Original Research Journal of Aging and Physical Activity, 2010, 18, 375-389 © 2010 Human Kinetics, Inc.

Correlates of Physical Activity Participation in Community-Dwelling Older Adults Christy Haley and Ross Andel The authors examined factors related to participation in walking, gardening or yard work, and sports or exercise in 686 community-dwelling adults 60–95 years of age from Wave IV of the population-based Americans’ Changing Lives Study. Logistic regression revealed that male gender, being married, and better functional health were associated with greater likelihood of participating in gardening or yard work (p < .05). Male gender, better functional health, and lower body-mass index were independently associated with greater likelihood of walking (p < .05). Increasing age, male gender, higher education, and better functional health were associated with greater likelihood of participating in sports or exercise (p < .05). Subsequent analyses yielded an interaction of functional health by gender in sport or exercise participation (p = .06), suggesting a greater association between functional health and participation in men. Gender and functional health appear to be particularly important for physical activity participation, which may be useful in guiding future research. Attention to different subgroups may be needed to promote participation in specific activities. Keywords: walking, gardening, sports, exercise promotion

The positive role of physical activity in promoting functional health, delaying or preventing the onset of disease and disability, and reducing mortality has been well established (e.g., Ferrucci et al., 1999; Hubert, Bloch, Oehlert, & Fries, 2002; Jonker et al., 2006; Leveille, Guralnik, Ferrucci, & Langlois, 1999). Cardiovascular, as well as all-cause, mortality risk appears to be lessened by physical activity participation even after other relevant risk factors are accounted for (Nocon et al., 2008). Maintenance or improvements in musculoskeletal function resulting in functional mobility, psychological well-being, and overall quality of life have also been noted (Warburton, Nicol, & Bredin, 2006). Despite the known benefits of exercise, physical activity participation rates are low among all age groups. Among adults over the age of 50, it is estimated that as many as 37–79% do not engage in regular physical activity (Jerome et al., 2006; unpublished paper prepared for the Robert Wood Johnson Foundation, 2000). As the proportion of older adults continues to increase, the potential impact on

The authors are with the School of Aging Studies, University of South Florida, Tampa, FL. 375

376  Haley and Andel

life expectancy and active life expectancy through physical activity promotion is significant. Given the low rates of physical activity and the subsequent disconnect between potential and recognized physical activity benefits, increasing participation in such activity has become a Healthy People 2010 national health objective for all age groups, including older adults (U.S. Department of Health and Human Services, 2000). The health-promotion model proposed by Pender (1987, 1996) posits that individual biological, sociocultural, and psychological characteristics influence subsequent health-promoting behaviors such as engagement in physical activity. Identifying associations between individual factors and physical activity involvement at the population level, therefore, is a useful first step in guiding research aimed at designing and testing targeted interventions to promote physical activity among older adults. Several such characteristics have been associated with physical activity participation in older adults.

Sociodemographic Factors and Physical Activity It is still not clear whether increasing age is associated with physical activity participation. Resnick, Palmer, Jenkins, and Spellbring (2000) observed no significant differences in the ages of those who self-reported regular physical activity (verified through attendance records) among residents in a continuing-care retirement community. Conversely, several studies have found that physical activity decreases with age. In a cross-sectional survey of Mexican Americans (Mouton, Calmbach, Dhanda, Espino, & Hazuda, 2000), younger participants were significantly more physically active than older participants. Weiss, O’Loughlin, Platt, and Paradis (2007) identified increasing age as a predictor of physical activity decline among adults in low-income communities, although the sample (age range 18–65 years) did not include oldest old adults. Older age was also associated with lower selfreported physical activity (Riebe et al., 2005), and type, intensity, frequency, and duration of activity were all factored to estimate weekly energy expenditure. In a study of older adults’ leisure habits, a significant decline in physical activity was observed between the ages of 70 and 80 (Janke, Davey, & Kleiber, 2006). Until the eighth decade, however, involvement in physical activity was relatively stable and was actually shown to increase as a function of retirement status. Similarly, Salmon, Owen, Crawford, Bauman, and Sallis (2003) found that participants over 60 spent more time walking, performing moderately intense activity, and engaging in more overall physical activity than younger participants. Other sociodemographic characteristics, namely, gender, education, and race/ethnicity, have been associated with physical activity participation. Previous research has consistently suggested that physical activity participation is lower among older women (e.g., Janke et al., 2006; Weiss et al., 2007) and less educated older adults (Droomers, Schrijvers, & Mackenbach, 2001; Janke et al., 2006; Mouton et al., 2000; Weiss et al., 2007). Lower physical activity rates have also been found in older individuals from several minority racial/ethnic groups. Mexican Americans age 60 years and older reported lower physical activity levels than European Americans of the same age, despite greater perceived benefits and fewer perceived barriers among the Mexican American group (Mouton et al., 2000).

