Skip to main content
NCBI home page
The Journals of Gerontology Series B: Psychological Sciences and Social Sciences logoLink to The Journals of Gerontology Series B: Psychological Sciences and Social Sciences
. 2016 Sep 13;73(8):1491–1500. doi: 10.1093/geronb/gbw116

Gender Differences in Longitudinal Trajectories of Change in Physical, Social, and Cognitive/Sedentary Leisure Activities

Deborah Finkel 1,✉, Ross Andel 2,3, Nancy L Pedersen 4,5
PMCID: PMC6890525  PMID: 27624718

Abstract

Objective

We examined changes in participation in cognitive, social, and physical leisure activities across middle and older adulthood and tested moderation of trajectories of change in participation by gender.

Method

In all, 1,398 participants in the Swedish Adoption/Twin Study of Aging (SATSA) completed a 7-item leisure activity questionnaire up to 4 times over 17 years. Mean baseline age was 64.9 years (range = 36–91); 59% were women. Factor analysis identified physical, social, and cognitive/sedentary leisure activity participation factors. Age-based latent growth curve models adjusted for marital status, gender, education, depressive symptoms, and physical health were used.

Results

Overall, results indicated stability in social activities, increase in cognitive/sedentary activities, and decrease in physical activities, as well as accelerated decline in all three types of activities after about the age of 70 years. Social activity remained mostly stable for women and declined for men. Women reported higher levels of cognitive/sedentary leisure activity across the study. Both men and women declined in physical leisure activity. Variance in leisure activities increased with age; men demonstrated more variance in social activities and women in physical activities.

Conclusions

Understanding change in leisure activities with age and by gender can have important implications for interventions and for use of leisure activity data in epidemiological research.

Keywords: Gender differences, Growth curve modeling, Swedish Adoption/Twin Study of Aging

Participation in leisure activities has been associated with favorable late-life psychological (Adams, Leibbrandt, & Moon, 2011; Cacioppo, Hughes, Waite, Hawkley, & Thisted, 2006; Silverstein & Parker, 2002), physical (Haskell et al., 2007), and cognitive (Crowe, Andel, Gatz, Pedersen, & Johansson, 2003; Hultsch, Hertzog, Small, & Dixon, 1999; Wilson et al., 2003) outcomes, although a downward adjustment in activities (Baltes, 1987), or in social relationships specifically (Carstensen, 1992), may be adaptive in aging. Multiple theories of aging discuss activity participation. For example, the continuity theory of aging (Atchley, 1989) has highlighted steady participation in leisure activity of the same type as the most adaptive strategy to aging even in the midst of age-related physical and psychological changes. The activity theory of aging posits that greater participation in leisure activities and substitution of activities that are no longer available with new activities (as opposed to simply reduction in activity) are the best ways to promote overall well-being with aging (Havighurst, 1961; Lemon, Bengtson, & Peterson, 1972). Building on these two theories, the innovation theory of successful aging (Nimrod & Kleiber, 2007) points out that innovation in activities, albeit infrequent in later life (Nimrod, 2007), can be a particularly satisfying and meaningful, potentially leading to favorable late-life outcomes.

Life-span developmental theory (Baltes, 1987) points to fewer gains and increasing losses as the driver of human aging. Adaptation to this process by selective optimization with compensation, which includes reduction in activities and/or moving on to activities where age-related limitation plays a lesser role, is central to this theory (Baltes & Carstensen, 1996). Kleiber and colleagues (Kleiber, McGuire, Aybar-Damali, & Norman, 2008) propose the concept of selective disengagement, whereby doing less due to age-related physical constrains might be life enhancing. Finally, the idea of successful aging (Rowe & Kahn, 1997) is innately intertwined with the notion of engaged lifestyle. According to this theory, physical and social leisure activities are integral to all three principles of the successful aging theory: physical, mental, and social well-being, although the socioemotional selectivity theory (Carstensen, 1992) would argue that decline in social activity may be adaptive as aging individuals focus on fewer, more meaningful relationships.

Activity participation, defined here as the frequency of participating in physical, social, and cognitive/sedentary activities, appears central to the process of aging. Understanding changes in activity participation across older adulthood is crucial for further substantiating or refuting these existing theories. However, only a few studies to date have investigated how participation in leisure activities may change over time. For example, even though decline in overall activity participation was recently reported for both men and women, level of activity in late life tended to be a continuation of level of activity in earlier life (Agahi, Ahacic, & Parker, 2006). Stability in activity participation was also reported by the PATH Through Life Project (Bielak, Anstey, Christensen, & Windsor, 2012).

Conversely, a steady decline in participation over time was observed for cognitive, social and physical activities in the Victoria Longitudinal Study (Small, Dixon, McArdle, & Grimm, 2012), with decline in physical activity accelerated later in life. In data from the Americans’ Changing Lives (ACL) study (House et al., 1994), participation in leisure activities declined exponentially with increasing age across social and physical leisure activities (Janke, Davey, & Kleiber, 2006) and physical activities (Shaw, Liang, Krause, Gallant, & McGeever, 2010). Conversely, The PATH study reported an overall increase in physical activity over 8 years in participants initially aged 60–64 years (Bielak, Cherbuin, Bunce, & Anstey, 2014).

