Background and Purpose The Movement Continuum Theory (MCT) provides a potential basis for movement assessment and intervention, but “movement” lacks specificity. The purposes of this study were to propose and evaluate a subdivision of movement into multiple dimensions.
Subjects A convenience sample of 318 adults completed a 24-item self-report measure of movement ability.
Methods A multimethod approach was used to identify, operationalize, and test a multidimensional model of movement. Data analysis included a comparison of the fit of unidimensional and multidimensional models using item response theory methods and inspection of response patterns.
Results A model specifying 6 dimensions—flexibility, strength, accuracy, speed, adaptability, and endurance—fit respondent data significantly better than the unidimensional model, even with high pair-wise correlations between dimensions. Response patterns showed large differences rather than uniform scores across dimensions for over half of the respondents.
Discussion and Conclusion Subdividing movement into the proposed dimensions fits the data and potentially strengthens the usefulness of the MCT as a theoretical foundation for managing movement effectively.
The Movement Continuum Theory (MCT),1 first published in 1995, establishes links among movement sciences, the movement capability of individuals, and the role of movement specialists in maximizing people's movement capability. The MCT1 presents movement as the central unifying construct for the assessment and management of movement and movement disorders instead of the common clinical practice of focusing on function or disability.2 Its authors proposed it as a possible grand theory of physical therapy,1 but the MCT and its principles can enhance the understanding of movement and potential interventions by other professions as well.
Despite broad relevance and a need for theoretical foundations for clinical practice,1,3 the MCT has inspired little empirical research since its introduction. In a search of CINAHL and MEDLINE databases as of August 2005, none of the 24 articles referring to the MCT since its publication contained accounts of prospective testing of the MCT or any hypotheses stemming from it.
This study initiates testing of the MCT in a direction that could ease the application of this theory to empirical research. In this study, the construct of movement is subdivided into multiple components or dimensions that may prove more readily measurable than the singular generic movement construct presented in the MCT. A multidimensional model such as the model proposed here may stimulate both the generation of testable hypotheses and the association of current evidence of effectiveness with a unified theory. A multidimensional model of movement also may promote the characterization of people's different movement abilities, enhancing the specificity with which clients and movement specialists can pinpoint deficits and identify appropriate interventions. The purposes of this study were to propose a multidimensional model of movement as an extension of the MCT and to perform an initial evaluation of this new model of movement.
The MCT presents 3 general and 6 physical therapy principles that link movement science with movement capability and clinical practice.1 In essence, movement, defined as an actual change in position, occurs at multiple interacting levels along a continuum from microscopic to the level of a person acting in society. Each level is influenced by physical, social, psychological, and environmental factors. Physical agents and therapeutic exercise generally have entry points at the tissue level or higher, but because the levels interact, these interventions can affect molecular and cellular movement as well as body part and person movement. The MCT specifies that each person has maximum, current, and preferred movement capabilities. If a movement specialist successfully addresses movement problems with a patient or client, then current movement capability will increase and the gap between current and preferred movement capabilities will narrow.1
Testing the principles presented by the MCT requires an assessment of people's current and preferred movement capabilities and the effect of intervention on them. The construct of movement as presented in the MCT, however, is too generic for clinical assessment. Specifying subdivisions or dimensions of movement may assist in identifying clinically measurable constructs that have a definitive relationship to the movement capabilities presented in the MCT. Because the MCT already presents a framework for identifying what part of the person moves (eg, at the tissue, body part, or person level on the continuum) and for identifying physical, psychological, social, and environmental factors that influence the movement,1 these aspects of observable behavior do not require redundant description. Only the movement itself requires further specification.
The specification of multiple subdivisions or dimensions of movement has support in the movement science and clinical literature. Clinical4 and motor control5 sources present strength, flexibility, proprioception, and coordination as candidates for intervention following orthopedic or neurologic pathology. Some of these sensorimotor aspects overlap with the list that Hedman et al6 compiled as the “components of movement” or that Majsak7 identified as constraints delineating the “range of movement behaviors.” Additional overlap and alternative ways of specifying aspects of movement appear in Craik's discussion of issues for defining normal motor behavior8 and the classification that Scheets et al9 formulated for diagnosing impairment of the movement system. Each of the movement aspects and components mentioned in these sources could contribute to a multidimensional model of movement.
