Responsiveness of the movement ability measure: a self-report instrument proposed for assessing the effectiveness of physical therapy intervention.

When combined with performance-based data, self-report measures can provide evidence of the effectiveness of physical therapy intervention. Evidence of effectiveness frequently involves the recording of changes over time, between the beginning and the end of an episode of care, for example. Measures used to provide effectiveness evidence over time, therefore, must detect or respond to change. (1) Evaluating an instrument such as the Movement Ability Measure (MAM) for use in assessing the effectiveness of physical therapy intervention thus requires not only evidence of reliability and validity (see the article by Allen on the validity and reliability of data for the Movement Ability Measure in this Special Series) but also evidence of responsiveness. This article presents evidence of the ability of the MAM to respond to clinically relevant changes over a course of physical therapy intervention.

The MAM is a self-report measure created with a focus on movement, a concept common to broad diagnostic groups and specific to the work of physical therapists. The MAM has its foundation in the Movement Continuum Theory (MCT) of physical therapy (2) and in a multidimensional extension of the MCT (see the article by Allen on 6 dimensions within the construct of movement in the Movement Continuum Theory in this Special Series). The MCT posits that people have a maximum achievable movement potential and a current and preferred movement capability. The multidimensional extension of the MCT specifies the idea of movement further by subdividing the construct of movement into 6 dimensions: flexibility, strength, accuracy, speed, adaptability, and endurance. The MAM puts the MCT and the 6 dimensions of movement into operation by asking people to record how they think they move now and how they would like to be able to move in the 6 dimensions. If people's perceptions of movement ability change with an episode of care, then their responses on the MAM should change.

Assessing how well an instrument detects change involves different criteria, depending on the purpose of the instrument. (3) Instruments designed for longitudinal testing in effectiveness research should reflect significant and meaningful changes between measurements obtained before intervention and measurements obtained after intervention. Evaluating such instruments involves examination of sensitivity to change and responsiveness; the definitions of these terms overlap. (3) Sensitivity to change means that the instrument can accurately detect change when it occurs, a concept predicated on the reliability of the instrument. That is, different responses over different occasions should reflect true differences in ability rather than measurement error. Measures of sensitivity to change include effect size, standardized response mean (SRM), and paired t values for single-group comparisons. (3) Responsiveness incorporates the concept of sensitivity to change but also involves comparison of changes in a group with changes in a source external to the group, such as changes in a no-intervention control group. Measures of responsiveness include the responsiveness index (4) and independent t values. (3)

Even with evidence of sensitivity to change and responsiveness, the changes recorded may mean very little to the physical therapist or to the patient or client. (5) The meaningfulness or clinical value of reported differences following intervention depends on additional criteria, which are sometimes reported as the minimal clinically important difference (MCID) (4) or the minimal important difference. (6) Calculating the MCID helps to quantify the size of the difference necessary to have meaning for the clinician or patient. For the clinician, meaningful differences may signify a turning point in the course of the intervention; for the patient, meaningful differences may signify a new freedom to resume some previous activities. (5) In a study in which the Berg Balance Scale was used to record balance changes with intervention in participants with multiple sclerosis, a 1-point difference out of a total of 56 points reflected insufficient change for the participants to find it meaningful; a 6-point change was designated the minimum needed to have clinical importance for the participants. (7)

Determination of the MCID relies on either normative or criterion-based methods. (6,8) In a normative method, the MCID is based on the distribution of scores within the group tested. This method equates the MCID with the standard error of measurement (SEM), which is mathematically related to the standard deviation of the scores at baseline and the reliability of the scores obtained with the instrument. In a criterion-based method, the MCID is anchored on some external source. Possible external sources include therapist experience with a particular measure or consensus with other therapists, achievement of intervention goals or therapist-recognized improvement, or patient statement or response on a concurrent measure, such as a global rating of difference scale. (8) Once calculated, the MCID may serve as a standard against which to compare changes in an individual or as a cutoff point for proportions of people in an intervention group who achieve the MCID and the proportions of people who do not. (5)

For this study, multiple criteria served to help interpret and place into context (1) changes noted in MAM responses following intervention. Evaluation of sensitivity to change incorporated calculation of effect size, calculation of SRM, and single-group t test comparisons. Evaluation of responsiveness incorporated calculation of the responsiveness index and other comparisons with a no-intervention group and with criteria obtained from the attending physical therapists regarding the success of therapy. Attending physical therapists also completed the MAM for each participant in the intervention group after the initial examination to provide an additional external source against which to compare the participants' self-reported movement ability. Interpreting the meaning of different responses following intervention involved calculation of the MCID and comparison of the proportions of participants in different groups who achieved the MCID. Calculation of the MCID relied on a normative method because no previous experience with the MAM could be used to establish a criterion anchor.