The role of gender in trajectories of activity participation is not well understood, and no theory describes gender differences in leisure activities specifically. Intuitively, gender differences in leisure activities may be expected based upon the notion that interests and hobbies tend to differ along certain patterns for men and women, gender-based social roles exist, and men and women tend to differ to at least some extent in their experiences as they go through life. For example, Nimrod and Kleiber (2007) have suggested that women may be more likely than men to start new activities in older adulthood, possibly leading to more stability in activity participation among women. They also suggest that women may value social activity more than men. With respect to physical activities, there is evidence that men participate in these activities more frequently (Janke et al., 2006) and are less likely to show decline (Shaw et al., 2010), probably as a function of more physical limitations in women compared with men.

Building on previous research, we set out to investigate change in leisure activity participation, measured as frequency of participation of seven activities that were cognitive/sedentary, social, or physical in nature, across four waves of data collection spanning 17 years, both overall and in men and women separately. We posit that better understanding of trajectories of change in leisure activities and of gender differences in these changes is needed and can help in designing future observational and intervention studies investigating leisure activities in relation to health outcomes.

We used four identical assessments of participation in leisure activities in three domains (physical, social, and cognitive/sedentary) collected from members of the Swedish/Adoption Twin Study of Aging (SATSA; Finkel & Pedersen, 2004) in years 1993, 2004, 2007, and 2010. In addition to being able to model change directly, we also (a) look at three major leisure activity domains—cognitive/sedentary, social, and physical—within one study and (b) base our analysis on data spanning 17 years. As a result, our approach can improve our understanding of leisure activity participation profiles across older adulthood. Given the age range of participants, and the fact that we modeled change on the basis of age (rather than simply time as done previously), we were able to estimate change in leisure activity participation between 36 and 91 years of age. We hypothesized that steady decline in participation would be observed in all three leisure activity domains over time, but we also suspected an accelerated decline toward the end of the age range. We also hypothesized that women would report more participation in social activities whereas men would participate in physical activities more often than women.

Method

Participants

Ascertainment procedures for SATSA have been described previously (Finkel & Pedersen, 2004). In brief, the sample is a subset of twins from the population-based Swedish Twin Registry. The base population comprises all pairs of twins who indicated that they had been separated before the age of 11 years and reared apart, and a sample of twins reared together matched on the basis of gender and date and county of birth. Twins were initially mailed a questionnaire (Q1) in 1984, and afterwards, questionnaire data were collected at irregular intervals. The leisure activity data included in the present analyses were first collected at the fourth questionnaire wave (Q4) in 1993 and included in Waves Q5 (2004), Q6 (2007), and Q7 (2010).

The current sample was 58% women, and there were no gender differences in rates of participation in questionnaire waves, χ2(4,1398) = 4.64, not significant. Nearly half of the sample (46%) had data from three or more waves. Sample size and descriptive information are presented in Table 1.

Table 1.

Sample Demographics

Variable Year assessed Men Women
N Mean (SD) N Mean (SD)
Q4 Age* 1993 505 63.7 (12.2) 690 65.4 (13.3)
Q5 Age 2004 268 69.9 (10.6) 381 70.5 (11.9)
Q6 Age 2007 227 71.6 (9.9) 334 72.3 (11.0)
Q7 Age 2010 181 74.6 (9.8) 262 75.3 (11.0)
Education* 1984 511 1.8 (1.0) 698 1.6 (0.8)
Depression* 1993 511 10.4 (7.9) 698 13.1 (9.5)
Illness summary* 1993 511 3.6 (3.4) 698 4.6 (2.6)
Open in a new tab

Note: *Difference between men and women is significant at p < .01.

Measures

Starting at Q4, participants were asked how often, on a scale of 1 (daily) to 5 (never), they participated in the seven leisure activities listed in Table 2. Factor analysis across the four questionnaire waves indicated that responses consistently clustered into three factors: social activities, cognitive/sedentary activities, and physical activities. The factor structures at the four measurement occasions are presented in Table 2. Taken together, the factors explained 58.9% to 62.7% of the variance. Inter-factor correlations were modest, indicating little overlap among the three types of leisure activities. Because the factor loadings differed somewhat at each Q, relevant items were summed to construct each leisure factor to ensure measurement invariance. The cognitive and physical factors were fairly normally distributed, but the social factor demonstrated positive skew; therefore, all factors were rank-normalized to minimize skew. The factors were then translated to T-scores using means and variances from Q4 such that high scores indicated more activity.

Table 2.

