Longitudinal Associations Between Physical Activity and Purpose in Life Among Older Adults: A Cross-Lagged Panel Analysis

Abstract

Objective: The purpose of the current study was to examine the bidirectional relationship between physical activity (PA) and purpose in life in an older adult sample using longitudinal data. Method: Data were collected from three waves (2006, 2010, and 2014) of the Health and Retirement Study. A total of 4591 participants who had complete information of PA and purpose in life at baseline were included in the current study. A series of cross-lagged models were tested separately for vigorous-intensity PA, moderate-intensity PA, and light-intensity PA. Results: While higher levels of purpose in life were associated with more frequent engagement in future vigorous-intensity PA, moderate-intensity PA, and light-intensity PA, none of the PA variables predicted subsequent purpose in life. Discussion: The results did not support a reciprocal relationship between PA and purpose in life in older adults. It is more likely that purpose in life predicts PA, rather than vice versa.

Keywords

physical activity purpose meaning older adults

Early studies on health psychology in old adults have mainly focused on alleviating or treating mental disorders and dysfunction (Johnson & Acabchuk, 2018; Schmidt et al., 2011). This line of research is indeed important because mental disorders are linked with various health outcomes in older adults (Meeks et al., 2011). However, health psychology research in older adults should be concerned not only with mental illness and dysfunction but also supporting methods to enhance positive mental states. As defined by the World Health Organization (WHO), health is defined as “not merely the absence of disease or infirmity,” but also “a state of complete physical, mental, and social well-being” (World Health Organization, 2014). More recently, there is an increasing interest in positive psychology which has led to growing awareness of positive mental well-being, commonly conceptualized in terms of hedonic well-being and eudaimonic well-being and distinct from merely the absence of mental ill-being (Seligman & Csikszentmihalyi, 2000).

Hedonic well-being focuses on one’s experience of happiness or pleasure through the satisfaction of preference (Diener, 1984), and eudaimonic well-being highlights the importance of purposeful or goal-directed activities in actualizing one’s fullest potential (Ryan & Deci, 2001). Evidence has shown that both hedonic well-being and eudaimonic well-being are related to a number of health outcomes (Ryff & Boylan, 2016). However, compared with hedonic well-being, the current understanding on eudaimonic well-being is limited as it is less represented in the well-being literature, although evidence has shown eudaimonic well-being might be a more reliable and long-lasting indicator of individual or collective well-being (McMahan & Estes, 2011; Ryan et al., 2008).

Among several facets of eudaimonic well-being, purpose in life is a construct that has received increasing interest due to its potential in predicting and promoting a range of health outcomes. Purpose in life is generally defined as an individual’s sense of meaning, purpose, and direction in his or her life and is central to one’s well-being (Ryff, 2014; Steger et al., 2006). A small but increasing number of studies have shown that purpose in life is linked with a lower risk of stroke, Alzheimer’s disease, myocardial infarction, and all-cause mortality (Boyle et al., 2010; EKim et al., 2013; Roepke et al., 2014). In addition, literature has documented favorable associations between purpose in life and an array of physiological markers such as interleukin-6 receptors, salivary cortisol, and high-density lipoproteins, all of which are further linked with a range of diseases (Friedman et al., 2007; Jacobs et al., 2011).

From a lifespan perspective, purpose in life may even be more important in the older population than other age-groups because older adults on average have a lower level of purpose in life than young and middle-aged adults (Pinquart, 2002). As aging is associated with multiple physical and cognitive declines (Chen et al., 2015), identifying sources to maintain or increase the level of purpose in life may provide opportunities for novel interventions to promote health and protect or slow down the age-related declines in the older population.

Physical activity (PA) has been identified as a promising source of purpose in life among many other factors. Evidence supporting the positive relationship between PA and purpose in life has been emerging across different age-groups. For example, in a 6-month longitudinal investigation among female college students, Mack et al. (2012) found increases in health-enhancing PA were associated with increases in purpose in life. A randomized controlled trial showed that an 8-week Zumba intervention with 60 minutes each time and three times per week significantly improved purpose in life in adult women compared with the control group (Delextrat et al., 2016). The positive association between PA and purpose in life was documented among older adults by two cross-sectional studies (Ju, 2017; Kim et al., 2017).