Correlates of Three Physical Activity Types   377

Higher rates of physical inactivity among Mexican American and African American than among White men and women of all ages were reported by Crespo, Smit, Andersen, Carter-Pokras, and Ainsworth (2000). Janke et al. found lower rates of physical activity among older African American adults than other racial groups. Notably, although other races showed a slight decline in physical activity with age, African American individuals significantly increased physical activity with age. Overall rates were still lower than for other racial groups, however. The relationship between physical activity and marital status is less clear. Higher levels of physical activity for older married individuals have been observed (Janke et al., 2006; Pettee et al., 2006), and active men were much more likely to have an active spouse in Pettee et al.’s study. Crespo et al. (2000) found that currently or formerly married men were more likely to be more physically active than never-married men, whereas physical activity participation did not seem to vary by marital status for women.

Physical and Mental Health Indicators of poor mental and physical health have been associated with reduced physical activity participation. High body-mass index has been linked with reduced physical activity (Jancey et al., 2007; Mouton et al., 2000; Weiss et al., 2007). More functional limitations (Janke et al., 2006), low self-rated health (Droomers et al., 2001; Flegal, Kishiyama, Zajdel, Haas, & Oken, 2007; Resnick et al., 2000; Weiss et al., 2007), and more chronic conditions (Mouton et al.) have also been shown to predict lower physical activity. In addition, increased depressive symptoms have been associated with lower or declining levels of physical activity in older adults (Sizer Fitts et al., 2008; Janke et al., 2006; Mouton et al., 2000).

Purpose and Hypotheses Using the Health Promotion Model (Pender 1987, 1996), we examined sociodemographic factors and physical and mental health indicators in relation to three categories of frequency of physical activity participation—namely, walking, gardening or yard work, and active sports or exercise—among a representative sample of community-dwelling adults age 60 and older. Numerous studies have examined the association between participant characteristics and physical activity of participants; however, most did not include multiple outcome measures and the variety of independent variables that was available to us. Although assessment of the intensity and duration of activity was beyond the scope of the current study, two of the three activity types, gardening and walking, have been cited as the most popular among older adults (Centers for Disease Control and Prevention, 1999). We expected to find that those more likely to engage in physical activity would be younger, men, and married and have a higher level of education and better physical and mental health. However, we also hypothesized that the correlates of physical activity participation would vary based on the type of activity. Finally, we expected that gender, often cited as an important factor in physical activity participation, would modify the significant associations between other independent variables and physical activity.

378  Haley and Andel

Methods Sample The data used in this study came from a subset of Wave IV of the Americans’ Changing Lives Study, conducted in 2002 (House, 2006), which is an ongoing panel study of the lives of adults living in the continental United States. It is a multistage stratified area probability sample of people 25 years of age and older, with an oversampling of African Americans and individuals 60 years of age and older during Wave I. Data collection was done mostly by telephone or face-to-face interview with participants (when possible) or proxy respondents. A complete description of the study and methods has been published elsewhere (House et al., 1994; Lantz et al., 2001). A total of 3,617 respondents were included in Wave I. Of those, 1,184 were dead by the end of Wave IV. Wave IV nonrespondents totaled 646. Self-reporting interview respondents at Wave IV totaled 1,692, resulting in a response rate of 73% among the original sample still living by the end of Wave IV. The current analyses were limited to community-dwelling individuals age 60 years and older (N = 686). Data from individuals living in nursing homes, assistedliving facilities, and retirement communities (n = 26) and proxy respondents (n = 60) were excluded.

Outcome Measures The outcome measures were self-reported participation in three categories of physical activity. Participants were asked “How often do you typically work in the garden or yard?” “How often do you take walks?” and “Other than taking walks, how often do you engage in active sports or exercise?” Answer options for all three questions were often, sometimes, rarely, or never. There was an uneven distribution of responses across the three activity types. Therefore, physical activity status was dichotomized, with participants considered active if they answered often and otherwise not active in the analyses.

Independent Measures Sociodemographic Characteristics.  Sociodemographic characteristics included age, gender (female = 1; male = 0), race (White = 1; self-identified as any race other than White = 0), level of education (0–12 years vs. 13 or more years), and marital status (married or unmarried—including divorced, separated, marriage annulment, and never married, or widowed). Physical Health Indicators.  Self-reported activity limitation because of health

was assessed by asking participants, “How much are your daily activities limited by your health or health problems?” Answer choices were a great deal, quite a bit, some, a little, or none at all. Self-reported activity limitation because of health was then dichotomized, with those answering none at all considered to have no limitations (0) and all others considered to have limitations (1). Specifically, the activity-limitations variable was used to identify individuals who may have difficulty with certain activities but scored high on the Functional Health Index because of the specificity of the activity-limitation questions it contained.