Method

The MAM was developed and tested for psychometric soundness by following procedures specified elsewhere (see the article by Allen on the validity and reliability of data for the MAM in this Special Series). (9)

Each of the 24 items in the MAM consists of 6 statements indicating levels of movement ability from 1 at the lowest to 6 at the highest. Respondents choose the statement that most closely matches how they think they move now and how they would like to be able to move. The total possible raw score is 144 if a respondent chooses the highest item response levels across all 24 items. The Figure shows an item for the flexibility dimension. Only "now" responses were analyzed for this article, and all analyses were performed on the basis of responses to items without regard to separate dimensions.

All participants in this study met the following criteria: age of 18 years or over, able to read and write English, and physically able to complete the questionnaire. Recruitment for the no-intervention group proceeded via invitation to associates of the author who were not undergoing or about to start a course of physical therapy intervention. Participants in this group completed the MAM twice, at about a 2-week interval.

Recruitment for the intervention group proceeded via verbal invitation when patients began a course of physical therapy intervention at participating outpatient clinical sites. Participants in the intervention group completed the MAM once at the beginning of physical therapy intervention (time 1), again at about 2 weeks into the episode of care (time 2), and again at 2 months or discharge, if the patient was discharged before 2 months had passed (time 3). Participants were included in the analysis if they completed at least 2 MAMs. In addition to the MAMs completed by the participants, each attending physical therapist completed the MAM once after the initial examination (time PT) for each participant in the intervention group. At the end of the episode of care, the attending physical therapist also determined the success of physical therapy intervention and recorded the answer. The training of attending physical therapists consisted of written and verbal instructions to complete 2 tasks: to complete the MAM after the first visit and to indicate at the end of the discharge visit whether physical therapy intervention for the participant was completely successful, partially successful, or not successful (therapists were instructed to choose one). All participants were informed that completing and returning the questionnaire constituted consent for their (anonymous) responses to be included in the study.

Forty-three participants began the study at 9 different outpatient clinics with 12 different attending physical therapists; 35 completed at least 2 MAM questionnaires. The average age of the 35 participants in the intervention group was 47 years, with a minimum and a maximum of 19 and 85 years, respectively. Twenty-five participants were female. The reasons for physical therapy intervention in this group included low back pain; neck or upper back pain; shoulder pain; knee, wrist, or foot surgery; ankle tendinitis; carpal tunnel syndrome; and thoracic outlet syndrome. Of the 12 attending physical therapists, I had a professional doctoral degree (at least 2 have since earned transitional doctor of physical therapy degrees), I had a certification in orthopedic manual therapy, 5 were owners of their clinics, and 2 were managers of corporation-owned clinical facilities. Years of physical therapy experience ranged from about 2 years to over 20 years.

Thirty-six volunteers from the community completed the MAM twice, at about a 2-week interval. Two were undergoing or about to start physical therapy intervention and so were removed from further analysis for the no-intervention group. The average age of the remaining 34 participants in the no-intervention group was 54 years, with a minimum and a maximum of 19 and 78 years, respectively. Seventeen participants were female.

After recruitment, participants in the intervention group received physical therapy intervention as directed by their therapists. Attending physical therapists managed the movement disorders of people in the intervention group without specification or constraint from this study, and intervention records were not collected. If people perceived an improvement in their movement ability with intervention, then they were expected to choose responses at a higher level at time 2 or time 3 than at time 1. Gains on the MAM were expected to be greater for people who had completely successful courses of physical therapy than for those who had partially successful or unsuccessful courses, as designated by their attending physical therapists. People in the no-intervention group were not expected to change their responses on the MAM.