On the other hand, several studies have documented the reverse effect that purpose in life might also have an impact on PA. For example, cross-sectional studies demonstrated that higher levels of purpose in life predicted greater PA in middle-aged women (Holahan et al., 2011), cardiac patients (Holahan et al., 2008), and older adults (Ruuskanen & Ruoppila, 1995). The positive association between purpose in life and PA was also supported when PA was objectively measured by accelerometers among community-living adults (Hooker & Masters, 2016). In addition, a longitudinal study found that higher levels of purpose in life at baseline were associated with higher levels of self-report PA at 13-month follow-up among adolescents (Brassai et al., 2015).

In summary, previous studies have either treated PA as a predictor of purpose in life and thus examined the PA-to-purpose relationship or considered purpose in life as a predictor of PA and thus examined the purpose-to-PA relationship. There is both evidence for the effect of PA on purpose in life and the effect of purpose in life on PA, suggesting that PA and purpose in life might be reciprocally related. Although several articles (Homan & Boyatzis, 2010; Ju, 2017; Ruuskanen & Ruoppila, 1995) have discussed the possible bidirectional relationship between PA and purpose in life, no previous study has simultaneously tested the reciprocal relationship between them in a same sample. The majority of previous studies examining the association between PA and purpose in life were cross-sectional in nature and tested only one association of the direction, which precludes us from answering the question “which-came-first.” The possible bidirectional relationship and the extent to which the association might be stronger in one direction than the other could be examined using longitudinal data in which PA and purpose in life are evaluated on two or more occasions. The existence of the bidirectional relationship between the two constructs would be inferred if PA predicts purpose in life over time, and in the meantime, purpose in life predicts PA over time.

It is well-known that the ideal way to test causality is by conducting rigorously controlled experimental studies. However, while true experiments are relatively controllable when testing the PA-to-purpose relationship, it is somewhat unwieldy to manipulate the purpose in life. Consequently, cross-lagged panel designs are recommended with both variables being measured repeatedly over time (Kenny, 1975). The cross-lagged panel model allows us to estimate the directional influence between variables while simultaneously controlling for correlations within time points and autoregressive effects (Kearney, 2017). The cross-lagged panel analysis has been previously used in exploring the reciprocal relationship between PA and other psychological constructs (Leonhardt et al., 2009). In addition, cross-lagged panel analysis can be performed using the structural equation modeling framework, which enables us to control for the measurement error of latent variables (Kline, 2015).

Taken together, given the importance of both PA and purpose in life for the older population and the largely unexplored reciprocal relationship between these two constructs, the purpose of the current study was to investigate the association between PA and purpose in life in older adults using longitudinal data while taking into account the possibility of a bidirectional relationship. More specifically, the current study aimed to use a three-wave cross-lagged panel analysis in the structural equation modeling framework to examine whether PA would predict subsequent purpose in life in the older population or vice versa, or both. In addition, previous studies investigating the relationship between PA and purpose in life did not differentiate PA of different intensities. As PA intensity is an important parameter of PA, the current study aimed to separately investigate the longitudinal relationship between vigorous-intensity PA, moderate-intensity PA, light-intensity PA, and purpose in life in older adults. Based on previous evidence, we hypothesized that (1) on average, the levels of PA and purpose in life would both decrease over time; (2) higher levels of PA would be associated with higher levels of subsequent purpose in life; and (3) higher levels of purpose in life at baseline would be associated with higher levels of subsequent PA.

Methods

Study Design and Participants

The current study used data collected from three waves (2006, 2010, and 2014) of the Health and Retirement Study (HRS). The HRS, conducted by the Institute for Social Research at the University of Michigan, is an ongoing nationally representative longitudinal study of more than 37,000 individuals aged 50 and over in the United States (Sonnega et al., 2014). The survey has been fielded every two years since its original data collection in 1992. The HRS has focused on a wide range of variables of health, cognition, and economics during the process of aging and retirement (Sonnega & Smith, 2015). In 2006, the HRS added the assessment of various psychosocial variables such as social support, loneliness, and purpose in life. The HRS has been widely used as a useful source in studying the changes in health associated with aging (Choi et al., 2018; Crosswell et al., 2018).