Correlates of Three Physical Activity Types   379

The Functional Health Index, previously used in several studies (Janke et al., 2006; Kahng, 2008), was created from individual variables measuring bed- or chairbound status, self-bathing difficulties, difficulty climbing stairs, difficulty walking several blocks, and difficulty doing heavy housework. The index was scaled such that respondents who reported being functionally impaired on bed or bath were not asked about the next lower levels of impairment, and so on. Those who reported being in bed or a chair most or all of the day because of health or having “a lot” of difficulty or “cannot” bathe self were categorized as most functionally impaired and assigned an index value of 1. A value of 2 was assigned to participants with moderate functional impairment, who reported having “a lot” of difficulty or “cannot” climb a few flights of stairs or walk several blocks. Participants with some functional impairment who reported having “a lot” of difficulty or “cannot” do heavy work around the house were assigned an index value of 3. A value of 4 was given to the least functionally impaired. These participants reported having a little, some, or no difficulty doing heavy work around the house. Values were reverse coded for our analyses, with a score of 4 indicating most functional impairment and little or no reported functional impairment indicated by a score of 1. Body-Mass Index.  Body-mass index (BMI) was calculated by dividing self-

reported body weight in Wave IV (kilograms) by self-reported height from Wave I (in meters; height was not reported in subsequent waves) and then squaring. The resulting scores were divided into four categories based on the World Health Organization’s (2006) classifications of underweight (29.99). Quadratic BMI was also included in the analyses as a continuous variable.

Mental Health Indicators.  Depressive symptoms were measured with the 11-item

version of the Centers for Epidemiological Studies Depression Scale (CESD-11), composed of items assessing the extent of depressive symptoms (α = .83; Kohout, Berkman, Evans, & Cornoni-Huntley, 1993; Radloff, 1977). The specific measure was a standardized CESD-11 index that takes the mean across the 11 input items (see Kohout et al., 1993 and Radloff, 1977). All items were standardized using the Wave I weighted means and standard deviations before being combined. The index was restandardized after the items were combined using the Wave I weighted index mean and standard deviation. Cases with missing data equaling less than half the values were imputed with means of the nonmissing items. Cases missing data on more than 50% of input items were imputed using an OLS regression technique. This measure has been used in previous studies (e.g., Janke et al., 2006; Kahng, 2008). The index score was dichotomized such that participants with standardized scores ≥0 (the same or improved CES-D scores from Wave I to Wave IV; reference group) were compared with those with standardized scores below 0 (worse scores in Wave IV compared with Wave I).

Data Analyses Data were analyzed using the Statistical Package for the Social Sciences (SPSS) version 16 (SPSS Inc., Chicago, IL). Bivariate correlations were performed to examine interrelation among the independent variables. Then, binary logisticregression analyses for each physical activity outcome were performed sequentially in three blocks. The first block consisted of sociodemographic variables representing

380  Haley and Andel

sociocultural factors referenced in the Health Promotion Model. The second and third blocks added physical and mental health indicators, respectively, representing biological and psychological factors. Block 1 variables entered were age, gender, race, marital status, and education level. Block 2 variables were activities limited by health, functional health, and BMI. The Block 3 variable was the CES-D measure. To test for multicollinearity, we examined the variance inflation factor (VIF), with VIF > 2.5 as the criterion, as suggested by Allison (1999).

Results Sample Characteristics The sample characteristics are summarized in Table 1. The participants were mostly women and White and had 12 years or less of education. Married and widowed respondents each accounted for just under half the respondents, with the remaining unmarried. Age ranged from 60 to 95, with a mean of 73 years (SD = 8.42). Over half reported having at least some difficulty with daily activities because of health. However, functional health as measured with questions regarding specific activity limitations was high, with over half the sample having few if any limitations. Most respondents were normal or overweight according to the World Health Organization’s BMI classification system, with each representing just over one third of the sample. Nearly a quarter of participants were classified as obese. Most of the sample had more depressive symptoms than at Wave I. Over half the respondents were categorized as inactive in all three categories of physical activity. More respondents reported being active in walking than in gardening or active sports or exercise.