Responses on the MAM were analyzed with item response theory (IRT) methods and ConQuest software. (10), * Other sources provide an overview of IRT methods and the associated terminology (11-13); therefore, only summary statements appear here. With IRT methods, the responses of all participants to the 24 items are converted into the probabilities of those responses and the natural log of the odds (logits) of each participant responding in a particular way to these items. The resulting logits designate the location of each item and participant on an interval scale. The scale for the MAM is the self-perception of movement ability in logit units.

Reliably estimating item locations on a logit scale requires a fairly large heterogeneous sample so that participants' responses cover all actual response levels for all items. Such a requirement could restrict clinical studies to prohibitively large and unrealistically heterogeneous samples. However, IRT methods provide the possibility of anchoring item locations on previous calibrations. With item locations anchored, participant locations can be estimated and compared, even with small and relatively homogeneous samples. For this study, all analyses were performed with the item locations calibrated from a previous study. (9) The previous study included 318 community-dwelling adults representing a heterogeneous mixture of movement ability; about 10% of the adults were starting outpatient physical therapy intervention. The logits for the levels of item responses (technically, item steps, or boundaries between 2 response levels) in the calibration study ranged from -7.7 to +9.8. The mean for the calibration population was set at 0 logit; the standard deviation was unconstrained. Standard errors ranged from 0.7 logit for a few of the lowest levels of responses to items to 0.1 logit for most of the highest levels of responses to items. The item separation reliability for the calibration study was .98. (10)

With item response level locations anchored on these previous calibrations, participant locations in this study were estimated and differences were calculated in 4 pair-wise comparisons of test occasions: from test to retest, time 1 to time 2, time 1 to time 3 (or discharge), and time 1 to time PT. Within each pair, the first test occasion was the baseline. For 6 participants in the intervention group who were discharged at 2 weeks, the completed time 2 MAM response sets also were counted with the time 3 MAM response sets in the analyses comparing baseline and discharge data.

Assessing the sizes of differences within each pair-wise comparison involved 2 procedures. In one procedure, the main effects for each pair were computed in ConQuest (10) and assessed for significant differences. In the other procedure, each pair was examined with paired t tests. For computing the main effect, test occasion could be modeled as a main effect within the item response model for each pair-wise comparison. ConQuest produced estimated logit values and standard errors for each pair, and the differences within pairs were observed. The main effect of test occasion was deemed significant if the difference within the pairs of estimated logit values was greater than 1.96 times the standard error. (10) For examining the results of paired t tests, differences in participant location estimates over time were calculated for each participant by subtracting the estimated baseline location from the designated comparison location. Average differences across participants were assessed with paired t tests. One-tailed analyses were performed for time 1-to-time 2 and time 1-to-time 3 comparisons because of the anticipated increase in perceived movement ability following intervention. Two-tailed analyses were performed for test-to-retest and time 1-to-time PT comparisons. The test-to-retest [alpha] value was set at .05. Because the same baseline data were used for the other 3 pair-wise comparisons, the Bonferroni correction for 3 comparisons (14) converted the a value to .017.

Effect size (14) was calculated as the average difference in the locations of participants from baseline to comparison occasion divided by the standard deviation of the distribution of baseline locations. The SRM (3) was calculated with the same numerator divided by the standard deviation of the distribution of differences in locations from baseline to comparison occasion. The responsiveness index (4) was calculated with the same numerator divided by the standard deviation of the distribution of differences in locations from baseline to comparison occasion for the no-intervention participants. Effect size thus relates change to the initial variability in the same group, SRM relates change to the variability of changes over test occasions in the same group, and the responsiveness index relates change to the variability of changes over test occasions specifically in the no-intervention group.

The MCID was calculated with the equation:

SEM = [s.sub.1] x [(1 - Cronbach [alpha]).sup.1/2],

as recommended by Wells et al, (8) to signify important change for individuals. In this equation, [s.sub.1] is the standard deviation of the participant locations at baseline and Cronbach [alpha] is a measure of internal consistency, a type of reliability.

Comparing the MAM responses of intervention group participants and attending physical therapists was a form of interrater reliability testing; therefore, an additional calculation, the intraclass correlation coefficient [ICC(2,k)], was carried out with SPSS software (version 11.0) ([dagger]) to examine their relationship. Intraclass correlation coefficient model 2 was selected so that results could be generalized to different therapists who might treat a participant during an episode of care. Intraclass correlation coefficient form k=24 was selected because the locations of participants were estimates based on responses to 24 items.