Since 2006, half of HRS participants were randomly chosen and assigned to an enhanced face-to-face interview with physical, biological, and psychosocial measures. The other half of the sample only completed the core interview in 2006 and received enhanced face-to-face interviews in 2008. The two half samples alternatively received the enhanced face-to-face interviews, and therefore, the longitudinal information for the psychosocial variables (e.g., purpose in life) is available every four years. Additional information about the design and procedures of the HRS has been published elsewhere (Sonnega et al., 2014). The current study used the data collected in 2006 (T1), 2010 (T2), and 2014 (T3) that contained the repeated assessment of PA and purpose in life in the same sample.

Participants who were 65 years or older and had complete measurement of PA and purpose in life at T1 were included in the current study. Data were available from 4591 participants at T1. It should be noted that the sample size was relatively small compared with the overall HRS sample. It was mainly due to the unavailable data of purpose in life at T1. Among them, 3687 participants were successfully reinterviewed at T2, and 2818 participants were successfully reinterviewed at T3.

Measures

Physical activity

Physical activity was measured using three items assessing the frequency of PA involved in one’s daily life including vigorous-intensity PA, moderate-intensity PA, and light-intensity PA. For example, the item for vigorous-intensity PA was “How often do you take part in sports or activities that are vigorous, such as running or jogging, swimming, cycling, aerobics or gym workout, tennis, or digging with a spade or shovel?” The response option was based on a 7-point scale ranging from 1 (hardly ever or never) to 7 (every day). Higher scores indicated higher levels of vigorous-intensity PA, moderate-intensity PA, and light-intensity PA, respectively. Evidence for validity indicated that these single-item measures were moderately correlated with PA measured by accelerometers and validated self-report tools (Milton et al., 2011; Milton et al., 2013; Smith et al., 2005).

Purpose in life

Purpose in life was measured using a seven-item subscale from the Ryff Scales of Psychological Well-Being (Ryff, 1989). A sample item was “I enjoy making plans for the future and working to make them a reality.” Respondents were asked to rate the extent to which they endorse each item on a 6-point Likert scale from 1 (strongly disagree) to 6 (strongly agree). After reversely coding the negatively worded items, a composite score was calculated by summing all items with higher scores representing higher levels of purpose in life. Previous studies have documented the adequate validity and reliability for the use of the seven-item scale for purpose in life in older adults (Homan, 2016; Vella-Brodrick & Stanley, 2013). Cronbach’s alpha for the current study was .72 at T1, .76 at T2, and .75 at T3, indicating good internal consistency.

Covariates

Several sociodemographic and health variables at T1 were included as covariates since evidence has shown they were potentially linked with purpose in life and PA.

Sociodemographic variables included self-reported age (in years), gender (male/female), race/ethnicity (white/Caucasian, Black/African American, and others), and marital status (married/not married), and educational level (no degree/high school diploma or GED/college degree or higher).

Health-related variables included self-rated health and an index of chronic conditions. Self-rated health was assessed using a single-item question “would you say your health is excellent, very good, good, fair, or poor?” with a 5-point Likert scale ranging from 1 (poor) to 5 (excellent). The number of chronic conditions was assessed using eight items asking participants if they have been diagnosed with eight conditions, including (1) high blood pressure, (2) diabetes, (3) cancer or malignant tumor (excluding minor skin cancer), (4) lung disease, (5) heart attack, coronary heart disease, angina, congestive heart failure, or other heart problems, (6) stroke, (7) emotional, nervous, or psychiatric problems, and (8) arthritis or rheumatism. An index ranging from 0 to 8 was created by summing all items with higher scores indicating the presence of more chronic conditions.