Bivariate Correlations and Multicollinearity Diagnostics Although some significant correlations were found, they did not reach the magnitude that would indicate multicollinearity. The strongest association was between Functional Health Index and having any activity limitations because of health (r = .40, p < .01). Collinearity diagnostic analysis revealed no VIF greater than 1.3, suggesting that multicollinearity was unlikely to influence the results.

Binary Logistic Regression Binary logistic-regression results for activity in gardening or yard work are presented in Table 2. Female gender, being unmarried, and additional functional limitations were all independently associated with about 50% lower odds of participation in gardening and yard activity. Table 3 shows the results for binary logistic-regression analyses for participation in walking. Women had lower odds of participation in walking, as did those with more functional limitations and higher BMI. The results for association between independent variables and participation in sports and exercise are presented in Table 4. Older participants, men, and those with more education and fewer functional limitations were more likely to participate in sports or exercise.

Table 1  Americans’ Changing Lives Wave 4— Community-Dwelling Adults Age 60 and Older Variable Predictor Variables  Age, M ± SD  Gender    male    female  Race    White    non-White   Marital status    married    unmarrieda    widowed  Education    0–12 years    13–17 years   Daily activities limited by health    any    not at all   Functional Health Indexb    least impaired    somewhat impaired    moderately impaired    most impaired   Body-mass indexc    underweight    normal range    overweight    obese   Depressive symptomsd    ≤Wave I    >Wave I Outcome Variables   Active in yard work    not active    active   Active in walking    not active    active   Active in sports or exercise    not active    active

73.0 ± 8.42 30.0 70.0 76.5 23.5 44.3 15.2 40.5 65.3 34.7 67.6 32.4 62.7 14.0 11.5 11.8 1.6 37.8 37.3 23.2 30.8 69.2 65.5 34.5 57.2 42.8 70.9 29.1

Note. Values are percentages except where noted otherwise. Because of missing data, sample range is 663–686. a Divorced, separated, marriage annulled, or unmarried. bSelf-reported, a sum of difficulties with bed or bathing, stair climbing, walking, and housework; 1 = least functional impairment, 4 = most functional impairment. cCalculated by body weight (kg) × height (m),2 then classified as underweight (29.99). dMeasured using 11-item CES-D and then standardized based on Wave I index scores; dichotomized into same or fewer versus more depressive symptoms than Wave I.

381

382

0.99

0.97

0–12 years education

Activity limitations (any)

e

0.63–1.34 0.44–0.67 0.91–1.42

0.92 1.13

0.82–1.68

0.32–0.94

0.66–1.5

0.90–2.12

0.37–0.81

0.98–1.02

95% CI

0.54***

1.17

0.55*

1.00

1.38

0.55**

1.00

OR

Model 2

1.15

1.13

0.55***

0.95

1.19

0.55*

1.02

1.36

0.56**

1.00

OR

0.77–1.72

0.90–1.40

0.45–0.69

0.66–1.38

0.83–1.70

0.32–0.94

0.68–1.53

0.89–2.09

0.38–0.83

0.98–1.02

95% CI

Model 3

*p < .05. **p < .01. ***p < .001.

a Divorced, separated, marriage annulled, or unmarried. bSelf-reported, dichotomized into any or none. cSelf-reported, a sum of difficulties with bed/bathing, stair climbing, walking, and housework; 1 = least functional impairment, 4 = most functional impairment. dCalculated by body weight (kg) × height (m)2 then classified as underweight (29.99). eMeasured using 11-item CES-D and then standardized based on Wave I index scores; dichotomized into same or fewer versus more depressive symptoms than Wave I.

Note. OR = odds ratio; CI = confidence interval.

Depressive symptoms (≤Wave I)

Body-mass indexd

Functional limitationsc

b

0.69–1.37

0.30–0.84

0.50**

a

Unmarried

0.60–1.30

0.88

0.93–2.12

1.40

Married

0.35–0.75

0.97–1.01

95% CI

White race

0.51***

Age

Female gender

OR

Variable

Model 1

Table 2  Results of Binary Logistic Regression for Activity in Gardening or Yard Work

383

0–12 years education

Activity limitations (any)

e

0.61–1.26 0.41–0.61 0.58–0.89

0.88 0.72**

0.62–1.25

0.91–2.46

0.75–1.66

0.46–1.03

0.43–0.92

0.99–1.03

95% CI

0.50***

0.88

1.49

1.11

0.68

0.63*

1.01

OR

Model 2

0.84

0.72**

0.49***

0.86

0.87

1.49

1.11

0.69

0.63*

1.01

OR

0.57–1.24

0.58–0.90

0.40–0.61

0.59–1.23

0.62–1.24

0.90–2.45

0.74–1.65

0.46–1.04

0.43–0.93

0.99–1.03

95% CI

Model 3

*p < .05. **p < .01. ***p < .001.

a Divorced, separated, marriage annulled, or unmarried. bSelf-reported, dichotomized into any or none. cSelf-reported, a sum of difficulties with bed/bathing, stair climbing, walking, and housework; 1 = least functional impairment, 4 = most functional impairment. dCalculated by body weight (kg) × height (m)2 then classified as underweight (29.99). eMeasured using 11-item CES-D and then standardized based on Wave I index scores; dichotomized into same or fewer versus more depressive symptoms than Wave I.

