There is a growing awareness in the geriatric community that expectations and motives of older adults are important to the process of physical disablement (16). For example, research has shown that self-efficacy expectations are important in understanding falls (22) and the decline of basic functional skills such as ambulation and transfer (18). However, in the context of social cognitive theory, the impact of self-efficacy expectations on behavior is clearly linked to the incentive value of behavior. Unfortunately, there are no existing measures of older adults' motives to maintain their physical function.
The central role of motivational concepts in human behavior has a long history in the field of psychology (19,20). For example, they are represented in Rotter's Social Learning Theory of Personality (19) as “reinforcement values” and in Bandura's (2) social cognitive theory as “incentives.” Consistent with these theories is the realization that older adults will neither attempt nor persist with attempts to perform a valued behavior if they lack confidence in their ability (12). However, if a person has low self-efficacy for a functional task, yet has a strong desire to perform it, then this conflict is likely to lead to adjustment disorders, such as anxiety and/or depression. Hence, the assessment of motives is essential to a comprehensive understanding of disablement and the sequel of disabling conditions.
With this in mind, the objective of the current research was to develop a scale that would enable researchers to assess older adults' desire for physical competence (DPC). The conceptual definition of this new construct is as follows: DPC reflects older adults' motivation to possess the ability to perform tasks that require different elements and levels of physical functioning.
The primary goal of this investigation was to develop and to provide initial validation of a measure that would enable investigators to assess older adults' desire to acquire or maintain their physical competence. The present investigation involved two distinct phases: item development and psychometric evaluation of the measure. All methods and procedures were approved by a federally registered institution review board for human subject research.
Phase 1: Item Development
In developing items for the DPC scale, we first reviewed several measures designed to assess the perceived difficulty of performing various tasks that represent either basic or advanced components of physical function created specifically for older adults (9,11). This procedure led to the creation of an initial pool of 46 items that tapped the universe of content for this construct.
A team of three investigators with expertise in behavioral health, fitness, and gerontology then examined the initial item pool and eliminated items based on the following criteria: a) there was ambiguity in the physical demand or redundancy with other items, b) the task was gender specific, c) the task appeared to be dependent on external influence, or d) the task relied heavily on dexterity (e.g., buttoning a shirt). This process yielded a subsequent pool of 25 items. These 25 items were then reworded for specificity and clarity.
As recommended by Nunnally (13), a five-point Likert scale was used to rate the desire to be able to perform each task employing the following verbal anchors: 0) no desire whatsoever, 1) low desire, 2) moderate desire, 3) strong desire, 4) very strong desire. Instructions for completion of the measure read as follows: “Place an X in the box that best describes your current desire to be able to perform each task. It is very important to remember that we are not interested in whether you can do the task or not; rather, we are interested in your level of desire to possess the physical ability that would enable you to do each task.”
Phase 2: Psychometric Evaluation of the DPC Measure
An alpha factor analysis with a varimax rotation was used to explore the factor structure of the original item pool. Subsequently, item response theory was used to establish the appropriate scale for measuring DPC. Once scores could be computed, correlational analyses were conducted to examine both convergent and discriminant validity. Subsequently, a “known groups” construct validation procedure was employed by hypothesizing that older adults with arthritis would have lower DPC sores than those older adults without arthritis. This hypothesis was based on the premise that one way older individuals may cope with functional decline is to lower their expectations and desire to pursue activities that are perceived to be unrealistic (7,17). In a final analysis, the DPC was given to a group of older adults (N = 30) on two different occasions separated by a 2-wk interval to evaluate test-retest reliability of the DPC.
To examine the factor structure and construct validity of the measure, we recruited 157 participants (53 men, 104 women) 60 yr of age and older with a range of functional abilities employing an age-stratified sampling procedure; that is, we recruited both older men and women across eight age groups: 60–64, 65–69, 70–74, 75–79, 80–84, 85–89, 90–94, 95+. In addition, another 30 older adults were recruited to evaluate the test-retest reliability of the measure. Both of these groups were community samples recruited through senior centers and local health screening events for older adults. To qualify for the study, individuals had to first agree to read and sign an informed consent that had been approved by our institutional review board. There was a 100% compliance with this first step. Table 1 provides selected demographic information on each sample. The mean age (±SD) of participants in the measure development phase was 73.28 (±8.0) yr with a range of 60–95 yr. The sample population was moderately heterogeneous relative to age and reported a variety of ethnic and socioeconomic backgrounds. These individuals also had an assortment of chronic diseases.
