Nationwide health agendas, including the National Arthritis Action Plan (1) and Healthy People 2010 (2), have advocated participation in and adherence to a regular program of exercise as an effective self-management technique for individuals with arthritis. Adherence to aerobic and/or resistance exercise results in numerous health benefits in individuals with arthritis (3–13). Physical health benefits may include improved balance, flexibility, duration of morning stiffness, aerobic capacity, 50-foot walk time, and 6-minute walking distance. Other physical benefits may include improved strength, range of motion, pain, physical disability, and functional status (6–12). Psychological health benefits may include improvements in anxiety, depression, health desirability, and quality of life (11, 13, 14).
However, only a limited number of individuals with arthritis are active at sufficient levels over time to achieve these benefits (15, 16). Individuals with arthritis are less active in vigorous, moderate, and light exercise compared with individuals without arthritis (15). Furthermore, of those individuals with arthritis who begin an exercise program, 45–60% will not maintain the behavior (17, 18). Thus, identification of theory-based correlates of exercise initiation and adherence is critical, with a future goal of examining the efficacy of intervention programs aimed at influencing the correlates (19). According to goal-setting and self-efficacy theories (20, 21), exercise-related goals and self-efficacy may be 2 correlates that influence whether individuals initiate and persist in the performance of health behaviors, including exercise.
Exercise-related goals are defined as what an individual is consciously trying to accomplish, what an individual is working toward, and/or the result being sought (22–26). Goal-setting theory (20) distinguishes between 2 components of goals. First, goal difficulty is the degree of proficiency or level of behavioral performance being sought, such as weekly exercise frequency. Second, goal specificity is the degree of precision with which the aim is specified.
Goal-setting theory (20) contains 2 hypotheses relating to the influence of goal difficulty and specificity on the performance of a behavior such as exercise. First, difficult goals are hypothesized to be more effective than easy goals in positively influencing performance. According to the theory (20), this effect occurs because difficult goals produce greater effort and persistence than do easy goals. Second, the interaction of difficult and specific goals is hypothesized to lead to higher levels of behavioral performance than those associated with vague, do-your-best, or no goals. According to the theory (20), this effect occurs because individuals exert great effort and persistence when striving for high (i.e., difficult) levels of a specific behavioral performance.
Although numerous studies have found positive relationships between difficult goals and behavioral performance as well as between difficult and specific goals and performance in organizational/business domains (27), the generalizability of these findings to the exercise domain is unclear. Only one study to date has examined the relationships between goal difficulty, specificity, the interaction of difficult and specific goals, and exercise adherence. Frahm-Templar and colleagues found that goal specificity and the interaction of goal difficulty and specificity significantly predicted exercise adherence in a sample of regularly active, apparently healthy, undergraduate students (28). In individuals with arthritis, Stenstrom investigated the effects of a 12-week home exercise program that incorporated either a goal-setting condition or a pain attention condition on a variety of outcomes including exercise load (29). At the end of the 12 weeks, 17 participants in the goal-setting group (n = 22) increased their exercise loads, while only 4 participants in the pain attention group (n = 20) increased their exercise loads. Results suggest that goal setting may be a correlate of exercise in individuals with arthritis.
Self-efficacy is defined as beliefs in one's specific abilities to organize and execute the courses of action required to produce a given outcome (21). According to self-efficacy theory, efficacy beliefs influence the amount of time and effort put forth in a given situation, the level of persistence in the face of obstacles, and the level of success achieved from performing a task (21). In the exercise domain, recent reviews suggested that at least 2 specific types of self-efficacy be assessed when the research goal is the prediction of exercise (30, 31).
First, task self-efficacy is defined as beliefs in one's ability to engage in an increasing number of weekly exercise sessions such as confidence to exercise on 1, 2, and 3 days each week (30). Second, scheduling self-efficacy is defined as beliefs in one's ability to regularly complete specific behaviors that help one to exercise regularly (32). Examples include confidence to plan and prepare in advance so one's exercise time is not compromised and confidence to find an exercise time that most suitably fits one's daily lifestyle (32).