Factor Analysis of Leisure Questions at Each Questionnaire Wave

Item Q4 Q5 Q6 Q7
SOC COG PHY SOC COG PHY SOC COG PHY SOC COG PHY
Exercise sports .08 −.08 .84 .14 −.16 .75 .12 −.18 .85 .13 −.20 .83
Walking −.10 .31 .51 −.09 .19 .71 −.10 .40 .54 −.12 .28 .70
Club meetings .76 .01 .19 .76 −.02 .10 .81 −.01 .12 .77 .04 .10
Church activities .79 .02 −.32 .81 .04 −.24 .72 −.02 −.24 .76 .00 −.20
Study circle or class .56 .07 .34 .59 .05 .25 .62 .10 .17 .72 .05 .14
Reading books .12 .69 .00 −.02 .80 .07 .08 .70 .12 .13 .71 .07
Puzzles, chess, etc. −.03 .79 −.16 .07 .76 −.06 .02 .82 −.21 −.03 .81 −.08
Variance explained (%) 27.5 17.3 14.1 25.0 19.1 18.6 25.6 18.7 15.0 29.1 17.9 14.9
Inter-factor correlations
 Cognitive .12 .11 .07 .12
 Physical .13 .14 .17 .11 .09 .14 .13 .16
Open in a new tab

Notes: Factor loadings greater than .50 are in bold face. Q4 through Q7 refer to questionnaire waves.

COG = cognitive/sedentary factor; PHY = physical factor; SOC = social factor.

Education

In SATSA, education is rated on a 4-point scale from 1 (elementary school) to 4 (university or higher).

Depressive symptoms

Depression was measured at Q4 via the Center for Epidemiological Studies of Depression (CES-D) scale (Radloff, 1977), which assesses self-report of depressive feelings and behaviors during the last week.

Physical health

A summary of self-reported illnesses was collected at Q4 that incorporated measures of 13 organ systems, including cardiovascular disorders, respiratory disorders, musculoskeletal disorders, and central nervous system disorders (Harris, Pedersen, Stacey, McClearn, & Nesselroade, 1992). The illness summary indicates the number of chronic illnesses the participant has experienced, and it ranges from 0 to 12 in this sample.

Marital status

Marital status at baseline (Q4) was indicated using a dichotomous variable: married versus widowed/single.

Statistical Method

A growth curve model was used to examine longitudinal trajectories for the leisure activities and the impact of gender, education, depression, and physical health on those trajectories. Marital status was controlled in all models. The structural model can be considered as a multilevel random coefficients model (Bryk & Raudenbush, 1992; McArdle & Anderson, 1990). The model provides estimation of fixed effects, that is, fixed population parameters as estimated by the average growth model of the entire sample, and random effects, that is, interindividual variability in intraindividual change in growth model parameters. Growth curve models take into account missing data by giving more weight to individuals with the most time points.

The quadratic latent growth curve model (LGCM) that was applied to the data (Bryk & Raudenbush, 1992; Finkel, Reynolds, McArdle, Gatz, & Pedersen, 2003) is presented in Figure 1. Individual scores at any one time are a function of a latent intercept (I), the linear slope (S), the quadratic term (Q), and random error (U0 − U3). The paths from the latent slope factors to the observed scores are the age basis coefficients, B1t and B1t2. The age basis serves as a marker for the age of the individual at each time of measurement, adjusted for the centering age (age 65 years). The model fitting procedure entails fitting individual growth models to all available data; repeated measurements are indicated by the y0 through y3 variables. The random errors or uniquenesses (u0 − u3) represent unaccounted variation from fitting the growth model to the activity participation measures; these time-specific residual variances were constrained to be equal over time. The means (Mi = mean intercept; Ms = mean slope; Mq = mean quadratic; Mc = mean covariate) are the estimates of the average performance and average amount of change. Finally, the relationships among the intercept and rates of change are represented by the correlations Ris, Riq, and Rsq. Although the covariate in the figure (C) is measured with a single indicator (Cov), this can also be expanded to include multiple indicators: education, depression, health, and marital status. Correlations between the covariate and the LGCM parameters (Rci, Rcs, and Rcq) allow us to examine the impact of covariates on components of the growth curve. Finally, using gender as a class variable allows us to test the fit of separate growth models for men and women.

Figure 1.

Figure 1.

Open in a new tab

Latent growth curve model.

The current analyses focus on individual performance, making it necessary to eliminate any bias to standard errors resulting from the inclusion of twins, by including a correction for twin pairs in the modeling. Variation and covariation in the latent growth curve parameters, that is, the random effects, were partitioned into within- and between-pair components to account for pair dependency. The random- and fixed-effects parameter estimates were obtained using PROC Mixed in SAS 9.4 (SAS, 2002–2012). Nested models were compared using the difference chi-square test obtained by taking the difference between the obtained model fit statistics [−2LL: −2ln(Likelihood)] and testing its significance with the degrees of freedom equal to the difference in the number of parameters of the two models.

Results

As presented in Table 1, women who participated at Q4 were on average 1.7 years older than men, t(1193) = 3.01, p < 01, but men/women did not differ significantly at the later waves. Education was significantly higher for men than for women, t(1207) = 3.33, p < .01, and so was CES-D score, t(1207) = 5.32, p < .01. Women reported on average one more illness than men, t(1207) = 7.21, p < .01, and more men (75%) than women (56%) were married, χ2(1,1398) = 44.67, p < .01.