The intervention group was further subdivided into groups based on whether the attending physical therapist deemed the intervention completely successful, partially successful, or not successful. Average gains for each of the smaller groups were calculated and compared with 1-tailed independent t tests.

Results

The main effect of test occasion in the pair-wise comparisons, obtained with ConQuest, was significant in comparisons of time 1 and time 2, time 1 and time 3, and time 1 and time PT (P<.05 for each comparison) but not for test and retest. Standard errors for the main effect ranged from 0.03 logit to 0.05 logit. The sizes of the main effect of test occasion ranged from 0.01 logit for analysis of test and retest to 2.09 logits for analysis of time 1 and time 3.

Paired t tests revealed that gains in the locations for participants who underwent outpatient physical therapy averaged 0.98 logit at 2 weeks, equivalent to about 7 raw score points for the average participant (n=35, P<.00002). Gains in the locations for participants averaged 2.53 logits at 2 months or discharge, equivalent to about 17 raw score points for the average participant (n=25, P<.00005). In comparison, participants in the no-intervention group showed no significant change in locations, with an average difference of -0.10 logit after approximately 2 weeks (n=34, P=.22). In contrast to the results obtained with the main effects procedure, paired t tests revealed no significant difference between time 1 and time PT locations (n=34, P=.27).

With a change of 2.53 logits at 2 months or discharge, the effect size was 0.90, the SRM was 0.93, and the responsiveness index (4) was 5.62 (Tab. 1). The effect size was much smaller (0.38) after only 2 weeks of intervention.

The MCID was calculated to be 0.61 logit by the normative (SEM) methods with the standard deviation of 2.55 logits from the initial locations of the 35 participants in the intervention group (time 1 to time 2 pairwise comparison) and the Cronbach a of .94 calculated for the MAM (see the article by Allen on the validity and reliability of the Movement Ability Measure in this Special Series). Twenty of 35 participants (57%) with time 1 and time 2 data showed a change on the MAM equal to or greater than the MCID at 2 weeks. Twenty of 25 participants (80%) with time 3 or discharge data showed a change on the MAM equal to or greater than the MCID at 2 months or discharge.

For participant locations estimated from participant time 1 data and from physical therapist time PT data (n=34), the ICC(2,24) was .78. Locations resulting from physical therapist perceptions of participants' movement ability averaged 0.44 logit lower than locations resulting from participants' perceptions of their own movement ability. Physical therapists' responses across items were much less variable than participants' responses on the MAM (variances of 4.81 and 9.80, respectively).

For the 25 participants for whom a physical therapist assessed the success of the intervention, a 1-tailed t test revealed a significant difference (P<.05) between those assessed as having a partially successful intervention (n = 12, average gain of 1.32 logits) and those assessed as having a completely successful intervention (n = 13, average gain of 3.30 logits). No physical therapist designated a course of intervention as not successful for any of these participants.

Discussion and Conclusion

In this study, MAM responses changed between test occasions during a course of physical therapy, as indicated by a significant main effect of test occasion and significant gains over time for the intervention group but not for the no-intervention group. The lack of gains in MAM responses for the no-intervention group indicates that completing the MAM a second time does not necessarily change people's movement ability responses. For the intervention group, the effect size was 0.90 at 2 months or discharge; this value was considered a large standardized measure of change. (14) Because of the large effect size, the power of this 1-tailed analysis of paired data was over 90% even with a sample size of 25 pairs. (14) The SRM was close to 1, indicating that the average change was approximately equal to the standard deviation of changes across occasions. The responsiveness index revealed that the 2-month average change was more than 5.5 times the standard deviation of change in the no-intervention group. The change in most participants' MAM responses at 2 months or discharge was larger than the size of the distribution-calculated MCID, implying that the change was large enough to make a difference in how well participants thought they moved for their daily activities. In addition, the change in MAM responses was larger for participants who had a completely successful course of physical therapy intervention than for those who had a course deemed only partially successful by their physical therapists.