Although task self-efficacy has been found to predict exercise, the majority of the research has been conducted with healthy adults (33–36). Preliminary evidence also suggests that task self-efficacy may be a correlate of exercise in populations with chronic disease. For example, task self-efficacy predicted self-reported exercise levels in a sample of chronic heart failure patients (37) and has predicted changes in daily activity in individuals who suffered a myocardial infarction (38). Although no research has examined the relationship between task self-efficacy and exercise in individuals with arthritis, a positive relationship would be expected. That is, task self-efficacy revolves around beliefs in abilities to perform the specific behavior of exercise and thus, should be an influential correlate of behavior (30).
Similar to the task self-efficacy literature, studies examining the relationship between scheduling self-efficacy and exercise have been conducted in predominantly healthy adult samples (30). This research has found scheduling self-efficacy to be a significant predictor of exercise attendance in both novice and experienced exercisers (32, 39, 40). Because scheduling self-efficacy taps individuals' confidence in abilities related to carrying out exercise (i.e., scheduling exercise), this type of confidence would also be expected to be associated with exercise behavior in individuals with arthritis.
The purpose of the study was to examine whether exercise-related goals, task self-efficacy, and scheduling self-efficacy were predictive of 8-week attendance to an aquatics exercise program in individuals with arthritis. Based on contentions from goal setting and self-efficacy theories (20, 21) and findings from past research described earlier, goal difficulty, specificity, the interaction, task self-efficacy, and scheduling self-efficacy were hypothesized to be significant independent predictors of attendance. A secondary purpose was to examine whether high aquatic class attendees differed from low aquatic class attendees on goal difficulty, specificity, the interaction, task and scheduling self-efficacy. Based on contentions from goal setting theory (20) that the setting of difficult and specific goals leads to better performance, high attendees were expected to report the setting of significantly more difficult and specific goals compared to low attendees. Further, based on findings from prior research of significant differences in various types of exercise-related self-efficacy (e.g., coping, attendance) between healthy adult high and low exercise attendees (41, 42), high attendees were hypothesized to have significantly higher task and scheduling self-efficacy compared with low attendees.
PARTICIPANTS AND METHODS
Participants included 216 individuals with arthritis participating in Arthritis Foundation Aquatics Programs for a mean of 2 years and 5 months. Participants reported having arthritis for a mean ± SD of 3.70 ± 1.75 years. One hundred eighty-three participants reported that a doctor diagnosed their arthritis. The most common forms of arthritis reported by participants were osteoarthritis (n = 114), rheumatoid arthritis (n = 31), and fibromyalgia (n = 15). The age of participants ranged from 32 years to 91 years (mean ± SD age 69.21 ± 11.11 years). The majority of participants were women (n = 188), white (n = 203), married (n = 142), and retired (n = 145). The highest level of education attained by participants was 2 ± 3.14 years of college undergraduate education, and the average total household income ranged from $30,000 to $39,999.
A series of demographic variables were obtained including age, sex, ethnicity, marital status, employment status, education, income, type of arthritis, disease duration, and length of prior enrollment in the aquatics class.
This 3-item measure assessed the perceived difficulty of the goals that participants set for aquatic exercise class attendance in the subsequent week, month, and year. A sample item was “The goals I set for the number of times I will attend my aquatic aerobics class in a month are.” Each item was rated on a 0–7-point Likert scale (0 = I don't set goals, 7 = very difficult). For each participant, a mean value for the 3 items was calculated and used in the analyses. This measure had good reliability, factorial validity, and predictive validity in previous research (28) and exhibited good internal consistency in this study (Cronbach's α = 0.92).
This 3-item measure assessed the perceived specificity of the goals that participants set for aquatic exercise class attendance in the subsequent week, month, and year. A sample item was “The goals I set for the number of times I will attend my aquatic aerobics class in a week are.” Each item was rated on a 0–7-point Likert scale (0 = I don't set goals; 7 = very specific). For each participant, a mean value for the 3 items was calculated and used in the analyses. This measure had good reliability, factorial validity, and predictive validity in previous research (28) and exhibited good internal consistency in this study (Cronbach's α = 0.88).