To examine trajectories of change in the leisure activity factors, six nested latent growth curve models were fit to the data. Results of progressive comparisons of the fit of each model are presented in Table 3. Although including marital status as a covariate did not significantly affect model fit, it did result in small adjustments in parameter estimates. Therefore, marital status was included as a covariate of the three LGCM parameters in all models. The first model was the baseline model, including fixed effects (intercept, linear slope, and quadratic terms) and random effects (variances around each parameter, correlations among the parameters, and residual variance). In Model 2, gender was incorporated in the model as a class variable for fixed effects, only allowing us to examine gender differences in mean LGCM parameters. Results of model comparisons indicated significant improvement in model fit for two of the leisure factors: social and cognitive/sedentary. For example, for the social factor change in model fit from Model 1 to Model 2 was 8.0 (df = 3, p < .05). In Model 3, gender was added in the random effects as well, allowing all variances and covariances, as well as mean parameter estimates, to differ for men and women. Significant improvement in model fit occurred for all three factors, although the effect was largest for the social factor. Thus, significant gender differences in changes in variance in the leisure factors with age were indicated.

Table 3.

Model Fit Statistics: −2LL

Models Number of parameters Leisure factors
Social Cognitive Physical
1. Baseline 13 19,350.1 19,328.4 19,361.9
2. Add gender in fixed effects 16 19,342.1* 19,267.5** 19,357.6
3. Add gender in fixed and random effects 23 19,326.3* 19,251.5* 19,342.8*
4. Add education in fixed effects 26 19,314.0** 19,201.0** 19,331.1**
5. Add depression in fixed effects 29 19,262.6** 19,196.2 19,328.3
6. Add illness summary in fixed effects 32 19,217.9** 19,191.2 19,309.5**
Open in a new tab

Notes: *Model fit is significantly different from previous model at p < .05.

**Model fit is significantly different from previous model at p < .01.

The purpose of Models 4–6 was to examine the impact of education, depression, and health on the fixed effects of the latent growth curve models: Education was added in Model 4, then depression was added as well in Model 5. Finally, the illness summary variable was added in Model 6. Model comparisons reported in Table 3 indicated that education had a significant impact on the LGCM parameters for all three leisure factors, but depression had a significant impact for the social factor, only. The illness summary variable added significantly to the fit of the model to the data for the social and physical factors.

The fixed-effects parameters estimated by Model 6 are presented in Table 4, allowing for an examination of the nature of the influence that gender, education, depression, and health had on trajectories of change in the leisure factors. Overall, we found significant quadratic effects (Age2) for the cognitive/sedentary and physical factors, indicating nonlinear change with age. Consider the social factor: the significant Education × Intercept parameter (1.58) indicates the extent to which the mean intercept is adjusted based on education. In other words, more education was associated with higher mean social activity. Physical health also significantly impacted the intercept (0.41), such that more illnesses were associated with a higher mean level of (relatively sedentary) social activities. The significant Age × Gender parameter (−0.07) indicates the correction from the mean linear slope for men; thus, at the centering age (65 years) social activity was declining significantly faster in men than in women. The significantly negative Age2 × Gender term means declines in social activity accelerated faster for men than for women. These gender differences are evident in the trajectories reported in Figure 2: Trajectories estimated by Models 3–6 are reported to allow comparison of the impact of the covariates. In the top panel, it is clear that social activity was fairly stable for women, but declined significantly after the age of 65 years for men. Although the parameter estimates reported in Table 3 indicate significant Age2 × Depression effects, the effect size (depicted in Figure 2) was much smaller than the effect of gender.

Table 4.

Parameter Estimates (SE) of Fixed Effects From Model 6

Fixed effect Social Cognitive Physical
Number of observations 2,746 2,817 2,718
Intercept 50.02 (.41)** 52.52 (.40)** 51.61 (.37)**
Intercept × Gender 0.16 (.59) −4.07 (.62)** −0.26 (.60)
Intercept × Education 1.58 (.30)** 1.61 (.32)** 1.23 (.30)**
Intercept × Depression −0.05 (.04) −0.01 (.04) −0.02 (.03)
Intercept × Illness 0.41 (.13)** 0.27 (.13)* −0.01 (.13)
Age 0.034 (.023) 0.060 (.023)** −0.106 (.023)**
Age × Gender −0.072 (.033)* −0.072 (.032)* 0.068 (.036)
Age × Education −0.003 (0.17) 0.014 (.017) −0.005 (.019)
Age × Depression 0.001 (.002) −0.001 (.002) −0.001 (.002)
Age × Illness −0.008 (.007) −0.008 (.007) −0.015 (.007)*
Age2 −0.0012 (.0012) −0.0063 (.0013)** −0.0084 (.0013)**
Age2 × Gender −0.0059 (.0018)** −0.0015 (.0020) −0.0017 (.0023)
Age2 × Education 0.0017 (.0009) 0.0016 (.0010) 0.0025 (.0011)*
Age2 × Depression −0.0003 (.0001)** −0.0002 (.0001) −0.0001 (.0001)
Age2 × Illness −0.0004 (.0004) −0.0006 (.0004) −0.0008 (.0004)
Open in a new tab

Notes: *Parameter estimate is significantly different from zero at p < .05.

**Parameter estimate is significantly different from zero at p < .01.

Figure 2.