Note. OR = odds ratio; CI = confidence interval.

Depressive symptoms (≤Wave I)

d

Body-mass index

Functional limitationsc

b

0.52–0.99

0.78–1.99

1.25 0.72*

a

Unmarried

0.68–1.44

0.55–1.17

1.00

0.40–0.82

0.98–1.02

0.81

Female gender

Married

1.00 0.57**

Age

95% CI

White race

OR

Variable

Model 1

Table 3  Results of Binary Logistic Regression for Walking Activity

384

1.00–1.04

Activity limitations (any)

e

0.68–1.45 0.54–0.82 0.70–1.04

0.99 0.83

0.37–0.76

0.57–1.64

0.54–1.27

0.52–1.23

0.41–0.90

1.01–1.01

95% CI

0.67***

0.53***

0.97

0.83

0.80

0.61*

1.03*

OR

Model 2

1.07

0.83

0.67***

1.00

0.53**

0.97

0.84

0.79

0.61*

1.03*

OR

0.71–1.62

0.66–1.04

0.54–0.83

0.68–1.47

0.37–0.76

0.57–1.64

0.55–1.28

0.52–1.22

0.41–0.90

1.01–1.01

95% CI

Model 3

*p < .05. **p < .01. ***p < .001.

a Divorced, separated, marriage annulled, or unmarried. bSelf-reported, dichotomized into any or none. cSelf-reported, a sum of difficulties with bed/bathing, stair climbing, walking, and housework; 1 = least functional impairment, 4 = most functional impairment. dCalculated by body weight (kg) × height (m)2 then classified as underweight (29.99). eMeasured using 11-item CES-D and then standardized based on Wave I index scores; dichotomized into same or fewer versus more depressive symptoms than Wave I.

Note. OR = odds ratio; CI = confidence interval.

Depressive symptoms (≤Wave I)

Body-mass index

d

Functional limitationsc

b

0.33–0.67

0.54–1.50

0.47***

0.90

0–12 years education

Unmarrieda

0.57–1.31 0.52–1.20

0.86 0.79

0.39–0.85

Married

Female gender

95% CI

White race

1.02*

0.58**

Age

OR

Variable

Model 1

Table 4  Results of Binary Logistic Regression for Active Sports and Exercise

Correlates of Three Physical Activity Types   385

We tested for nonlinear relationships between the three activity types, age, and BMI. No significant relationship was detected between either variable in any physical activity type. Inclusion of quadratic BMI resulted in a nonsignificant linear association between BMI and participation in walking. Results reported in Tables 2, 3, and 4 reflect analyses performed with only the linear BMI variable. Finally, we examined the interaction of variables significantly associated with each outcome by gender. Only the interaction of functional health by gender emerged as marginally significant (p = .06), whereby greater functional disability was associated with lower participation in sports or exercise among men but not women.

Discussion According to the Health Promotion Model (Pender 1987, 1996), successful promotion of physical activity relies on identifying characteristics that influence physical activity adoption and participation. We examined sociodemographic and healthrelated correlates of three types of physical activity—walking, gardening or yard work, and sports or exercise—in the Americans’ Changing Lives Study, a study representative of the U.S. noninstitutionalized population. The unique features of this study include the examination of three types of activity within one study, the inclusion of multiple sociodemographic and health-related variables, the exploration of the potential modifying effect of gender, and the use of a representative sample. Although the correlates of activity participation varied somewhat across the three outcomes, gender and functional health appeared to be consistent, and particularly strong, correlates of activity. Women were far less likely than men to be active in all three categories of physical activity. This remained true even after controlling for functional limitations, which have been found to be more prevalent among older women than men (Crimmins, Saito, & Ingegneri, 1997). Given that greater disability has been demonstrated among older women, and functional fitness has been shown to improve with physical activity (Simons & Andel, 2006), it seems that great strides could be made in old-age disability by increasing the numbers of women who are physically active. Functional health was a strong correlate of being physically active in all three physical activity categories, which is consistent with previous findings (e.g., Janke et al., 2006). However, we also found that functional health modified the association between gender and participation in sports or exercise such that functional health was inversely related to participation in men only. This finding offers additional insight into potential avenues for investigation to determine how functional health may differentially relate to older adults’ participation in active sports and exercise by gender. Having any activity limitations because of health was not significantly associated with participation in physical activity. This suggests that although most of the sample reported having at least a few activity limitations, it does not necessarily result in an inverse relationship with physical activity participation. It is possible that they tend to select activities unrelated to their specific limitations. It is also possible that individuals learn to adapt or modify their chosen activities to enable participation despite the perceived limitations. More specific information about perceived activity limitations and physical activity participation would be necessary to determine how the two interact.