Although the MAM was able to record change in the intervention group, potential biases limit the generalizability of the changes observed here. Participants in both groups were volunteers, possibly having a bias toward interest in movement ability. In addition, participants in the intervention group had multiple physical therapy intervention sessions, which would help focus their attention on movement, thereby enhancing this potential bias in a way not duplicated in the no-intervention group. Participants who completed the study may have had a positive bias toward physical therapy intervention; those who did not have a positive bias or who were initially skeptical about physical therapy intervention could have dropped out of the study or refused to start. Also, participants who completed the study may have experienced greater changes along the way than, for example, the 8 participants in the intervention group who dropped out before completing a second MAM. The physical therapists' bias toward greater change is revealed by the fact that no physical therapist judged a course of intervention not successful for participants who completed the study. Physical therapists may have overestimated the success of the intervention, because for 3 of 13 participants judged to have partially successful therapy and for 2 of 12 participants judged to have completely successful therapy, there was a difference between baseline and discharge that was less than the MCID.

To examine further the possibility of participants' bias toward change, data and gains at times 2 and 3 were analyzed. Of the 15 participants who showed a change of less than 0.61 logit (MCID) at 2 weeks, 3 participants were specified as being discharged at that point, 6 participants had no time 3 data, 2 participants continued but still showed a change of less than 0.61 logit at 2 months, and 4 participants continued and showed a change greater than the MCID at discharge. In comparison, 5 of the 20 participants who showed a change greater than the MCID at 2 weeks had no time 3 data. Therefore, dropping out of the study occurred among both participants who were making good progress and those who were not. Replication of this study is needed to determine whether the apparent difference in proportions of participants dropping out (6 of 15 showing less change and 5 of 20 showing change equal to or greater than the MCID) is significant or whether similar changes recorded with the MAM across occasions occur in other samples.

Note that any changes recorded in this study cannot be attributed definitely to the effect of physical therapy intervention, although the time course coincided with an episode of care. Nor can the lack of change in MAM responses for the no-intervention group be attributed definitely to a lack of change in movement ability for this group, although no intervention took place. A randomized controlled trial with the MAM as a dependent variable could help to establish the usefulness of the MAM in providing evidence of the effectiveness of a particular intervention compared with the lack of that intervention.

Although the ConQuest modeling method indicated that the main effect of test occasion was significant between time 1 and time PT, the paired t test for this comparison revealed no significant difference. The ICC for this comparison was relatively strong, at .78. The mixed results indicated that there probably were some differences in how physical therapists and participants judged movement ability but that their judgments were fairly strongly correlated. The time PT data would provide stronger evidence supporting the accuracy of movement ability reports from participants if not for large differences in variances for the 2 groups. The physical therapists tended to mark one level across most items on some of their completed MAM questionnaires, possibly reflecting time constraints, with a resultant restriction in variability unmatched by participants' responses. Participants tended to perceive their own movement ability at a higher level than their therapists did at the beginning of the course of physical therapy intervention. Future studies could use performance-based data or therapists' written examinations for comparison with patients' MAM responses to determine whether the patients' perceptions of change in movement ability have corollary evidence to support them.

Few clinicians have the expertise locally available to run IRT analyses of data gathered from their patients. The classical scoring method of adding up points across items on a measure like the MAM can provide some indication of people's perceptions of movement ability. Performing statistical analyses and comparisons such as those used in this study, however, assumes that the underlying scale, like the logit scale, has the same intervals between units. The units in a raw score do not meet this criterion in measures such as the MAM, in which item response levels are ordinal. To facilitate clinical utilization, Table 2 shows the conversion of some raw scores into logits on the basis of a previously performed calibration study. The raw scores for the intervention group at time 3 in this study averaged about 96; the raw scores for the no-intervention group at both test and retest averaged about 109.

In a self-report instrument, responses depend on people's perceptions and how they interpret the instrument at the moment. People could interpret items on the MAM to mean a range of movement abilities through a day or a week, and responses could depend on whether medication for pain, stiffness, or dizziness works well or whether people focus on gross or fine motor movements. Thus, responses reflect some mixture of actual movement ability and perception of that movement ability. Given possible differences in the interpretation of items and the mixture of movement ability and perception, the evidence obtained for the responsiveness of the MAM in this project seems quite extraordinary; perhaps people decide what their reference point for movement ability is and stick to it, or perhaps the presentation of the MAM helps to direct them in some way. Gathering evidence of validity through more formalized response processes in future projects could help to clarify these issues. (15)

In this study, the MAM reflected changes that were both statistically significant and clinically meaningful. The MAM shows promise as a way to document gains in perceived movement ability across a population receiving outpatient physical therapy intervention. Generalization to diagnostic groups other than the ones represented here or settings other than outpatient clinics will require further research.