Figure 2.

Open in a new tab

Trajectories in mean leisure activity estimated by the latent growth curve Models 3, 4, 5, and 6.

The results for the cognitive/sedentary leisure factor are quite different. Parameter estimates in Table 4 indicated that the impacts of education and physical illness were limited to the intercept, only, whereas gender impacted intercept and age. The longitudinal trajectories for the cognitive/sedentary factor in Figure 2 show a clear mean difference in cognitive/sedentary leisure activity: On average, men scored 4.5 lower on the cognitive/sedentary T-score factor, or about one half a standard deviation. Both men and women showed modest increases in cognitive/sedentary leisure activity up to the age of 65 years, followed by declines in cognitive/sedentary leisure activity beginning around the age of 75 years.

Finally, gender differences were most modest for the physical factor: Model comparisons indicated gender differences in random effects (or variance) only, not in fixed effects. None of the interactions with gender reported in Table 4 achieved statistical significance. Education significantly impacted both the mean and quadratic rate of change in the physical factor. More education resulted in greater physical activity and slower decline in physical activity. Physical health impacted linear changes with age, with better health being associated with slower decline. The trajectories presented in Figure 2 show consistently decreasing physical leisure activity with age, with marginally faster decline for women.

Gender differences in change trajectories for total variance (estimated from the random effects) resulting from Model 6 for the three leisure factors are presented in Figure 3. Variance increased with age for all three leisure factors, but the increases were most dramatic for cognitive/sedentary and physical activity factors in late adulthood. Even though the data were corrected for physical illness, the likelihood of which increases with age, we still observed increasing variability in participation in leisure activities with age. Gender differences in these two factors were generally modest, with men demonstrating more variability than women in cognitive/sedentary activities and women demonstrating more variance than men in physical activities. The largest gender difference was found for social leisure activities: Men demonstrated about 30% more variance in social activities than women.

Figure 3.

Figure 3.

Open in a new tab

Trajectories in total variance in leisure activities estimated by the latent growth curve Model 6.

Because birth year ranged from 1886 to 1958 in the current sample, it is possible that these results vary across birth cohorts. Therefore, we subsequently divided the sample into two cohorts at the median birth year (1927) to create cohorts born before and after the onset of the Great Depression. Results indicated no cohort differences in the LGCM results for the physical leisure factor. A significant cohort effect (p < .001) on intercept of the LGCM was found for the social factor: The mean intercepts at the age of 65 years were 4 points higher for both men and women born before 1927. Finally, both cohort (p < .001) and cohort by sex (p < .001) effects were found for the intercept in the cognitive/sedentary factor. Men born before 1927 had a slightly higher mean (1.5 points) than later born men; however, women born before 1927 had a lower mean (3 points) than later born women. (Detailed results are available from authors.)

Discussion

We investigated trajectories of change in participation in cognitive/sedentary, social, and physical leisure activities over four waves and 17 years of follow-up. Our hypothesis regarding steady decline in all three leisure activity domains, based on previous research, was supported only to some extent. Controlling for gender, education, depression, physical health, and marital status, we found that social activity participation declined, but only toward the latter part of the age range. We also found an overall decline in participation in physical activities, which appeared to be a function of (a) decline after the age of 65 years and (b) accelerated decline in advanced old age. Finally, we found an increase in participation in cognitive/sedentary activities, mainly attributable to an increase in participation up until about the age of 70 years.

Our results also offer partial support regarding our second hypothesis that women would participate more in social activities and men in physical activities. We found that women were in fact participating more in social activities, but these gender-based differences emerged only in advanced old age. Participation in physical activities was quite similar, with men being potentially slightly more active in advanced old age only.

In sum, we found support for a relative consistency overall for cognitive/sedentary activity in men and women and for social activity among women, which may be viewed as consistent with the continuity theory of aging (Atchley, 1989), but may also reflect replacement of activities, hence plausibly providing support for the innovation theory (Nimrod, 2007). However, this pattern did not hold for social activity in men, where a significant relative decline was observed in advanced old age, and for physical activity, where both men and women declined, with even more decline observed in women. These results provide additional evidence for the effect of age on participation in physical activity reported previously (Shaw et al., 2010; Small et al., 2012).

With respect to social activity, we found stability until at least 70 years of age and the level of participation in social activities remained steady throughout the study for women. This may again suggest that participation in social activities is continuous throughout older adulthood (Atchley, 1989), but it may also suggest that individuals effectively replace activities in this domain as they age (Nimrod, 2007). Similar to previous research (Janke et al., 2006; Small et al., 2012), we found decline in social activities in advanced old age, but we also found that this decline was largely the function of decline in men, not in women, suggesting that men may stop participating in these types of activities with age, either spontaneously or by choice as part of “selective disengagement” (Kleiber et al., 2008).

Men have smaller support networks than women and thus marriage constitutes a larger portion of men’s social life (Dykstra & Fokkema, 2007). Moreover, especially in these older cohorts, women are often responsible for maintaining and fostering the social interactions of the couple (Dykstra & de Jong Gierveld, 1994; Rosenthal, 1985), particularly after the husband retires and loses contact with the social network he had at work. Therefore, widowhood may affect social activity participation in men more than in women. The role of retirement and widowhood in social activity participation among men versus women deserves additional attention.