386  Haley and Andel

The influence of gender on physical activity participation throughout the life course has been observed previously. Globally, participation rates tend to be lower among women than men of any age (Ransdell, Vener, & Sell, 2004), a result confirmed by our data. Explanations for this phenomenon are likely multifaceted and will require further study to determine key factors that influence physical activity across the life cycle. Several studies have reported family obligations as barriers to physical activity participation in younger women (Dunn, 2008; Hoebeke, 2008; Sorensen & Gill, 2008; Walcott-McQuigg & Prohaska, 2001). Although these obligations may or may not decline in older age, it has been shown that previous patterns of physical activity affect later life activity habits (McAuley, 1992). In several interventional studies examining adherence and attrition rates, gender differences in adherence were not observed (Flegal et al., 2007; Jancey et al., 2007). This suggests that although overall physical activity participation is lower among females, once involved in a regular program, the genders’ adherence rates may be similar. A key factor in improving physical activity rates among older women, therefore, may be in promoting adoption of physical activity in earlier life. It is possible that improvements in physical activity participation in older women may benefit from better understanding barriers to physical activity in young and middle age, thereby paving the way for greater involvement at older ages. Contrary to previous observations, race was not significantly associated with participation in physical activity of any type in the main analyses, although it was significantly correlated with lower activity at the bivariate level. Evidently, other factors explained race differences in physical activity in our study. Consistent with most of the literature on education and physical activity, higher education about doubled the likelihood of participation in sports or exercise. There are several explanations for the education differences in physical activity involvement. First, it is possible that more educated people simply have more knowledge about the benefits of being physically active. Second, more education generally indicates higher socioeconomic status and, as such, may also indicate more time and financial resources to devote to physically active leisure activities. Finally, it is possible that more educated individuals incur greater exposure to these activities through the educational system, thereby promoting adoption. Education was not associated with being active in walking or gardening or yard work, which may be viewed as counterintuitive. It may be that the higher socioeconomic status among individuals with higher education leads some to delegate yard work to others. Walking, being an inexpensive form of exercise and a mode of transportation for urban dwellers, may not have been as susceptible to the effects of education as it relates to socioeconomic status in this sample. Interventional studies have also shown no differences in education levels with respect to attrition or adherence (Flegal et al., 2007; Jancey et al., 2007). Similar to gender differences in participation and adherence rates, this suggests that understanding barriers affecting initiation of physical activity may be the key to determining educational differences in physical activity participation. Furthermore, type of activity may be more influenced by education level than engagement in any type of activity. The only significant relationship between marital status and being physically active was in yard work or gardening. Married individuals were more likely to be

Correlates of Three Physical Activity Types   387

active in this category than unmarried individuals. There was no significant difference between married and widowed participants. It is possible that married individuals are simply more likely than unmarried individuals to live in single-family homes, which necessitates performing yard work and gardening in the upkeep of such dwellings. BMI was negatively associated with walking status only. The nature of the selective relationship observed in this sample is unclear. It suggests, however, that further study may reveal a stronger relationship between BMI and specific types of physical activity involvement, rather than overall involvement. Neither age nor quadratic age was associated with decreased participation in physical activity, and advancing age was associated with increased participation in sports and exercise. This finding partially goes along with previous findings that physical activity participation may not vary significantly in older adulthood (Resnick et al., 2000), or at least not until the eighth decade (Janke et al., 2006). Our finding of an inverse association between age and participation in sports or exercise may reflect trends of increasing activity after retirement that were only partially weakened by advanced old age. As with all cross-sectional examinations, caution should be exercised when interpreting the results. It is possible, if not probable, that study participants who were still alive, community dwelling, and able to participate in Wave IV interviews were the most active and healthy of the original participants. Furthermore, the directionality of observed relationships cannot be established with cross-sectional data. It is unclear whether individual characteristics influence physical activity participation or if those factors are affected by physical activity participation. More longitudinal evidence is needed to determine the direction of influence. An additional limitation of the study was the use of single-item self-reported physical activity measures, which may limit the reliability and generalizability of the results. Future investigations would benefit from the use of validated measures that assess intensity and duration components of physical activity, as well. It is important to note that the independent variables in this study were not correlated with all three physical activity outcomes equally. Although direct comparison of the three outcomes and their correlates is not possible with the current analyses, it suggests that different personal and health characteristics may be associated with various types of physical activity. Given that characteristics may differ depending on activity type and subpopulations such as older adults, information relevant to these relationships may help develop a better understanding of how to best promote physical activity among these subgroups. Further research in this area should focus on the temporal relationships between biological, sociocultural, and psychological factors and specific physical activity outcomes. Other implications for future research and practice based on the current findings include the need for more longitudinal studies of physical activity in older adults to determine age’s role in activity status. More gender-specific research is needed to examine specific barriers to physical activity across the life course, as well as ways to overcome these barriers. Similarly, identification of barriers to certain types of physical activity associated with less education may be useful in guiding promotion of physical activity among this subgroup.