Phases of Study
This article describes 3 phases of a multimethod study. The purposes were to formulate and evaluate a multidimensional model of movement to extend the MCT. In the identification phase, components of movement from the literature were evaluated on the basis of a set of criteria for inclusion into an economical model. In the operation phase, the set of dimensions and the MCT formed the basis of a new measure constructed to incorporate both generic and multidimensional constructs of movement. In the test phase, data were collected with the new measure. The proposed multidimensional model then was compared with a unidimensional model of movement and with a multidimensional model with randomly attributed dimensions. Because the phases necessarily occurred sequentially, the results follow the method for each phase in sequence.
Identification Phase: Method and Results
Generating the set of potential movement dimensions consisted of setting evaluative criteria, identifying from literature sources common features of movement to propose as candidates, and comparing those candidates with the criteria to ensure alignment. The criteria for potential dimensions of movement to extend the MCT included the following:
Descriptive: The complete set of dimensions, with an added reference to the body parts or substances doing the moving, should fully describe normal human movement, a series of movements, or actively holding a position against a force.
Efficient: The set of dimensions should describe movement efficiently, subsuming related concepts, with the fewest number of separate dimensions while completely describing movement.
Distinct: The dimensions should identify observable features of movement distinct from the part of the body doing the moving or different physical, psychological, social, or environmental factors that influence movement.
Measurable: The dimensions should be measurable.
Understandable: The dimensions should make sense to both movement specialists and their patients or clients.
A comparison of possible movement dimensions with the criteria led to the addition, modification, or elimination of candidates. Tables 1 and 2 show comparisons of the first 4 criteria with the proposed (Tab. 1) and some of the rejected (Tab. 2) candidates for movement dimensions. The fifth criterion implies that people can differentiate among and use the various dimensions in their observations and descriptions of movement. Testing this implication or otherwise providing evidence of understanding of any of the movement dimensions will require empirical data.
The resulting set of dimensions includes flexibility, strength (force exerted), accuracy, speed, adaptability, and endurance. These 6 dimensions describe observed movement comprehensively and efficiently (criteria 1 and 2). The proposed dimensions of flexibility, strength, and speed apply to all human movement; accuracy applies specifically to purposeful movement; and adaptability and endurance apply to movement when encountering unexpected obstacles or when approaching the limits of a person's capacity. All of these dimensions have direct relationships with but remain distinct from the physical, psychological, social, and environmental factors that influence movement (criterion 3). Each candidate dimension can be measured clinically (criterion 4). Although further research may justify modification of this set, these 6 dimensions provide a starting point for characterizing movements readily observed by movement specialists and their patients or clients (toward criterion 5). In addition, these 6 dimensions present interesting possibilities for categorizing movement abilities maximized by athletes or performing artists or diminished in people with a particular pathologic condition (Tab. 3).
Operation Phase: Method and Results
The next step in determining the usefulness of this set of dimensions was to construct or locate measures for assessing movement. If the same measure could evaluate both generic and multidimensional movements, then it would facilitate the direct comparison of a generic or overall idea of movement with the dimensions of movement proposed in the model. In addition, because the MCT and the proposed model apply to a broad range of ability levels and to the movements of people with or without pathologic conditions, the ideal measure for comparing generic and multidimensional constructs would apply to a similar range. Many measures of movement exist for testing individual dimensions, specific diagnostic groups, or particular body parts exist, but few existing measures assess generic movement ability or apply to multiple groups or across the proposed dimensions.
Generation of the self-report Movement Ability Measure (MAM) operationalized the MCT and the proposed model and facilitated direct comparison of unidimensional and multidimensional models of movement. For addressing a generic or unidimensional construct of movement, all items in the MAM were given a similar item construction and standard levels of item responses. If people marked every item with the same level of response, then a generic movement construct could specify their movement ability quite adequately. For addressing a multidimensional construct of movement, variations in the wording of items in the MAM referred specifically to the 6 proposed dimensions of movement. If people marked items associated with one dimension quite differently from items associated with other dimensions, then specification of their ability on that dimension could enhance the description of their movement ability.
The self-report format allowed subjects to interpret movement as a whole or differentiate movement dimensions within the context of their own lives. The MAM placed minimal constraints on subject interpretation. In avoiding the specification of tasks that may have limited relevance across groups, the MAM also applied to a broad range of subjects across movement ability levels and with or without pathologic conditions.