It is important to note that our list of social activities was restricted to what Janke and colleagues referred to as “formal” activities (church, clubs, and study circles) while informal activities such as social visits were not available. It might be important to capture gender-based trajectories of these informal activities in future research.

Our results for cognitive/sedentary leisure activities, which indicated increased participation until about the age of 70 years, are quite unique. One possibility is that the observed results may be a function of the great age range that was available to us. It may be that as individuals free themselves of work-related responsibilities and approach/enter retirement, they increase their cognitive/sedentary activity participation. Studies that begin follow-up in older adulthood may not be able to capture this trend. However, Small and colleagues (2012), who included participants as young as 55 years of age, report a linear decline in activity, although it is not clear whether nonlinear trends were tested. It may be that modeling on the basis of time of study, which Small and colleagues used, as opposed to chronological age potentially obscured stability in cognitive activities during later portions of middle adulthood and early older adulthood.

Another possibility is that as physical function declines with age, older individuals turn to activities that are more sedentary, boosting participation in the cognitive/sedentary domain. This is consistent with the concept of selective optimization with compensation (Baltes & Carstensen, 1996) and should be tested in future research.

It is also important to note that women consistently showed greater participation in the cognitive/sedentary activities than men, possibly as a function of less involvement in occupation among women in this cohort. In this regard, it is interesting, surprising maybe, that men did not show an accelerated increase in cognitive/sedentary participation following retirement age.

Finally, results by Bielak and colleagues (2012), whose activity variable resembles a combination of our cognitive/sedentary and social activity variables, reported stability over 8 years in a cohort of people aged 60–64 years at baseline, providing support for our results in this domain.

With respect to physical leisure activities, our results corroborate previous findings that a decline in physical activity is to be expected, especially in women, and that this decline tends to be accelerated in later parts of older adulthood (Janke et al., 2006; Shaw et al., 2010; Small et al., 2012). It is likely that the decline is either an artifact of greater disability with age (Beltrán-Sánchez, Finch, & Crimmins, 2015) or the tendency of older adults to select fewer, more meaningful activities as they age (Kleiber et al., 2008). Men did not show greater participation in physical activities than women to the same extent as previous studies. Given that our study included adults who were relatively young at baseline, we can speculate that gender-based differences in physical activities may emerge relatively late in life only.

Variance in leisure activities increased with age overall, particularly starting in the 80s. Specifically, men demonstrated more variance in social and cognitive/sedentary activities, whereas women showed more variance in physical activities. Most likely, factors associated with physical health and thus mobility in later life contribute to greater variability in leisure activity participation. Future studies should consider this fact and apply information about leisure activity participation in advanced old age with caution.

It should be noted that education, depressive symptoms, and health all improved model fit. Education played an important role in overall participation in all three domains, and it seemed to moderate (reduce) the more precipitous decline in participation in social and physical activity observed toward the end of the age range. Depressive symptoms were related to accelerated decline in social activities in later life, and physical health was related to mean levels of social activity and accelerated decline in physical activities.

Finally, we found greater participation in social activity in participants born before 1927 than thereafter, which may be a function of generational changes (decline) in church attendance and involvement in clubs. We also found that men born before 1927 reported slightly higher levels of cognitive/sedentary activity than men born later, whereas women born before 1927 reported less cognitive/sedentary activity than women born later. The cohort by sex interaction likely reflects generational changes in access to education, particularly for women.

The study is not without limitations. First, we only had seven specific items describing leisure activity participation. A greater variety of items and potentially a diary-based approach could help further refine the findings. In addition, the categorization of activities as physical, social, and cognitive/sedentary cannot be considered “pure” as some physical (group sports) or cognitive/sedentary (chess) activities can also be partially social. In addition, motivation for engaging in activities may change with age. For example, some may be motivated by need for social interaction earlier in older adulthood and by health-related need to exercise later, whereas it may be the opposite for others. Second, our measure of social activities was limited to more formal activities as opposed to casual visits. However, the measure still taps into the extent of social activity participation, which has been found important for late-life health outcomes (Fratiglioni, Paillard-Borg, & Winblad, 2004; Seeman & Crimmins, 2001). Third, there are many variables that can influence or explain participation in activity such as, for example, having children. However, searching for these was not our goal. Instead, our goal was to bring attention to changes in activities with age and by gender, controlling for education, depression, and health.

In conclusion, using data from middle and older adulthood, we found a relatively steady participation in social activities in the sample overall and a decline in activity after about the age of 70 years in men. We also found an increase in cognitive/sedentary activity up until about the age of 70 years, with subsequent leveling and decline. Women reported significantly higher average participation in cognitive/sedentary activity across ages. Finally, we found decline in physical activity that was accelerated after about the age of 70 years, with the trajectory of decline somewhat steeper for women than for men.