388  Haley and Andel

References Allison, P.D. (1999). Logistic regression using the SAS system. Cary, NC: SAS Institute. Centers for Disease Control and Prevention. (1999). Physical activity and health: A report of the Surgeon General. Retrieved from http://www.cdc.gov/nccdphp/sgr/summary.htm Crespo, C.J., Smit, E., Andersen, R.E., Carter-Pokras, O., & Ainsworth, B.E. (2000). Race/ ethnicity, social class and their relation to physical inactivity during leisure time: Results from the Third National Health and Nutrition Examination Survey, 1988–1994. American Journal of Preventive Medicine, 18(1), 46–53. Crimmins, E.M., Saito, Y., & Ingegneri, D. (1997). Trends in disability-free life expectancy in the United States, 1970–1990. Population and Development Review, 23(3), 555–572. Droomers, M., Schrijvers, C.T.M., & Mackenbach, J.P. (2001). Educational level and decreases in leisure time physical activity: Predictors from the longitudinal GLOBE study. Journal of Epidemiology and Community Health, 55, 562–568. Dunn, M.Z. (2008). Psychosocial mediators of a walking intervention among African American women. Journal of Transcultural Nursing, 19(1), 40–46. Ferrucci, L., Izmirlian, G., Leveille, S., Phillips, C.L., Corti, M.C., Brock, D.B., & Guralnik, J.M. (1999). Smoking, physical activity, and active life expectancy. American Journal of Epidemiology, 149, 645–653. Sizer Fitts, S., Won, C.W., Williams, B., Snyder, S.J., Yukawa, M., Legner, V.J., . . . Phelan, E.A. (2008). What is the optimal duration of participation in a community-based health promotion program for older adults? Journal of Applied Gerontology, 27, 201–214. Flegal, K.E., Kishiyama, S., Zajdel, D., Haas, M., & Oken, B.S. (2007). Adherence to yoga and exercise interventions in a 6-month clinical trial. BMC Complementary and Alternative Medicine, 7, 37–43. Hoebeke, R. (2008). Low-income women’s perceived barriers to physical activity: Focus group results. Applied Nursing Research, 21(2), 60–65. House, J.S. (2006). ICPSR version 2. Ann Arbor, MI: Inter-university Consortium for Political and Social Research, 2007. Codebook: SIND-014(2006). ICPSRNO# 4690. Retrieved from http://www.icpsr.umich.edu/8080/ICPSR-STUDY/03394.xml House, J.S., Lepkowski, J.M., Kinney, A.N., Mero, R.P., Kessler, R.C., & Herzog, A.R. (1994). The social stratification of aging and health. Journal of Health and Social Behavior, 35(3), 213–234. Hubert, H.B., Bloch, D.A., Oehlert, J.W., & Fries, J.F. (2002). Lifestyle habits and compression of morbidity. Journal of Gerontology: Medical Sciences, 57A(6), M347–M351. Jancey, J., Lee, A., Howat, P., Clarke, A., Wang, K., & Shilton, T. (2007). Reducing attrition in physical activity programs for older adults. Journal of Aging and Physical Activity, 15, 152–165. Janke, M., Davey, A., & Kleiber, D. (2006). Modeling change in older adults’ leisure activities. Leisure Sciences, 28, 285–303. Jerome, G.J., Glass, T.A., Mielke, M., Xue, Q., Andersen, R.E., & Fried, L.P. (2006). Physical activity participation by presence and type of functional deficits in older women: The women’s health and aging studies. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences, 61, 1171–1176. Jonker, J.T., De Laet, C., Franco, O.H., Peeters, A., Mackenbach, J., & Nusselder, W.J. (2006). Physical activity and life expectancy with and without diabetes. Diabetes Care, 29(1), 38–43. Kahng, S.K. (2008). Overall successful aging: Its factorial structure and predictive factors. Asian Social Work and Policy Review, 2, 61–74. Kohout, F.J., Berkman, L.F., Evans, D.A., & Cornoni-Huntley, J. (1993). Two short forms of the CESD-D Depression Symptoms Index. Journal of Aging and Health, 5, 179–193. Lantz, P.M., Lynch, J.W., House, J.S., Lepkowski, J.M., Mero, R.P., Musick, M.A., & Williams, D.R. (2001). Socioeconomic disparities in health change in a longitudinal study of US adults: The role of health-risk behaviors. Social Science & Medicine, 53(1), 29–40.