The MAM was developed and tested for reliability and for content and construct validity with procedures recommended by Wilson10; evidence of reliability and validity is presented elsewhere (see the article by Allen on the validity and reliability of the Movement Ability Measure in this Special Series).11 Each item in the MAM consisted of 6 statements indicating levels of movement ability. Respondents were instructed to choose the statement that most closely matched how they thought they moved now and how they would like to be able to move. Three sample items and instructions are shown in Figure 1. The MAM included 4 items for each of the 6 dimensions, for a total of 24 items. The same instructions applied to all items. Consistency of responses across items was high, with person separation reliability ranging from .92 to .96 for the 6 dimensions and equaling .98 for the whole measure.
Test Phase: Method
For the test phase, a heterogeneous sample of people completed the MAM. The expectation was that most people who move normally might perceive themselves to have about the same level of movement ability on all 6 dimensions; therefore, a unidimensional model would fit the data very well. If people perceive differences in the effects of different dimensions on their movement ability, then they might respond quite differently to items associated with those dimensions. In this situation, a multidimensional model would fit the data better than a unidimensional model. The proposed multidimensional model was compared with a unidimensional model and with a multidimensional model in which items were randomly assigned to dimensions.
Recruitment of volunteers to respond to the MAM targeted a broad spectrum of groups in order to obtain a heterogeneous representation of movement abilities. Adults volunteered from religious and community groups, personal contacts, a college sports team, physical therapy outpatient clinics, and a senior day activity event. In addition to the MAM, respondents completed a cover sheet of information about health status and any movement problems. Respondents were informed that completing and returning the questionnaire constituted consent for their (anonymous) responses to be included in the study.
The data were analyzed with item response theory (IRT) methods12 and ConQuest13,* software, and only the “now” responses to items were analyzed. Two models were compared. One model assigned all items to 1 dimension in a unidimensional construct; the other assigned items to the 6 dimensions in a multidimensional construct. Fit was analyzed on the basis of the differences in the deviances and the numbers of parameters (obtained from ConQuest) by use of the G2 likelihood ratio statistic. For a more complex (multidimensional) model to fit better than a simpler nested (unidimensional) model, it must result in a lower deviance (a measure of lack of fit of the data to the model) than can be accounted for simply by the greater number of parameters estimated. The difference between the deviances for the 2 models functions like a chi-square distribution with the difference in the number of parameters as the degrees of freedom. Correlations also were obtained for each pair of dimensions in the multidimensional model.
To assess whether any multidimensional model would fit better than the unidimensional model for these data, a random multidimensional model was generated, with items assigned randomly, but without replication, to generic dimensions. That is, no more than one item from any proposed dimension was allowed per generic dimension. This random multidimensional model also was compared with the unidimensional model with the G2 likelihood ratio statistic as described previously.
In addition to the comparisons of models with the G2 statistic, the patterns of responses of individual respondents were examined. Examining uniform or uneven patterns of responses across dimensions might provide insight into the constructs in the proposed model. A sum of squares indicator, DI, was calculated to indicate the sizes of the differences in responses across dimensions.14 For this calculation, movement levels and respondent abilities (θ) were examined in logits, the log of the odds of choosing the statement indicating a given level of movement ability within each item. The DI sums differences from movement ability estimates across the 6 dimensions (d) for each person p, as follows: If the sum of the squared deviations from an average estimate is low, then that person perceives his or her movement to be about the same across all 6 dimensions. If DIp is high, then that person perceives movement ability on at least one of the dimensions to be quite different from the average of the rest. Representative respondents with low and high DI values were selected; movement ability plots (MAPs) depicted the asymmetry of dimensions for these selected respondents with low and high DI values. Designation of low and high DI values within any particular study is arbitrary.15 For this study, the lowest and highest average logits for any dimension were inspected for each respondent; the DI cutoff was assigned to the value above which all respondents had differences from their lowest to their highest dimensions that were large enough to be outside of a 98% confidence interval.
Test Phase: Results
A total of 318 adults completed the MAM. Respondent ages ranged from 18 to 101 years, with modes (10 each) at ages 49 and 76. Women constituted 206 (65%) of the respondents; 178 (56%) acknowledged at least a little movement difficulty in the previous week. Forty-six respondents (14%) indicated that they were starting or undergoing physical therapy at the time of responding to the MAM.