The results can inform future research. For example, advanced old age to be a strong moderator of physical activity. Therefore, considering the interaction of age and physical activity may be crucial for valid examination of the role of leisure in late-life outcomes. Further, it may be useful to test whether intervening to boost men’s social engagement in late life has health or psychological benefits/drawbacks. Finally, uncovering the reasons for and potential benefits/drawbacks of the relatively low participation in cognitive/sedentary activity in men may be helpful.

Funding

The Swedish Adoption/Twin Study of Aging (SATSA) is supported by the National Institute of Aging (AG04563, AG10175, and AG08724), the MacArthur Foundation Research Network on Successful Aging, Swedish Council for Working Life and Social Research (97:0147:1B, 2009-0795), and Swedish Research Council (825-2007-7460 and 825-2009-6141).

Conflict of Interest

The authors declare no conflict of interest.

References

  1. Adams K. B. Leibbrandt S., & Moon H (2011). A critical review of the literature on social and leisure activity and wellbeing in later life. Ageing and Society, 31, 683–712. doi:10.1017/S0144686X10001091 [Google Scholar]
  2. Agahi N., Ahacic K., Parker M. G. (2006). Continuity of leisure participation from middle age to old age. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 61, S340–S346. doi:10.1093/geronb/61.6.S340 [DOI] [PubMed] [Google Scholar]
  3. Atchley R. C. (1989). A continuity theory of normal aging. The Gerontologist, 29, 183–190. doi:10.1093/geront/29.2.183 [DOI] [PubMed] [Google Scholar]
  4. Baltes M. M., & Carstensen L. L (1996). The process of successful ageing. Ageing and Society, 16, 397–422. doi:10.1017/S0144686X00003603 [Google Scholar]
  5. Baltes P. B. (1987). Theoretical propositions of life-span developmental psychology: On the dynamics between growth and decline. Developmental Psychology, 23, 611. doi:10.1037/0012-1649.23.5.611 [Google Scholar]
  6. Beltrán-Sánchez H. Finch C. E., & Crimmins E. M (2015). Twentieth century surge of excess adult male mortality. Proceedings of the National Academy of Sciences USA, 112, 8993–8998. doi:10.1073/pnas.1421942112 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bielak A. A. Anstey K. J. Christensen H., & Windsor T. D (2012). Activity engagement is related to level, but not change in cognitive ability across adulthood. Psychology and Aging, 27, 219. doi:10.1037/a0024667 [DOI] [PubMed] [Google Scholar]
  8. Bielak A. A. Cherbuin N. Bunce D., & Anstey K. J (2014). Preserved differentiation between physical activity and cognitive performance across young, middle, and older adulthood over 8 years. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 69, 523–532. doi:10.1093/geronb/gbu016 [DOI] [PubMed] [Google Scholar]
  9. Bryk A. S., & Raudenbush S. W (1992). Hierarchical linear models. London: Sage. [Google Scholar]
  10. Cacioppo J. T. Hughes M. E. Waite L. J. Hawkley L. C., & Thisted R. A (2006). Loneliness as a specific risk factor for depressive symptoms: cross-sectional and longitudinal analyses. Psychology and Aging, 21, 140. doi:10.1037/0882-7974.21.1.140 [DOI] [PubMed] [Google Scholar]
  11. Carstensen L. L. (1992). Social and emotional patterns in adulthood: support for socioemotional selectivity theory. Psychology and Aging, 7, 331–338. doi:10.1037/0882-7974.7.3.331 [DOI] [PubMed] [Google Scholar]
  12. Crowe M. Andel R. Gatz M. Pedersen N. L., & Johansson B (2003). Does participation in leisure activities lead to a reduced risk of dementia?The Journals of Gerontology, Series B: Psychological and Social Sciences, 58, P294–P255. PMID: 14507930 [DOI] [PubMed] [Google Scholar]
  13. Dykstra P. A., & de Jong Gierveld J (1994). The theory of mental incongruity, with a specific application to loneliness among widowed men and women. In: Erber R., Gilmour R. (eds), Theoretical frameworks for personal relationships (pp. 235–259). Hillsdale N.J: Erlbaum. [Google Scholar]
  14. Dykstra P. A., & Fokkema T (2007). Social and emotional loneliness among divorced and married men and women: Comparing the deficit and cognitive perspectives. Basic and Applied Social Psychology, 29, 1–12. doi:10.1080/01973530701330843 [Google Scholar]
  15. Finkel D., & Pedersen N. L (2004). Processing speed and longitudinal trajectories of change for cognitive abilities: The Swedish Adoption/Twin Study of Aging. Aging, Neuropsychology, and Cognition, 11, 325–345. doi:10.1080/13825580490511152 [Google Scholar]
  16. Finkel D., Reynolds C. A., McArdle J. J., Gatz M., Pedersen N. L. (2003). Latent growth curve analyses of accelerating decline in cognitive abilities in late adulthood. Developmental Psychology, 39, 535–550. doi:10.1037/0012-1649.39.3.535 [DOI] [PubMed] [Google Scholar]