This 3-item measure assessed participants' confidence in their abilities to attend aquatics classes 1 day, 2 days, and 3 days per week for the subsequent 8 weeks. Consistent with other research in this area (30), participants answered each item on a 0–100% Likert scale (0% = not at all confident, 100% = completely confident). For each participant, an overall mean value for the 3 items was calculated and used in the analyses. Adequate internal consistency for this measure was found in the present study (Cronbach's α = 0.69).
This measure assessed participants' confidence in their abilities to perform 8 tasks related to scheduling exercise into their daily routine for the subsequent 8 weeks (32). Sample items included making aquatic exercise sessions high on their priority list of weekly activities and making sure they did not miss a whole week of aquatic classes. Participants answered each item on a 0–100% Likert scale (0% = not at all confident, 100% = completely confident). For each participant, an overall mean value for the 8 items was calculated and used in the analyses. This measure had adequate internal consistency, predictive validity, and concurrent validity in previous research (42, 43) and had good internal consistency in the present study (Cronbach's α = 0.94).
Aquatic exercise attendance.
Attendance at scheduled aquatic classes over the 8-week period following questionnaire administration served as a measure of exercise. Attendance of study participants was recorded by the aquatics instructors. Because participants were registered for 2–5 aquatics class sessions each week, a standardization procedure was conducted. In order to standardize the unequal number of aquatic class sessions, a mean percentage of 8-week attendance was calculated for each participant by first dividing the total number of classes attended by the total number of registered classes for each participant. This value was multiplied by 100 to arrive at an overall mean attendance percentage. This procedure has been used in previously published exercise research (32, 34, 41).
Upon study approval by the Institutional Review Board at Kansas State University, participants were recruited from 14 Arthritis Foundation Aquatics Program sites in a midwestern state. The sites offered the program, which involved range of motion, strengthening, and endurance exercises, for 1 hour on 2–5 days each week. Participants were recruited immediately after a selected class or at a scheduled meeting that took place outside of the normally scheduled class time. Individuals who agreed to participate completed the questionnaire at this time (n = 64) or returned the questionnaire to their instructor within 2 days (n = 152). In the latter case, instructors then mailed the completed questionnaires to the researchers. Finally, instructors recorded aquatic class attendance of study participants for the following 8 weeks.
Statistical package for the social sciences (SPSS) version 10.1 software (Chicago, IL) was used to analyze the data. Before proceeding with the analyses, identification and selection of individuals who had full data on the goal setting, self-efficacy, and aquatic exercise attendance variables were conducted. All other individuals would have been omitted from the SPSS analyses due to listwise deletion. Of the 216 individuals who returned the questionnaire, full data were obtained for 142 individuals. Of this group, 45 individuals completed the questionnaire at the aquatic sites and 97 individuals returned the questionnaire to their instructor within 2 days. A one-way between-groups multivariate analysis of variance revealed no significant differences between individuals who returned the questionnaire immediately and those who returned the questionnaire within 2 days in goal difficulty, specificity, interaction of difficulty and specificity, task self-efficacy, and scheduling self-efficacy (F[5,136] = 1.98, Pillai's trace = 0.07, P < 0.05). As a result, the following analyses included all 142 individuals.
In order to obtain descriptive information on the study variables, means and standard deviations were calculated in SPSS. In order to examine the simple associations between goal setting, self-efficacy, and aquatic exercise attendance, the bivariate correlation function was used in SPSS. To examine whether exercise-related goals and self-efficacy were predictive of aquatic exercise attendance as hypothesized, a multiple hierarchical regression was conducted. Because the correlation and regression analyses involved the main effect goal difficulty and specificity variables and the interaction of these variables, a standardization procedure was conducted according to published guidelines (44, 45). First, the main effect goal difficulty and specificity variables were centered (i.e., standardized) by subtracting the sample mean from the score of each participant for each variable (44, 45). The product of the centered goal difficulty and goal specificity variables for each participant was then calculated and used as the interaction term. The centered goal difficulty, specificity, and interaction were used in the correlation and regression analyses.