Data Analysis

Descriptive statistics for participant characteristics at baseline including sociodemographic and health variables were presented using means, SDs, or counts, as appropriate. The differences in baseline characteristics between individuals with and without missing data were examined using independent samples t-tests or chi-square tests. Means, SDs, and bivariate correlations for PA of different intensities and purpose in life at each wave were also presented. To examine the changes of PA and purpose in life over time, separate linear mixed models were conducted with vigorous-intensity PA, moderate-intensity PA, light-intensity PA, and purpose in life as the dependent variables and time as the independent variable while controlling for the baseline sociodemographic and health covariates. The descriptive statistics, independent samples t-tests, chi-square tests, and linear mixed models were performed using R version 3.5.

To estimate the reciprocal influences between PA and purpose in life among older adults, a series of cross-lagged structural equation models were performed. In the current study, purpose in life was conceived as a latent factor which has seven indicators (i.e., the seven items from the Scales of Psychological Well-Being). The univariate normality of each observed variable was examined by checking the skewness and kurtosis. The absolute value of skewness greater than 3 and/or the absolute value of kurtosis greater than 10 was considered as non-normal (Kline, 2015). For the non-normal observed variables, possible transformations (e.g., log transform and square transformation) would be applied.

For PA of each intensity, four models were tested to examine the bidirectional hypothesis between PA and purpose in life over time. Model 1 was considered as the baseline model in which only autoregressive paths are included (e.g., the path from PA at T1 to PA at T2 and the path from PA at T2 to PA at T3). These autoregressive paths provided information about the stability of the variables with higher path coefficients indicating greater stability. Model 2 had the autoregressive paths and paths from PA to subsequent purpose in life. Model 3 had the autoregressive paths and paths from purpose in life to subsequent PA. Model 4 was a fully cross-lagged model which included autoregressive paths, paths from PA to subsequent purpose in life, and paths from purpose in life to subsequent PA. For all models, correlations between the error terms of PA and purpose in life were allowed to be freely estimated. In each model, covariates described previously were controlled for at T1.

All cross-lagged panel analyses were computed using Mplus version 8.1 (Muthén & Muthén, 2019). Parameters in each model were estimated using the full information maximum likelihood estimation method, which is an advanced method handling miss data and produces unbiased parameters on the basis of all available observed variables (Arbuckle, 1996). Model fit was evaluated using the model chi-square with its degrees of freedom and p value, Steiger–Lind root mean square error of approximation (RMSEA), Bentler Comparative Fit Index (CFI), Tucker–Lewis Index (TLI), and standardized root mean square residual (SRMR). A good model fit would be indicated by nonsignificant chi-square, CFI ≥.95, TLI ≥.95, SRMR ≤.05, and RMSEA ≤.05 along with its 90% confidence interval ≤.1 (Kline, 2015). An adequate model fit would be indicated by nonsignificant chi-square, CFI ≥.90, TLI ≥.90, SRMR ≤.08, and RMSEA ≤.08 (Brown, 2015). It should be noted that the significance level of the model chi-square is highly sensitive in the case of a large sample size (Bollen, 1989). Therefore, the model chi-square would not be used as a decisive model fit index for a single model in the current study. Chi-square difference tests were used to compare the model fit between nested models (e.g., Model 1 and Model 2).

Replication

As mentioned above, half of HRS participants were randomly chosen and assigned to an enhanced face-to-face interview since 2006. The main analysis of the current study included participants who were interviewed in 2006, 2010, and 2014. To test the replicability of the results, we used the other half of the sample, that is, participants who were interviewed in 2008, 2012, and 2016, as a replication sample. The statistical analyses of the replication sample were identical to that of the main sample.

Results

Baseline Characteristics

Descriptive statistics of baseline characteristics for the overall sample were presented in Table 1. The mean age of participants at baseline was 74.32 ± 7.11 years. The majority of participants were female (56.87%), white/Caucasian (85.28%), and married (61.31%), and obtained a high school diploma or a college degree (77.19%). The mean score of self-rated health was 3.10, indicating a good level of self-rated health, while participants had an average of 2.37 chronic conditions.

Table 1.

Characteristics of the Overall Sample at Baseline.