In addition to the DPC, all participants completed a demographic survey that included age, gender, race, education level, income level, marital status, living status, and the presence/absence of six chronic diseases. Participants also completed several questionnaires that assessed health status and a variety of psychosocial constructs, which were used to examine the construct validity of the DPC scale.
The SF-36 is a generic measure of health status that has two norm-based composite scales that range from 0 to 100: mental health and physical health (24). Higher scores on these scales indicate more favorable levels of function. For this study, we limited our attention to the physical health scale of this measure.
The short-form 13-item social desirability scale (SDS) (5) assesses participants' need for social approval. Our intent was to insure that the measure of desire for physical competence was not confounded by this construct. The SDS has been used extensively in instrument validation studies and possesses excellent psychometric properties (5).
Desire for control.
The desire for control scale was developed by Burger (4) and assesses individuals' personal control over several aspects of their lives. For the purpose of the current study, we used a 10-item modified version of the scale. Participants were asked to read each question and mark the box next to the appropriate answer; response options included a 4-point scale ranging from 0 to 3: never, sometimes, often, or always. The internal consistency of this instrument in the present study was acceptable (α = 0.66).
The body satisfaction measure consists of two subscales: physical function and body appearance. For this study, the body appearance questions (three items) were eliminated, because the focus of the study was on physical function. All remaining items were rated on a 7-point scale that was scored from −3 to +3 with numbers on the scale anchored by the following phrases: very dissatisfied (−3), somewhat dissatisfied (−2), a little dissatisfied (−1), neither (0), a little satisfied (+1), somewhat satisfied (+2), and very satisfied (+3) (15). Participants were asked to read each statement and respond by circling the answer that best describes their level of satisfaction with each of the items. The internal consistency of this measure was excellent (α > 0.90).
The Satisfaction with Life Scale measures global life satisfaction of an individual (6). Each of five items is rated on a scale that ranges from 1 (strongly disagree) to 7 (strongly agree). The scores for the five items are summed with a possible range of scores on this questionnaire from 5 (low satisfaction) to 35 (high satisfaction). The internal consistency of this measure was excellent (α > 0.90).
Following completion of the informed consent form, the questionnaire packets were then given to participants to complete and included the DPC and the six questionnaires described above. As a small token of appreciation, study participants had their names entered into a lottery for gift certificates to a restaurant.
As mentioned previously, the dimensional structure of the DPC scale was evaluated through exploratory factor analysis. Pearson correlations were used to evaluate convergent and discriminant validity, whereas a general linear model was employed to evaluate the known groups hypothesis concerning older adults with or without arthritis.
Results of the psychometric analyses.
The means (±SD) for the individual items of the DPC can be found in Table 2. All the SD for the items are close to being 1 or >1, a desirable feature of items for any psychometric instrument (13). Additionally, nine items were eliminated from the scale because they either had low factor loadings (<0.30) or specific items loaded on more than one factor. Thus, the final DPC scale consisted of 16 items. Inspection of the eigenvalues indicated the presence of two factors, which explained 70.67% of the variance in the items. Because these two factors reflected the underlying concept of item difficulty related to physical function (see final two columns of Table 2); that is, either low or high physical demand, item response theory (IRT) was used in the analysis of the data as opposed to using classic test theory (CTT) and the aforementioned two-factor solution. Using an IRT model, the items were mapped along a single component structure of item difficulty. The major advantages of modeling the data in this manner are that test scores remain independent of the DPC items, and item statistics are independent of desire for physical competence that characterize the sample used in item calibration (10). In the IRT literature, these properties are termed item parameter invariance and ability parameter invariance, respectively.
Rating scale results.
A Rasch rating scale model was applied to the DPC data (1,3). In the Rasch model, the item and the person parameters can be simultaneously displayed on the same scale or map. The map is a visualization of the parameters and can be used to locate the position of a specific item or person. The map found in Figure 1 illustrates that the data are characterized by a relatively normal distribution of scores. The asterisks shown below the x axis represent individual items on the DPC. Whereas the lower end of the construct seems to be adequately covered by the measure, 43 of the 157 participants are located above the most difficult item; however, not all participants rated strong desire to be able to perform this task. This pattern in the data is not unexpected because the measure was designed to capture desire for physical competence in community-dwelling older adults that may be targeted for physical activity interventions due to a less than optimal level of fitness.
The table of item difficulty parameters (Table 3) illustrates that the DPC has a rather diverse range of items in terms of their difficulty. Item parameters are expressed in logits, and a value of 0.0 is set as the average of all the estimates of item difficulty. The category threshold parameters (Table 4) show that movement along the item response scale is not uniform. Specifically, moving from a rating of strong to very strong desire contributes more to a positive total score that moving from low desire to moderate desire. Similarly, going down from low to no desire has a greater negative effect on the total score than stepping down from moderate to low desire.