Within the regression analysis, the order of entry of predictors was determined based upon theoretic considerations and findings from past research (see ref. 46). First, the length of time that participants were previously exercising in their aquatics classes was entered. Prior research with healthy populations has found past behavior to be a significant predictor of future behavior (32, 47). By controlling for the predictive influence of past behavior, the added and independent variation accounted for by the remaining constructs could be determined (46). The goal-setting variables were then entered, because goal setting is a form of behavioral intention (48) and, thus, most correspondent with the outcome of exercise behavior. Within this group of variables, the centered first-order goal difficulty and specificity variables were entered in the second step (44, 45). The product of the centered first-order variables (i.e., the interaction) was entered in the third step. Task self-efficacy was entered in the fourth step due to the greater correspondence of this measure—relative to scheduling self-efficacy—to the outcome of aquatic exercise attendance (30), followed by the entry of scheduling self-efficacy in the fifth step.
In order to examine study hypotheses regarding differences between high and low attendees, a one-way between-subjects multivariate analysis of variance (MANOVA) was conducted. The independent variable was attendees (high and low) and the dependent variables were the centered goal difficulty, specificity, and interaction, as well as task and scheduling self-efficacy. In order to identify high and low attendees, a tertile split of the aquatic exercise attendance data was conducted. This procedure resulted in individuals being classified as high attendees if they attended 74.28% or more of their aquatics classes (n = 47) or low attendees if they attended 54.55% or less of their classes (n = 48). A between-groups t-test confirmed that the 2 groups significantly differed in aquatic exercise attendance, t(70) = −21.72; P < 0.0001. Thus, analyses of between-group differences could be made with the assurance that truly different groups were examined.
Descriptives and correlations.
The mean ± SD for goal difficulty (2.77 ± 1.37), specificity (5.56 ± 1.61), and the interaction of difficulty and specificity (15.66 ± 8.92) reflected that participants reported the setting of relatively easy yet specific aquatic exercise goals. Participants reported relatively high task self-efficacy (80.05 ± 23.08) and scheduling self-efficacy (84.08 ± 17.95) and attended 63% of their aquatic exercise classes over the 8-week study period. Table 1 contains the bivariate correlations of the study variables. As expected, goal specificity, task self-efficacy, and scheduling self-efficacy were significantly and positively associated with aquatic exercise attendance. Interestingly, goal difficulty was significantly but negatively correlated with attendance. See Table 1 for the remaining associations.
Table 1. Bivariate correlations of study variables in 142 subjects*
|Goal specificity|| || ||−0.43†||0.28†||0.50†||0.22‡|
|Interaction of difficulty and specificity|| || || ||−0.04||−0.15||0.01|
|Task self-efficacy|| || || || ||0.63†||0.33†|
|Scheduling self-efficacy|| || || || || ||0.26†|
|Aquatic exercise attendance|| || || || || ||–|
Tests of study hypotheses.
Hierarchical multiple regression results.
As seen in Table 2, the overall model predicting aquatic exercise attendance was significant (F[6,135] = 5.74, P < 0.0001). After controlling for the significant effect of length of prior aquatic class attendance, goal difficulty, specificity, and task self-efficacy were significant independent predictors of attendance as expected. Contrary to hypotheses, the goal interaction and scheduling self-efficacy were not significant predictors.
Table 2. Prediction of aquatic exercise attendance in 142 subjects*
|Attendance||Overall model||0.17†|| || |
|Step 1||Length of prior attendance||–||0.02||0.16‡|
|Step 2||Goal difficulty||–||0.11†||−0.23†|
| ||Goal specificity||–|| ||0.21‡|
|Step 3||Interaction of difficulty and specificity||–||0.01||0.10|
|Step 4||Task self-efficacy||–||0.06†||0.26†|
|Step 5||Scheduling self-efficacy||–||0.0001||−0.02|
Between-subject differences results.
The overall MANOVA comparing high and low aquatic exercise attendees was significant (F[5,89] = 2.52, Pillai's trace = 0.12, P < 0.05). Subsequent univariate F tests revealed that, as hypothesized, high attendees had significantly higher task self-efficacy (F[1,93] = 8.86, P < 0.01; power = 0.84, η2 = 0.09) and scheduling self-efficacy (F[1,93] = 4.43, P < 0.05; power = 0.55, η2 = 0.05) than low attendees (see Table 3). Contrary to study hypotheses, univariate F tests also revealed that high attendees set significantly less difficult aquatic exercise goals (F[1,93] = 5.88, P < 0.05; power = 0.67, η2 = 0.06) (see Table 3), and the 2 groups did not differ in goal specificity (F[1,93] = 2.22, P > 0.05; power = 0.31, η2 = 0.02) and the goal interaction (F[1,93] = 0.0001, P > 0.05; power = 0.05, η2 = 0.0001) (see Table 3).