Variable	Sample n = 4591
Age (mean ± SD)	74.32 ± 7.11
Gender
Male	1980 (43.13%)
Female	2611 (56.87%)
Race
White/Caucasian	3519 (85.28%)
Black/African American	550 (11.98%)
Others	129 (2.74%)
Marital status
Married	2814 (61.31%)
Not married	1776 (38.69%)
Education
No degree	1047 (22.81%)
High school diploma or GED	2546 (55.46%)
College degree or higher	998 (21.74%)
Self-rated health (mean ± SD)	3.10 ± 1.07
Chronic conditions (mean ± SD)	2.37 ± 1.39

Attrition Analyses

A total of 2761 participants had complete data at all three waves, and 1830 participants had available data at either one or two waves. Compared to those without missing data, attrition analyses revealed that participants with missing data at follow-ups were older (77.17 vs. 72.43 years, p < .001), more likely to be male (47.4% vs. 40.3%, p < .001), not married (43.2% vs. 35.6%, p < .001), and not graduate from high school (28.1% vs. 19.2%, p < .001). In addition, participants with missing data had a greater number of chronic conditions (2.68 vs. 2.17, p < .001) and lowers levels of self-rated health (2.78 vs. 3.31, p < .001), vigorous-intensity PA (1.65 vs. 2.06, p < .001), moderate-intensity PA (2.93 vs. 3.51, p < .001), light-intensity PA (3.28 vs. 3.83, p < .001), and purpose in life (29.52 vs. 32.05, p < .001). There was no difference regarding the race/ethnicity between participants with and without missing data.

PA and Purpose in Life at Each Wave

Descriptive statistics of PA and purpose in life at each wave were shown in Table 2. Bivariate correlations (Table 3) indicated that PA of different intensities and purpose in life within and between waves were all positively correlated (p’s < .001). The correlations within each wave were greater than the correlations between waves. In addition, values of skewness and kurtosis indicated normality for vigorous-intensity PA, moderate-intensity PA, light-intensity PA, and purpose in life at each wave. Linear mixed models (Table 4) revealed a decreasing pattern of PA of each intensity and purpose in life over time after controlling for a range of covariates. Specifically, vigorous-intensity PA, moderate-intensity PA, light-intensity PA, and purpose in life on average decreased .02, .08, .10, and .18 points in the corresponding scale each year (p’s < .001).

Table 2.

Descriptive Statistics of PA and Purpose in Life at Each Wave.

	T1(n = 4591)		T2(n = 3687)		T3(n = 2818)
	Mean	SD	Mean	SD	Mean	SD
Vigorous-intensity PA	1.90	1.46	1.93	1.43	1.89	1.45
Moderate-intensity PA	3.28	1.76	2.91	1.69	2.82	1.74
Light-intensity PA	3.61	1.56	3.20	1.53	3.02	1.50
Purpose in life	31.05	6.60	31.66	6.73	30.64	6.70

Note. PA = physical activity.

Table 3.

Bivariate Correlations between PA and Purpose in Life at Each Wave.^a

	1	2	3	4	5	6	7	8	9	10	11	12
1. VPA: T1	—
2. MPA: T1	.32	—
3. LPA: T1	.17	.40	—
4. PIL: T1	.19	.21	.19	—
5. VPA: T2	.41	.25	.16	.15	—
6. MPA: T2	.27	.36	.21	.19	.38	—
7. LPA: T2	.18	.22	.33	.19	.24	.37	—
8. PIL: T2	.18	.15	.14	.59	.18	.19	.21	—
9. VPA: T3	.36	.21	.07	.13	.45	.28	.14	.13	—
10. MPA: T3	.24	.29	.17	.20	.31	.39	.19	.19	.38	—
11. LPA: T3	.18	.20	.25	.18	.23	.28	.42	.19	.25	.36	—
12. PIL: T3	.16	.17	.13	.53	.17	.21	.19	.58	.18	.24	.23	—

^aAll the bivariate correlations were significant with p values <.001.

Table 4.

Summary of Linear Mixed Models (Coefficients and Standard Errors).