With items specifically assigned to 6 corresponding dimensions, the multidimensional model fit significantly better than the unidimensional model (χ225=280.9, P<.0001), even with high internal consistency across all items (Cronbach α=.94) and high correlations between pairs of dimentions (r=.87–.99). In contrast, when items were randomly assigned to 6 generic dimensions, the multidimensional model fit no differently than the unidimensional model (χ225=23.3, P=.56).
When response patterns were examined with the DI statistic (mean=9.25 logits2, standard deviation=11.62), 5.3 logits2 was designated as the cutoff between low and high. No person who had a DI value above this cutoff had less than 2.5 logits between the lowest and the highest average dimension estimates. At 2.5 logits, the lowest and highest average dimension estimates were outside of their respective 98% confidence intervals (standard errors for average dimension estimates were about 1 logit), and the spread signified at least 0.5 and up to 1.25 movement ability level differences between the dimensions. At a DI value of 5.3, 165 (52%) of the respondents showed differences between the dimensions of movement rather than a uniform average across dimensions. Movement ability plots of sample cases (Figs. 2, 3, 4, 5, 6, and 7) chosen to represent low and high DI values depict dimensional abilities in logits along 6 respective axes in a hexagon (range for all axes=−11 to +9 logits). Greater asymmetry indicates larger differences between dimensions. Demographic information is provided when known from responses and comments on completed questionnaires.
Discussion and Conclusion
The 3 phases of this study resulted in a proposed set of dimensions to extend the construct of movement within the MCT. The proposed dimensions included aspects of movement that were described in the literature and that were aligned with evaluative criteria. Testing the proposed dimensions required the construction of a new measure targeting these movement constructs along with a generic movement construct. Model comparisons carried out with data obtained with the new measure showed that the proposed multidimensional model fit better than a unidimensional model.
Despite the dimension-specific wording of the MAM, many respondents provided no discernible indication that their movement was different across dimensions. For them, responses across the dimensions indicated about the same level of movement ability, although that movement ability might have been low or high, as shown in Figures 2 and 3. The associated demographic data indicated that symmetry in responses across dimensions might have been associated with debilitation or physical capability in general.
For more than half of the respondents in this study, MAM responses were different across dimensions. Some respondents showed exceptionally low levels of ability on some dimensions (Figs. 5 and 7), and others showed exceptionally high levels of ability on 1 or 2 dimensions (Fig. 6). These responses imply sufficient understanding of the dimensions in the MAM to reflect consistent differences (with person separation reliability ranging from .92 to .96) across designated groups of items. This is initial evidence that this set of dimensions may meet criterion 5. Determining whether such differences across dimensions have clinical meaning depends on future research. Comparing the demographic data to the MAPs suggested a link between responses and respondent characteristics rather than either uniform or random responses to items.
Although these results provide some initial evidence supporting the subdivision of the movement construct of the MCT into the 6 proposed dimensions, validation of the proposed model requires further research. For example, the MAM deliberately allowed respondents to interpret items without specifying standard tasks; this property increased its applicability across individuals with different experiences of functional activities but restricted the absolute comparison of one individual with another or of MAM responses with instrumented measures. To determine whether differences in perceived movement ability correlate with measurable differences in dimensions, future research might examine the association between MAM responses and performance-based measures or clinicians’ judgments of movement ability. To determine whether the magnitude of perceived movement ability has meaning, future research might examine group data for each dimension and compare healthy control subjects with subjects who have identified deficiencies. To explore the possible clinical meaning of the proposed dimensionality, future research might examine people before and after therapy to determine whether those who respond well to therapy started with a generic lack of movement ability across all dimensions or a specific and predictable lack of movement ability in one dimension or a few dimensions. Further research also might indicate that MAPs reveal identifiable patterns of asymmetry for certain clinical groups.
Asymmetry across different dimensions should follow predictable patterns according to the proposed multidimensional model of movement. For example, athletes should test higher in predictable subsets of these dimensions, depending on the requirements of their specific sporting events, as proposed in Table 3. Likewise, patients should test lower in predictable ways if they have diagnoses affecting 1 or several designated dimensions. Furthermore, if these dimensions extend the MCT, then patients should improve in affected dimensions upon successful completion of a clinical intervention. If research confirms predictable patterns among the dimensions related to athletic ability or pathology-related disability, then characterization of movement ability along the dimensions may prove useful in determining prognosis and planning for client intervention.