Table 3. Mean comparisons for high and low aquatic exercise attendees*
|Goal difficulty||2.48 ± 1.31||3.15 ± 1.41||0.05|
|Goal specificity||5.85 ± 1.68||5.40 ± 1.22||0.14|
|Interaction of difficulty and specificity||14.60 ± 9.17||17.12 ± 9.35||0.99|
|Task self-efficacy||88.51 ± 15.28||75.73 ± 25.26||0.01|
|Scheduling self-efficacy||89.34 ± 13.01||82.21 ± 19.33||0.05|
The unexpected finding of a significant negative correlation between goal difficulty and aquatic exercise attendance was of interest. Recent contentions are that self-efficacy may moderate the goal difficulty–behavior relationship (21, 27). A negative goal difficulty–behavior relationship may occur when individuals have low task self-efficacy, doubt their abilities to exercise, and, thus, experience a reduction in behavior as goal difficulty increases. A positive goal difficulty–behavior relationship may occur when individuals have high self-efficacy and thus exert persistent effort to achieve difficult goals (21, 27).
To examine whether self-efficacy moderated the goal difficulty–aquatic exercise attendance relationship, a tertile split of the task self-efficacy data was conducted such that high task self-efficacy individuals had a mean of 97.77% or higher (n = 47) and lower task self-efficacy individuals had a mean of 70.00% or lower (n = 48). These 2 groups significantly differed in task self-efficacy (t = −21.52, P < 0.0001). Two regression analyses were conducted to determine if the goal difficulty–aquatic exercise attendance relationship differed based upon a high or low categorization of self-efficacy. First, goal difficulty was regressed on attendance with only the lower task self-efficacy group. The model was significant (F[1,46] = 5.99, P < 0.05, R = 0.10; standardized β = −0.34, P < 0.02). Second, goal difficulty was regressed on attendance with only the high task self-efficacy group. The model was not significant (F[1,45] = 2.34, P > 0.05). Results provided support that task self-efficacy moderated the goal difficulty–aquatic exercise attendance relationship, because the relationship was different for low and high task self-efficacy groups.
This study was the first to examine both exercise-related goals and self-efficacy as correlates of aquatic exercise attendance in individuals with arthritis. The findings from the study allow for 2 main conclusions. First, exercise-related goals and specific types of self-efficacy may be important correlates of aquatic exercise attendance. Second, individual differences may exist in goal setting and self-efficacy variables between individuals who attend a high number of aquatic exercise classes versus individuals who attend a lower number of classes.
Correlates of aquatic exercise attendance.
Findings from the multiple hierarchical regression analysis provided partial support for study hypotheses. As expected, goal specificity was a significant predictor of aquatic exercise attendance such that as goal specificity increased, attendance also increased. This positive relationship paralleled findings from prior research conducted with healthy adult populations who performed discrete tasks, such as sit-ups (49–51), or the complex task of exercise (28).
Although goal difficulty predicted aquatic exercise attendance as hypothesized, the negative association between these variables was unexpected (20). This result may have occurred due to the moderation of the goal difficulty–attendance relationship by task self-efficacy. Recall from the exploratory analyses that in the lower task self-efficacy individuals, goal difficulty predicted attendance such that as goal difficulty increased, attendance decreased. Findings suggest that when individuals have low task self-efficacy, the setting of relatively easy exercise-related goals should be encouraged to promote exercise adherence. Although self-efficacy theory (21) would also contend that the setting of difficult goals should motivate individuals with high self-efficacy to adhere to exercise, the exploratory findings did not support this contention.
Task self-efficacy was also found to predict aquatic exercise attendance directly in this study such that as individuals with arthritis became more confident in their abilities to exercise, they attended more aquatics classes. This finding supported study and self-efficacy theory hypotheses (21) as well as prior research examining the relationship between task self-efficacy and exercise in healthy (33–36) and chronic heart failure populations (37).