	Vigorous-Intensity PA		Moderate-Intensity PA		Light-Intensity PA		Purpose in Life
	B	SE	B	SE	B	SE	B	SE
Year	−.02^***	.00	−.08^***	.00	−.10^***	.00	−.18^***	.02
Baseline age	−.02^***	.00	−.02^***	.00	−.04^***	.00	−.15^***	.01
Gender (female)	−.31^***	.03	−.23^***	.04	.47^***	.04	.20	.17
Race (Black)	.07	.05	−.08	.06	−.23^***	.05	1.90^***	.26
Race (others)	.14	.10	.23	.12	−.10	.10	.28	.50
Education (high school)	.03	.04	.14^**	.05	.09^*	.04	1.36^***	.21
Education (college)	.32^***	.05	.36^***	.06	.17^**	.05	2.85^***	.26
Marital status (not married)	−.04	.04	−.02	.04	−.14^***	.04	−.89^***	.18
Self-rated health	.25^***	.02	.35^***	.02	.27^***	.02	1.26^***	.09
Chronic conditions	−.07^***	.01	−.11^***	.02	−.06^***	.01	−.48^***	.06

Note. PA = physical activity.

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

Cross-Lagged Panel Analyses Between PA and Purpose in Life

The results of model fit indices and chi-square difference tests of the cross-lagged models for vigorous-intensity PA, moderate-intensity PA, and light-intensity PA were presented in Tables 5, 6, and 7. Findings were similar for PA of each intensity. Specifically, Model 3 (i.e., purpose in life-to-PA model) was chosen as the final model for PA of each intensity. This was determined by comparing the model fit between different models and examining the significance of paths in each model. Model 3 showed an adequate-to-good fit to the data for vigorous-intensity PA (CFI = .950, TLI = .937, RMSEA = .025, and SRMR = .035), moderate-intensity PA (CFI = .953, TLI = .940, RMSEA = .025, and SRMR = .032), and light-intensity PA (CFI = .940, TLI = .923, RMSEA = .028, and SRMR = .037). For PA of each intensity, the model fit of Model 3 was significantly better than Model 1 (baseline model) and was not significantly different from Model 4 (full cross-lagged model). In addition, in Model 4, the paths between PA of each intensity and subsequent purpose in life were all nonsignificant.

Table 5.

Cross-Lagged Models for Vigorous-Intensity PA.

	Chi-Square	CFI	TLI	RMSEA	SRMR	Chi-Square Difference
Model 1: Baseline model	1475.8	.947	.934	.026 (.025, .028)	.040
Model 2: PA to purpose in life	1470.6	.947	.933	.026 (.025, .028)	.040	M2 vs M1: 5.2
Model 3: Purpose in life to PA	1401.2	.950	.937	.025 (.024, .027)	.035	M3 vs M1: 74.6^***
Model 4: Full cross-lagged	1396.9	.950	.937	.025 (.024, .027)	.035	M4 vs M3: 4.2

Note. PA = physical activity; CFI = Comparative Fit Index; TLI = Tucker–Lewis Index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.

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

Table 6.

Cross-Lagged Models for Moderate-Intensity PA.

	Chi-Square	CFI	TLI	RMSEA (90% CI)	SRMR	Chi-Square Difference
Model 1: Baseline model	1523.7	.944	.929	.027 (.026, .028)	.044
Model 2: PA to purpose in life	1516.1	.944	.929	.027 (.026, .028)	.044	M2 vs M1: 7.6^*
Model 3: Purpose in life to PA	1324.8	.953	.940	.025 (.023, .026)	.032	M3 vs M1: 198.9^***
Model 4: Full cross-lagged	1318.9	.953	.940	.025 (.023, .026)	.031	M4 vs M3: 5.9

Note. PA = physical activity; CFI = Comparative Fit Index; TLI = Tucker–Lewis Index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.

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

Table 7.

Cross-Lagged Models for Light-Intensity PA.