Overall the items fit the model well as is evident from the infit and outfit statistics that fall within the range of measurement error. The average of the outfit mean square error was 1.02, and for the infit mean square error, it was 1.04, where values between 0.5 and 1.5 are considered productive for measurement (see Winstep manual section on diagnosing misfit (25)). On the other hand, the person fit statistic reveals that a few older adults exhibited deviant item responses. For example, one person almost always responded at the highest level of desire for all items except to the item, “do light work around the house.” This pattern of response suggests that the person in question may have been confused about the instructions for completing the measure; that is, he or she was responding to his or her desire to perform a specific household task that he or she had in mind as opposed to having the ability to perform various tasks around the home that require light work.
Table 5 provides descriptive statistics on the DPC for various subgroups. As can be seen from these data, DPC scores did not differ as a function of gender, race, or education, although there was a trend for women to have lower DPC scores than men, for blacks to have lower DPC scores than whites, and for high school graduates to have lower DPC scores than those who had education beyond high school. However, older adults who lived alone had lower DPC scores than those who were married, and individuals in the lowest income bracket (<$15,000 annually) had lower scores than any other income category.
Scores on the DPC were examined for their relationships with age, desire for control, body satisfaction, the SF-36 physical composite scale (PCS), life satisfaction, and social desirability. It was hypothesized that the DPC would be inversely correlated with age and directly correlated with desire for control, body satisfaction, the SF-36 PCS, and life satisfaction. No correlation was expected between the DPC and the social desirability scale. Results were generally consistent with these predictions in that the DPC had a correlation of r = −0.33 with age and r values of 0.23, 0.22, 0.25 (all P values <0.01) with desire for control, body satisfaction, and PCS scores on the SF-36, respectively. The DPC was unrelated to overall life satisfaction; however, life satisfaction was related to body satisfaction (r = 0.46), desire for control (r = 0.20), social desirability (r = 0.19), and the SF-36 PCS score (r = 0.24).
As a final test of construct validity, we compared DPC scores for those who had arthritis (N = 70) with those who were free of this chronic condition (N = 87). A general linear model using age as a covariate revealed a significant effect for arthritis, F(1,153) = 6.11, P = 0.15, with those having arthritis scoring lower on the DPC, mean (SD) = 0.96 (1.88) than those without arthritis, mean (SD) = 1.58 (1.68).
A group of 30 older adults was recruited from a cardiac rehabilitation program and senior activity groups in local communities to examine the test-retest reliability of the DPC. As shown in Table 1, this sample was comparable with participants who were involved in the validation phase of the study. A Pearson product moment correlation was calculated between the DPC scores that were collected 2 wk apart. The reliability of the DPC was supported by the strong relationship between scores across the 2-wk interval of time (r = 0.93) and the fact that the group means were essentially identical for the two assessments.
The DPC is a 16-item questionnaire that builds on available measurement tools that can be employed to examine the role of social cognitive theory (2) on the process of physical disablement (23). The DPC seems to have adequate convergent and discriminant validity that would permit its use in future research. Specifically, individuals with higher DPC scores had a higher general desire for control (4), were more satisfied with their physical function (14,15), and had more favorable perceptions of their physical health (24) than those with lower DPC scores. In a known group analysis, older adults with arthritis were found to have lower DPC scores than those who were free of this chronic medical condition.
It is also interesting to note that older adults who either lived alone or were in the lowest income bracket had lower DPC scores than those who were married or were in higher income brackets, respectively. People who are socially isolated participate in fewer physical activities than those who have a strong social support structure (21), and there is a well-known social gradient in physical function among older adults (8). Surprisingly, there was no evidence of a gender effect on the DPC, suggesting that older women may have a desire to be physically competent at a level that is comparable to men. Perhaps differences in physical activity levels typically observed between men and women are a function of differences in other constructs within the motivational process, such as self-efficacy beliefs (2). For example, there is good evidence from research on older adults that men have stronger self-efficacy beliefs for physical functioning than women (12)
Future Research and Limitations
A number of interesting research questions arise that might be pursued in future investigations with the DPC scale. Investigators might explore links between desire for physical competence and other social cognitive constructs. For example, the relationship between self-efficacy beliefs and involvement in physical activity programs on the part of older adults may be better understood by considering differences in desire for physical competence. Specifically, sedentary behavior might be due to either low self-efficacy or low desire for physical competence. Another interesting area of research is to pursue the development and testing of strategies to enhance desire for physical competence among older adults. A related hypothesis is that scores on the DPC reflect a state of readiness for physical activity.
