References

[1] Maitland GD. Vertebral Manipulation. 5th ed. London, England: Butterworth & Co (Publishers) Ltd; 1986:71-92, 144-170, 352-368. [2] Annett J. Learning a pressure under conditions of immediate and delayed knowledge of results. Q J Exp Psychol. 1959; 11:3-15. [3] Lee M, Moseley A, Refshauge K. Effect of feedback on learning a vertebral joint mobilization skill. Phys Ther. 1990;70:97-104. [4] Schmidt RA. Motor Control and Learning: A Behavioral Emphasis. 2nd ed. Champaign, ill: Human Kinetics Publishers Inc; 1988:337-456. [5] Ammons RB. Effects of knowledge of performance: a survey and tentative theoretical formulation. J Gen Psychol. 1956;54:279-299. [6] Lincoln RS. Learning and retaining a rate of improvement with the aid of kinesthetic and verbal cues. J Exp Psychol. 1956;51:199-204. [7] Bennet DM, Simmons RW. Effects of precision of knowledge of results on acquisition of a simple motor skill. Percept Mot Skills. 1984; 58:785-786. [8] Banting J. Intertherapist Reliability in Performing Grade H Mobilization Movement. Melbourne, Victoria, Australia: La Trobe University; 1982. Postgraduate diploma dissertation. [9] Grisold PM. Estimating Range of Movement from Passive Intervertebral Movements: The Nature of the kale and the Reliability of the Performance. Melbourne, Victoria, Australia: La Trobe University; 1983. Postgraduate diploma dissertation. [10] Mitchell WN. Reliability in the Performance of Grade II or IV Mobilization Movements. Melbourne, Victoria, Australia: La Trobe University; 1983. Postgraduate diploma dissertation. [11] Telin P. Estimating Range of Movement from Passive Intervertebral Movement Force Versus Displacement Cues. Melbourne, Victoria, Australia: La Trobe University; 1985. Postgraduate diploma dissertation. [12] Weeks P. Test-Retest Reliability of Stiffness Using PAIVM. Melbourne, Victoria, Australia: La Trobe University; 1982. Postgraduate diploma dissertation. [13] Wong M. Interobserver Reliability of Stiffness Onset Ratings Obtained from Passive Accessory Intervertebral Movements Melbourne, Victoria, Australia: La Trobe University; 1981. Postgraduate diploma dissertation. [14] Matyas TA, Bach TM. The reliability of selected techniques in clinical arthrometrics. Australian Journal of Physiotherapy. 1985;31: 175-199. [15] Harris RJ. A Primer of Multivariate Statistics. 2nd ed. London, England: Academic Press Inc (London) Ltd; 1985. [16] Sage GH. Introduction to Motor Behavior: A Neuropsychological Approach. Reading, Mass: Addison-Wesley Publishing Co Inc; 1977: 224-254. (*) Model 9281B with a Kistler 9807 amplification system, Kistler Instrument Corp, CH-8408 Winterthu Switzerland. [dagger] PDP 11/73 computer running under DAOS data-acquisition software, Digital Equipment Corp, 14 Main St, Maynard, MA 01754. [double dagger] Salter bathroom scales, Salter Weightronix Pty Ltd, 1 Apollo Ct, Blackburn, Victoria, Australia 3130.

J Keating, PT, is currently pursuing a master's degree by research thesis in the Department of Behavioural Health Sciences, Lincoln School of Health Sciences, La Trobe University, Bundoora, Victoria, Australia 3083. Address all correspondence to Ms Keating at 13 Warner St, Essendon, Victoria, Australia 3040.

TA Matyas, PhD, is Senior Lecturer, Department of Behavioural Health Sciences, and Convenor, Movement Rehabilitation Research Group, Lincoln School of Health Sciences, La Trobe University.

TM Bach is Senior Lecturer in Biomechanics, Department of Human Biosciences, Lincoln School of Health Sciences, La Trobe University.