	Chi-Square	CFI	TLI	RMSEA (90% CI)	SRMR	Chi-Square Difference
Model 1: Baseline model	1873.7	.928	.909	.031 (.029, .032)	.052
Model 2: PA to purpose in life	1871.7	.928	.909	.031 (.029, .032)	.051	M2 vs M1: 2.0
Model 3: Purpose in life to PA	1627.0	.940	.923	.028 (.027, .030)	.037	M3 vs M1: 246.7^***
Model 4: Full cross-lagged	1625.6	.939	.923	.028 (.027, .030)	.037	M4 vs M3: 1.4

Note. PA = physical activity; CFI = Comparative Fit Index; TLI = Tucker–Lewis Index; RMSEA = root mean square error of approximation; SRMR = standardized root mean square residual.

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

Standardized path coefficients of the final model for vigorous-intensity PA, moderate-intensity PA, and light-intensity PA were presented in Supplementary Figures 1–3. As expected, both PA and purpose in life were positively associated with subsequent measurements of the same construct. The standardized autoregressive coefficients for PA ranged from .283 to .365 from T1 to T2 and .287 to .342 from T2 to T3 (p’s < .001). The standardized autoregressive coefficients for purpose in life ranged from .940 to .970 from T1 to T2 and .834 to .898 from T2 to T3 (p’s < .001), indicating that purpose in life was rather stable over time.

Figure 1.

Standardized coefficients of the cross-lagged model between vigorous-intensity physical activity and purpose in life.

Figure 2.

Standardized coefficients of the cross-lagged model between moderate-intensity physical activity and purpose in life.

Figure 3.

Standardized coefficients of the cross-lagged model between light-intensity physical activity and purpose in life. Note. Only significant paths are presented (p’s < .05). Controlled covariates include age, gender, race/ethnicity, marital status, education, self-rated, and chronic condition at baseline.

Cross-lagged paths from purpose in life at T1 to PA at T2 (vigorous-intensity PA: β = .164, p < .001; moderate-intensity PA: β = .240, p < .001; light-intensity PA: β = .277, p < .001) and from purpose in life at T2 to PA at T3 (vigorous-intensity PA: β = .079, p < .001; moderate-intensity PA: β = .194, p < .001; light-intensity PA: β = .205, p < .001) were all positive and significant. In contrast, cross-lagged paths from PA at T1 to purpose in life at T2 as well as from PA at T2 to purpose in life at T3 were all nonsignificant for PA of each intensity. The results indicated that purpose in life positively predicted PA at later time points, but PA did not predict purpose in life at subsequent time points.

Additionally, we conducted a sensitivity analysis to confirm the final models by only including individuals with no missing data (n = 2761). Results indicated that the model fit and significance of path coefficients were not substantially changed for PA of each intensity. Regarding the covariates, higher levels of PA and purpose in life at baseline were in general associated with being female, married, white/Caucasian, older at baseline, higher levels of education, better self-rated health, and fewer chronic conditions. Removing each of these covariates significantly reduced the overall model fit. Especially, removing self-rated health and disease burden reduced the model fit more substantially compared with other covariates. Therefore, all these covariates were kept in the final models.

Replication

The replication sample consisted of 4624 participants at baseline, with the mean age being 74.71 ± 7.06 years. 42.0% of them were males and 84.8% of them were white/Caucasian. The demographic characteristics of the replication sample were similar to that of the main sample. Using the replication sample, the results were consistent with that derived from the main sample. For PA of each intensity, Model 3 was determined as the final model. In other words, purpose in life predicted subsequent PA while PA did not predict subsequent purpose in life among the replication sample. In addition, the magnitude of the coefficients in Model 3 for the replication sample was quite comparable to that for the main sample. The path coefficients of the final model of the replication sample can be found in the Supplementary Material.