In contrast to study hypotheses, the interaction of difficult and specific goals and scheduling self-efficacy did not predict aquatic exercise attendance. There may be at least 2 explanations for these findings. First, social cognitions may be influential correlates of behavior among individuals who are developing requisite performance skills and less influential among individuals who are skilled performers of a habitual task (47). Individuals in the present study were experienced aquatic exercisers who were attending their aquatics classes for >2 years. Thus, aquatic exercise may have been a more habitual behavior requiring less social cognitive control to perform. Second, in relation to self-efficacy, the influence of specific types of efficacy on exercise may vary over time (32). For example, scheduling self-efficacy has predicted attendance to structured fitness classes during weeks 9–16 but not during weeks 1–8, of an exercise class program for healthy, novice adult exercisers (32).
Differences between high and low attendees.
Group comparisons revealed that high aquatic class attendees had significantly higher task and scheduling self-efficacy than did low aquatic class attendees. These expected differences paralleled prior research conducted with healthy adult exercisers that has found significant differences in other types of exercise-related self-efficacy (i.e., attendance, coping) between consistent and inconsistent exercisers (41, 42). Contrary to study and goal-setting theory (20) hypotheses, low attendees reported the setting of significantly more difficult goals compared with high attendees. Findings suggest that in individuals with arthritis, high levels of performance (i.e., aquatic exercise attendance) may be facilitated by the setting of moderately difficult goals (i.e., goals that are neither too hard, such as exercise 7 days weekly, nor too easy, such as exercise 1 day per week). Indeed, recent research in the exercise and sport domains with healthy populations has found that the setting of moderately difficult goals was more effective than the setting of difficult goals (27). Furthermore, when individuals with arthritis set difficult exercise goals and carry through on these goals, the symptoms of their disease may exacerbate (e.g., more pain, stiffness) and, thus, individuals may then decrease their exercise frequency.
In contrast to study hypotheses, high and low attendees did not differ in goal specificity. Lack of differences may have occurred for at least one reason. Recall that both high and low attendees reported the setting of relatively specific goals and, according to goal-setting theory, such goals lead to consistency of performance (20). In this study, high and low attendees attended their aquatic exercise classes on a consistent basis—whether that was at a high or a low rate of attendance—over an 8-week period. Thus, goal specificity may have helped this sample of individuals maintain a consistent level of aquatic exercise performance.
The lack of significant differences in the interaction of difficult and specific goals between high and low attendees did not support study hypotheses. Although goal-setting theory (20) posits that higher levels of performance will result from the setting of difficult and specific goals, the type of performance under investigation may moderate such a relationship. Exercise is a complex task associated with a multitude of social psychological processes that operate to influence whether an individual actually performs the behavior (51). As a result, a direct relationship between goal setting and behavioral performance may not necessarily occur.
At least 2 study limitations should be considered. First, the findings pertain to a select group of individuals with arthritis who volunteered for the study, had arthritis for a relatively short duration, and had been exercising in their aquatics classes for >2 years. Thus, the findings may not generalize to individuals who have arthritis for longer periods of time, are attempting to initiate aquatic exercise, and/or are attempting to initiate or adhere to other types of exercise. Second, because the study did not involve experimental manipulation of the goal-setting and self-efficacy constructs, the direction of the relationship between these variables with aquatic exercise attendance can only be assumed. For example, as individuals with arthritis attend more aquatics classes, their task self-efficacy may increase as opposed to the interpretation provided in the preceding section that as task self-efficacy increased, attendance increased.
Future research should prospectively examine the relationship between goal setting, self-efficacy, and aquatic exercise attendance as individuals with arthritis progress from initiation to adherence. Information will then be obtained on whether the relationships observed in the current study are stable or dynamic over time. The influence of additional individual-level correlates, such as the actual number of exercise sessions that individuals set goals for attending, efficacy to cope with daily negative thoughts, and exercise enjoyment, and environmental-level correlates, such as exercise group cohesion, on exercise should be examined. Understanding the full scope of correlates would aid in the design of interventions that target the initiation of and adherence to exercise in individuals with arthritis. Finally, future research should examine correlates of other types of exercise, such as walking and resistance training, which have been found to be effective self-management techniques for individuals with arthritis.