The author acknowledges Rick Allen for support and editing advice throughout the process of conceptualizing, testing, and writing. The author thanks the physical therapists who volunteered to participate in the study and the respondents in both groups who completed the MAM.

The Committee for the Protection of Human Subjects at the University of California, Berkeley, designated this study exempt from further review.

Data and constructs of this manuscript were presented in a platform presentation at the International Objective Measurement Workshop; April 6, 2006; Berkeley, Calif.

This article arose from the author's doctoral dissertation at the University of California, Berkeley, December 2005.

This article was received July 12, 2006, and was accepted March 1, 2007.

DOI: 10.2522/ptj.20060198

Reference

(1) Beaton DE, Bombardier C, Katz JN, Wright JG. A taxonomy for responsiveness. J Clin Epidemiol. 2001;54:1204-1217.

(2) Cott CA, Finch E, Gasner D, et al. The movement continuum theory of physical therapy. Physiother Can. 1995;47:87-95.

(3) Stratford PW, Binkley JM, Riddle DL. Health status measures: strategies and analytic methods for assessing change scores. Phys Ther. 1996;76:1109-1123.

(4) Guyatt G, Walter S, Norman G. Measuring change over time: assessing the usefulness of evaluative instruments. J Chronic Dis. 1987;40:171-178.

(5) Haley SM, Fragala-Pinkham MA. Interpreting change scores of tests and measures used in physical therapy. Phys Ther. 2006; 86:735-743.

(6) Shikiar R, Harding G, Leahy M, Lennox RD. Minimal important difference (MID) of the Dermatology Life Quality Index (DLQI): results from patients with chronic idiopathic urticaria. Health Qual Life Outcomes. 2005;3:36.

(7) Lord SE, Wade DT, Halligan PW. A comparison of two physiotherapy treatment approaches to improve walking in multiple sclerosis: a pilot randomized controlled study. Clin Rehabil. 1998;12: 477-486.

(8) Wells G, Beaton D, Shea B, et al. Minimal clinically important differences: review of methods. J Rheumatol. 2001;28:406-412.

(9) Allen DD. Validity, Reliability, and Responsiveness of the Movement Ability Measure, a New Instrument Proposed for Assessing Physical Therapist Competence [dissertation]. Berkeley, Calif: Graduate School of Education, University of California; 2005.

(10) ACER ConQuest: Generalised Item Response Modelling Software. Version 2.0 [computer program]. Hawthorn, Victoria, Australia: ACER (Australian Council for Educational Research) Press; 2003.

(11) Wilson M. Constructing Measures: An Item Response Modeling Approach. Mahwah, NJ: Erlbaum; 2005.

(12) Wilson M, Allen DD, Li JC. Improving measurement in health education and health behavior research using item response modeling: introducing item response modeling. Health Educ Res. 2006;21(suppl 1): i4-i18.

(13) Wright BD, Masters GN. Rating Scale Analysis. Chicago, Ill: MESA Press; 1982.

(14) Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Practice. 2nd ed. Upper Saddle River, NJ: Prentice Hall Health; 2000.

(15) American Educational Research Association, American Psychological Association, National Council on Measurement in Education. Standards for Educational and Psychological Testing. Washington, DC: American Educational Research Association; 1999.

* Australian Council for Educational Research, Hawthorn, Victoria, Australia.

([dagger]) SPSS Inc, 233 S Wacker Dr, Chicago, IL 60606.

DD Allen, PT, PhD, is Adjunct Associate Professor, Department of Physical Therapy, Samuel Merritt College, Oakland, Calif, and Post-Doctoral Fellow, Health and Disability Research Institute, Boston University, Boston, Mass. Address all correspondence to Dr Allen at: [e-mail].

[Allen DD. Responsiveness of the Movement Ability Measure: a self-report instrument proposed for assessing the effectiveness of physical therapy intervention. Phys Ther. 2007;87:917-924.]