  17. Fratiglioni L. Paillard-Borg S., & Winblad B (2004). An active and socially integrated lifestyle in late life might protect against dementia. The Lancet Neurology, 3, 343–353. doi:10.1016/S1474-4422(04)00767-7 [DOI] [PubMed] [Google Scholar]
  18. Harris J. R. Pedersen N. L. Stacey C. McClearn G., & Nesselroade J. R (1992). Age differences in the etiology of the relationship between life satisfaction and self-rated health. Journal of Aging Health, 4, 349–368. doi:10.1177/089826439200400302 [Google Scholar]
  19. Haskell W. L. Lee I.-M. Pate R. R. Powell K. E. Blair S. N. Franklin B. A., … Bauman A (2007). Physical activity and public health: Updated recommendation for adults from the American College of Sports Medicine and the American Heart Association. Circulation, 116, 1081. doi:10.1161/CIRCULATIONAHA.107.185649 [DOI] [PubMed] [Google Scholar]
  20. Havighurst R. (1961). The nature and values of meaningful free-time activity, in Kleemeier R., (Ed), Aging and leisure (pp. 309– 44). New York: Oxford University Press. [Google Scholar]
  21. House J. S., Lepkowski J. M., Kinney A. M., Mero R. P., Kessler R. C., Herzog A. R. (1994). The social stratification of aging and health. Journal of Health and Social Behavior, 35, 213–234. doi:10.2307/2137277 [PubMed] [Google Scholar]
  22. Hultsch D. F., Hertzog C., Small B. J., Dixon R. A. (1999). Use it or lose it: Engaged lifestyle as a buffer of cognitive decline in aging?Psychology and Aging, 14, 245–263. doi:0.1037/0882-7974.14.2.245 [DOI] [PubMed] [Google Scholar]
  23. Janke M. Davey A., & Kleiber D (2006). Modeling change in older adults’ leisure activities. Leisure Sciences, 28, 285–303. doi:10.1080/01490400600598145 [Google Scholar]
  24. Kleiber D. McGuire F. A. Aybar-Damali B., & Norman W (2008). Having more by doing less: The paradox of leisure constraints in later life. Journal of Leisure Research, 40, 343–359. ISSN: 2159-6417 [Google Scholar]
  25. Lemon B. W., Bengtson V. L., Peterson J. A. (1972). An exploration of the activity theory of aging: activity types and life satisfaction among in-movers to a retirement community. Journal of Gerontology, 27, 511–523. doi:10.1093/geronj/27.4.511 [DOI] [PubMed] [Google Scholar]
  26. McArdle J. J., & Anderson E (1990). Latent variable growth models for research on aging. In Schaie J. E. B. K. W. (Ed.), Handbook of the psychology of aging (3rd ed, pp. 21–44). New York: Academic Press. doi:10.1016/B978-0-12-101280-9.50008-5 [Google Scholar]
  27. Nimrod G. (2007). Expanding, reducing, concentrating and diffusing: Post retirement leisure behavior and life satisfaction. Leisure Sciences, 29, 91–111. doi:10.1080/01490400600983446 [Google Scholar]
  28. Nimrod G., Kleiber D. A. (2007). Reconsidering change and continuity in later life: Toward an innovation theory of successful aging. International Journal of Aging & Human Development, 65, 1–22. doi:10.2190/Q4G5-7176-51Q2-3754 [DOI] [PubMed] [Google Scholar]
  29. 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. doi:10.1177/014662167700100306 [Google Scholar]
  30. Rosenthal C. J. (1985). Kinkeeping in the familial division of labor. Journal of Marriage and the Family, 965–974. doi:10.2307/352340 [Google Scholar]
  31. Rowe J. W., & Kahn R. L (1997). Successful aging. The Gerontologist, 37, 433–440. doi:10.1093/geront/37.4.433 [DOI] [PubMed] [Google Scholar]
  32. Seeman T. E., & Crimmins E (2001). Social environment effects on health and aging. Annals of the New York Academy of Sciences, 954, 88–117. doi:10.1111/j.1749-6632.2001.tb02749.x [DOI] [PubMed] [Google Scholar]
  33. Shaw B. A. Liang J. Krause N. Gallant M., & McGeever K (2010). Age differences and social stratification in the long-term trajectories of leisure-time physical activity. The Journals of Gerontology, Series B: Psychological Sciences and Social Sciences, 65, 756–766. doi:10.1093/geronb/gbq073 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Silverstein M., & Parker M. G (2002). Leisure activities and quality of life among the oldest old in Sweden. Research on Aging, 24, 528–547. doi:10.1177/0164027502245003 [Google Scholar]
  35. Small B. J. Dixon R. A. McArdle J. J., & Grimm K. J (2012). Do changes in lifestyle engagement moderate cognitive decline in normal aging? Evidence from the Victoria Longitudinal Study. Neuropsychology, 26, 144. doi:10.1037/a0026579 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Wilson R. S. Bennett D. Bienias J. de Leon C. M. Morris M., & Evans D (2003). Cognitive activity and cognitive decline in a biracial community population. Neurology, 61, 812–816. doi:10.1212/01.WNL.0000083989.44027.05 [DOI] [PubMed] [Google Scholar]

Articles from The Journals of Gerontology Series B: Psychological Sciences and Social Sciences are provided here courtesy of Oxford University Press

RESOURCES