Correlates of Three Physical Activity Types   389

Leveille, S.G., Guralnik, J.M., Ferrucci, L., & Langlois, J.A. (1999). Aging successfully until death in old age: Opportunities for increasing active life expectancy. American Journal of Epidemiology, 149(7), 654–664. McAuley, E. (1992). The role of efficacy cognitions in the prediction of exercise behavior in middle-aged adults. Journal of Behavioral Medicine, 15(1), 65–88. Mouton, C.P., Calmbach, W.L., Dhanda, R., Espino, D.V., & Hazuda, H. (2000). Barriers and benefits to leisure-time physical activity among older Mexican Americans. Archives of Family Medicine, 9, 892–897. Nocon, M., Hiemann, T., Muller-Riemenschneider, F., Thalau, F., Roll, S., & Willich, S.N. (2008). Association of physical activity with all-cause and cardiovascular mortality: A systematic review and meta-analysis. European Journal of Cardiovascular Prevention and Rehabilitation, 15(3), 239–246. Pender, N.J. (1987). Health promotion in nursing practice (2nd ed.). Norwalk, CT: Appleton and Lange. Pender, N.J. (1996). Health promotion in nursing practice (3rd ed.). Stanford, CT: Appleton and Lange. Pettee, K.K., Brach, J.S., Kriska, A.M., Boudreau, R., Richardson, C.R., Colbert, L.H., . . . Newman, A.B. (2006). Influence of marital status on physical activity levels among older adults. Medicine and Science in Sports and Exercise, 38(3), 541–546. Radloff, L.S. (1977). The CES-D Scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401. Ransdell, L.B., Vener, J.M., & Sell, K. (2004). International perspectives: The influence of gender on lifetime physical activity participation. The Journal of the Royal Society for the Promotion of Health, 124(1), 12–14. Resnick, B., Palmer, M.H., Jenkins, L.S., & Spellbring, A.M. (2000). Path analysis of efficacy expectations and exercise behavior in older adults. Journal of Advanced Nursing, 31(6), 1309–1315. Riebe, D., Garber, C.E., Rossi, J.S., Greaney, M.L., Nigg, C.R., Lees, F.D., . . . Clark, P.G. (2005). Physical activity, physical function and stages of change in older adults. American Journal of Health Behavior, 29(1), 70–80. Salmon, J., Owen, N., Crawford, D., Bauman, A., & Sallis, J.F. (2003). Physical activity and sedentary behavior: A population-based study of barriers, enjoyment and preference. Health Psychology, 22(2), 178–188. Simons, R., & Andel, R. (2006). The effects of resistance training and walking on functional fitness in advanced old age. Journal of Aging and Health, 18(1), 91–105. Sorensen, M., & Gill, D.L. (2008). Perceived barriers to physical activity in Norwegian adult age groups, gender and stages of change. Scandinavian Journal of Medicine & Science in Sports, 18(5), 651–663. U.S. Department of Health and Human Services. (2000). Healthy people 2010 (2nd ed.). Washington, DC: U.S. Government Printing Office. Walcott-McQuigg, J.A., & Prohaska, T.R. (2001). Factors influencing participation of African American elders in exercise behavior. Public Health Nursing (Boston, Mass.), 18(3), 194–203. Warburton, D.E.R., Nicol, C.W., & Bredin, S.S.D. (2006). Health benefits of physical activity: The evidence. Canadian Medical Association Journal, 174(6), 801–809. Weiss, D.R., O’Loughlin, J.L., Platt, R.W., & Paradis, G. (2007). Five-year predictors of physical activity decline among adults in low-income communities: A prospective study. The International Journal of Behavioral Nutrition and Physical Activity, 4(2), 1–7. World Health Organization. (2006). Global database on body mass index. Retrieved from http://apps.who.int/bmi/index.jsp?introPage=intro_3.html