A common alternative statistical method for determining dimensionality, factor analysis, proved unhelpful in this study. Exploratory or confirmatory factor analysis of an instrument relies on a lack of correlation between groups of items or dimensions to determine whether different factors are represented. For perceived movement ability as assessed with the MAM, the dimensions had an extremely high pair-wise correlation that negated confirmation of factors with factor analysis. Choosing IRT methods to test dimensionality proved more useful in this study because these methods estimate item and respondent locations on the same (logit) scale on the basis of all of the recorded responses to all of the items. Thus, IRT methods retain the distinctions between items and groups of items made by individual respondents rather than subsuming all of those differences in pooled correlation data across a sample.
Although the MCT describes movement at all levels, from the molecular and cellular levels to the level of a person acting in society, the MAM incorporates the 6 dimensions of readily perceivable movement only. Further research is needed to determine whether these 6 dimensions apply to the molecular and cellular levels of the continuum described by the MCT or whether separate movement descriptors are more applicable for these levels.
Although numerous discussions with professional informants helped refine the set of dimensions described here and although these dimensions met the evaluative criteria, the literature search for movement dimension candidates was neither exhaustive nor systematic. Further research may provide support for the exclusivity of these dimensions or provide some other criteria for accepting different dimension candidates. Research also may modify the concepts of these dimensions, splitting some into smaller subdivisions or merging others on the basis of some alternative criteria. It is hoped that the identification of the 6 dimensions in this study will promote discussion of movement and all of its possible dimensions.
The subjects in this study were not a randomized sample; subjects who volunteered to complete the self-report measure may have self-selected either because they thought they moved well or because they were conscious of movement problems. Neither of these motivations was thought to bias the results particularly, as this study focused on dimensionality and not the level of movement ability.
An alternative to the disablement models described as the basis of the Guide to Physical Therapist Practice,2 the MCT1 presents a potential grand theory of physical therapy3 that also could be relevant to movement specialists in other professions. Without testable hypotheses, however, the MCT will fail to provide a foundation for assessment and intervention. The proposed multidimensional model may promote hypothesis generation because the specificity of the dimensions makes measuring movement with the MCT more concrete. Strength, for example, as a dimension within the movement construct of the MCT, has links among the assessment of strength in the laboratory, the problems of weakness, and the intervention used to improve current ability to generate force. Characterizing movement capabilities across dimensions and testing any narrowing of the gap between current and preferred movement capabilities as a result of intervention become possible.
If the research suggested in this discussion further supports the MCT and the proposed dimensions of movement, it will have implications affecting research, education, and clinical practice. In research, the MCT and dimensions of movement could provide a framework for revealing relationships among flexibility, strength, and speed, for example, providing a needed unification for effectiveness evidence. In education, a focus on movement dimensions provides a natural link between basic and movement sciences and the movement deficits associated with particular pathologic conditions, perhaps improving student understanding of assessment and intervention relationships across diagnostic groups. In clinical practice, the dimensions of movement may help patients and movement specialists more readily specify and focus assessment and intervention on the dimensions having the most difficulty. Across all areas, dissemination and use of the MCT and dimensions of movement could enhance effectiveness in investigating and managing movement. Although this study addressed only the initial testing of the proposed multidimensional model of movement and the MCT, the potential usefulness of this theory makes further research worthwhile.
The author acknowledges Mark Wilson for sparking the original idea of dimensions of movement and for his direction in the methodology of testing. The author also thanks Rick Allen for support and editing advice throughout the process of conceptualizing, testing, and writing.
A version of this study was presented as a poster at the Combined Sections Meeting of the American Physical Therapy Association; February 1–5, 2006; San Diego, Calif. This study was completed as part of the author's doctoral dissertation at the University of California, Berkeley.
The Committee for the Protection of Human Subjects at the University of California, Berkeley, designated this study exempt from further review.
↵* Australian Council for Educational Research, Hawthorn, Victoria, Australia.
- Received June 27, 2006.
- Accepted March 1, 2007.
- Physical Therapy