Table 1.
Four Pair-Wise Comparisons (a) of Respondent Locations on the Movement
Ability Measure for the No-Intervention (NI) and Intervention (I)
Groups

Group:           [bar.X] (SD) Logits (b) for:
Pair-Wise

Comparison
(No. of Pairs)   Baseline             Comparison        Difference
                 Occasion             Occasion          Between
                                                        Baseline
                                                        Occasion and
                                                        Comparison
                                                        Occasion (c)

NI: test to      -1.79 (0.68)         -1.89 (0.67)      -0.10 (0.45)
  retest (34)

I: time 1 to     -3.02 (2.55)         -2.04 (2.69)       0.98 (1.24)
  time 2 (35)

I: time 1 to     -3.30 (2.81)         -0.77 (3.72)       2.53 (2.72)
  time 3 (25)

I: time 1 to     -2.79 (3.13)         -3.24 (2.19)      0.44 (2.29)
  time PT (34)

Group:
Pair-Wise        Effect Size          Standardized      Responsiveness
                 ([bar.X] for         Response          Index ([bar.X]
Comparison       Difference Between   Mean ([bar.X]     for Difference
(No. of Pairs)   Baseline             for Difference    Between
                 Occasion and         Between           Baseline
                 Comparison           Baseline          Occasion and
                 Occasion/SD          Occasion and      Comparison
                 for Baseline         Comparison        Occasion/SD
                 Occasion)            Occasion/SD       for NI
                                      for Difference)   Difference)

NI: test to      -0.15                -0.22             Not applicable
  retest (34)

I: time 1 to      0.38                 0.79              2.18
  time 2 (35)

I: time 1 to      0.90                 0.93              5.62
  time 3 (25)

I: time 1 to      0.14                -0.19             -0.98
  time PT (34)

(a) For group NI, baseline occasion=test and comparison
occasion=retest, for group I, baseline occasion=time 1 and comparison
occasions=time 2, time 3, or time PT; time PT indicates the occasion
at which the attending physical therapist completed the Movement
Ability Measure after the initial examination.

(b) A smaller (more negative) logit value means less perceived movement
ability.

(c) A negative value means less perceived movement ability at the
comparison occasion than at the baseline occasion.

Table 2.
Conversion Table for the Movement Ability Measure: From Raw Score'
to Logits (Standard Error) and Movement Ability Levels

Movement Ability Level                   Raw Score of     Lowest Logit
                                         at Least:        Value (SE)

Moves competitively                      121               3.54 (0.44)
Moves for normal activities plus extra    94              -1.12 (0.39)
  activities
Moves for normal activities               73              -4.16 (0.37)
Moves with difficulty but can do
  activities                              46              -6.84 (0.30)
Moves some, needs help                    28              -9.50 (0.56)
Needs someone else to do activities
  instead                                 24              -10.00 (0.65)

Movement Ability Level                   Raw Score Not    Highest Logit
                                         Greater Than:    Value (SE)

Moves competitively                      144              10.00 (1.14)
Moves for normal activities plus extra   120               3.34 (0.44)
  activities
Moves for normal activities               93              -1.27 (0.39)
Moves with difficulty but can do
  activities                              72              -4.30 (0.36)
Moves some, needs help                    45              -6.93 (0.31)
Needs someone else to do activities
  instead                                 27              -9.83 (0.62)

(a) For these raw scores, it is assumed that the respondents completed
all items in the Movement Ability Measure.

Figure.
Movement Ability Measure item directed toward the dimension of
flexibility. Participants were instructed to choose the one statement
that most closely matched their usual ability to move now, this week,
and the one statement that most closely matched the ability that they
would like to have even if they had to work hard for it. They were
instructed to mark one number on the left (Now) and one number on the
right (Would Like).

                                                                Would
Now                                                              Like

 6     I move so easily that I can stretch or reach extra         6
       far compared to others.

 5     I move easily enough for all my normal tasks plus          5
       free time and play activities.

 4     I move easily enough to do all my normal tasks but         4
       not more.

 3     I move with difficulty in my normal tasks because          3
       of stiffness or tightness.

 2     I have so much stiffness or tightness that I need          2
       special equipment or help to do some of my normal
       tasks.

 1     Stiffness or tightness keeps me from doing most of         1
       my daily care.