We recognize that the current study is the first step in validation of the DPC and that the current study design has limitations. For example, there is value in examining the predictive validity of the DPC in longitudinal studies of disability and conducting experimental research into the behavior of populations who score high and low on the DPC. A clear benefit of the DPC, however, is that it provides investigators with a measure of the relative value of physical competencies that can be related to other constructs in social cognitive theory that, when taken together, provide a more complete model of social-cognitive influences for studying physical disability in older adults.
1. Andrich, D. A rating formulation for ordered response categories. Psychometrika
2. Bandura, A. Social Foundations of Thought and Action: A Social Cognitive Theory.
Englewood Cliffs, NJ: Prentice-Hall, 1986.
3. Bond, T. G. F. C. M. Applying the Rasch Model.
Mahwah, NJ: Lawrence Erlbaum Associates, 2001.
4. Burger, J. M. Desire for control, locus of control, and proneness to depression. J. Pers.
5. Crowne, D. P. A new scale of social desirability independent of psychopathology. J. Consult. Psychol.
6. Diener, E., R. A. Emmons, R. J. Larsen, and S. Griffin. The Satisfaction with Life Scale. J. Pers. Assess.
7. Folkman. S. Personal control and stress and coping processes: a theoretical analysis. J. Pers. Soc. Psychol.
8. Ford, E., S. R. K. Merritt, G. W. Heath, et al. Physical activity behaviors in lower and higher socioeconomic status populations. Am. J. Epidemiol.
9. Haley, S. M,. A. M, Jette, W. J. Coster, et al. Late-life function and disability instrument: development and evaluation of the function component II. J. Gerontol. A Biol. Sci. Med. Sci.
10. Hambleton, R. K. Emergence of item response modeling in instrument development and data analysis. Med. Care
11. Jette, A. M. S. M,. Haley, W. J. Coster, et al. Late life function and disability instrument: I. Development and evaluation of the disability component. J. Gerontol. A Biol. Sci. Med. Sci.
12. McAuley, E., and D. Rudolph. Physical activity, aging, and psychological well-being. JAPA
13. Nunnally, J. C. Psychometric Theory.
New York: McGraw-Hill, 1967, pp. 104–558.
14. Ray, K. M., L. L. Hector, L. K. Lynes, A. L. Stewart, P. L. Painter, and S. L. Dibble. Assessment of satisfaction with physical fitness in kidney transplant recipients. Med. Sci. Sports Exerc.
15. Reboussin, B. A., W. I. Rejeski, and K. A. Martin. Correlates of satisfaction with body function and body appearance in middle- and older aged adults: The Activity Counseling Trial (ACT). Psychol. Health
16. Rejeski, W. J., and B. C. Focht. Aging and physical disability: on integrating group and individual counseling with the promotion of physical activity. Exerc. Sport Sci. Rev.
17. Rejeski, W. J., K. A. Martin, M. E. Miller, W. H. Ettinger, and S. Rapp. Perceived importance and satisfaction with physical function in patients with knee osteoarthritis. Ann. Behav. Med.
18. Rejeski, W. J., M. E. Miller, C. G. Foy, S. P. Messier, and S. Rapp. Self-efficacy and the progression of functional limitations and self-reported disability in older adults with knee pain. J. Gerontol. B Psychol. Sci. Soc. Sci.
19. Rotter, J. B. Social Learning Theory and Clinical Psychology.
Englewood Cliffs, NJ: Prentice-Hall, 1954, pp. 3–75.
20. Rotter, J. B., J. E. Chance, and E. J. Phares. Applications of a Social Learning Theory of Personality.
Oxford: Holt, Rinehart, and Winston, 1972, pp. 1–43.
21. Satariano, W. A., T. J. Haight, and I. B. Tager. Living arrangements and participation in leisure-time physical activities in an older population. J. Aging Health
22. Tinetti, M. E., and L. Powell. Fear of falling and low self-efficacy: a cause of dependence in elderly persons. J. Gerontol. A Biol. Sci. Med. Sci.
23. Verbrugge, L. M., and A. M. Jette. The disablement process. Soc. Sci. Med.
24. Ware, J. E., M. Kosinski, and S. K. Keller. SF-36 Physical and Mental Health Summary Scales: A User's Manual.
Boston: The Health Institute, 1994, pp. 6:1–6:22.