Discussion

The current study examined the longitudinal relationships between PA of different intensities and purpose in life in older adults using data collected at three time points across eight years. As expected, we found that PA of different intensities and purpose in life decreased slowly over time in older adults. These findings support several previous studies which also demonstrated an age-related decline of PA and purpose life in later life (Milanović et al., 2013; Musich et al., 2018). The current study also revealed that the strongest predictor of PA at subsequent waves was PA at previous waves, and the same was also found for purpose in life. This indicates a certain level of stability in PA and purpose in life over time and supports the control for the autoregressive effects in the models. The current study was the first study to investigate the bidirectionality between PA and purpose in life in older adults using a cross-lagged panel analysis. However, the reciprocal relationship between PA and purpose in life was not supported by the current study. While higher levels of purpose in life were associated with higher levels of subsequent vigorous-intensity PA, moderate-intensity PA, and light-intensity PA, none of the PA variables was associated with subsequent purpose in life. Comparing the strength of associations between purpose in life and PA of different intensities, it suggested that purpose in life had a stronger association with light-intensity PA than that with vigorous-intensity PA and moderate-intensity PA. This might be explained by the preferences for PA among older adults. Among many types of PA, walking, a typical form of light-intensity PA, was identified as the favorite type of PA among older adults (Amireault et al., 2018; Stathokostas & Jones, 2016). Furthermore, it appeared the unidirectional prospective relationship between purpose in life and PA was long-lasting since purpose in life at both T1 and T2 predicted future PA.

While some previous studies suggested PA was a predictor of purpose in life, none of the vigorous-intensity PA, moderate-intensity PA, and light-intensity PA predicted future purpose in life in the current study. The inconsistent findings might be explained by the difference in study designs. For example, a cross-sectional study involving 250 community-dwelling older adults with a mean age of 72.4 years found that PA frequency was significantly associated with purpose in life (Ju, 2017). A similar positive association between PA and purpose in life was also found in another cross-sectional study involving 2414 older adults with loneliness (Kim et al., 2017). Indeed, it might be true that PA had a cross-sectional relationship with purpose in life in older adults, as indicated by the significantly positive bivariate correlations between PA and purpose in life in the current study. However, when the longitudinal data were used and the autoregressive effects were controlled in the current study, PA was no longer a predictor of future purpose in life. This suggested that the cross-sectional relationship observed might predominantly be caused by the influence of purpose in life on PA rather than the opposite. So far, there was only one randomized controlled trial examining the effect of exercises on purpose in life which involved adult women with a mean age of 27.3 years (Delextrat et al., 2016). This study found that an 8-week exercise intervention based on Zumba was effective in improving purpose in life (Delextrat et al., 2016). The difference between PA increased by interventions and naturally occurring changes in PA might also be a possible explanation, and more studies examining the impact of exercise interventions on purpose in life in older adults are warranted in the future.

The observed associations in the current study between purpose in life and subsequent PA were consistent with findings from previous research. For example, a cross-sectional study that purpose in life was a significant predictor of PA in older adults aged 65–84 years after controlling for self-rated health, depressive symptoms, and education (Ruuskanen & Ruoppila, 1995). Such association was also found in two other cross-sectional studies involving healthy older adults (Holahan & Suzuki, 2006) and older cardiac patients (Holahan et al., 2008). The current study extends prior knowledge by providing more definitive evidence of a longitudinal association between purpose in life and PA in older adults. The prospective association between purpose in life and PA was found in other two longitudinal studies involving adolescents (Brassai et al., 2015) and adults (Hooker & Masters, 2016), indicating that purpose in life might be a consistent predictor of PA across the lifespan.

Although the current study did not explore the mechanisms underlying the relationship between purpose in life and PA, some theories may shed light on how purpose in life influences future PA. Ryff and Singer (1998) proposed a key theory that individuals with higher levels of purpose in life are more likely to pay attention to their health and thus engage in positive health behaviors as better health might enhance their ability to pursue their goals. This theory was supported in later studies (Kim et al., 2020; Wiesmann & Hannich, 2011) showing that life was associated with multiple health behaviors in older adults including PA. In this sense, purpose in life could be viewed as a source of motivation to participate in and maintain PA (Hooker & Masters, 2016). Social Cognitive Theory (Bandura, 1977), which highlights the importance of self-efficacy in behavioral change, may also help to explain the association between purpose in life and PA. It has been demonstrated that self-efficacy was a direct predictor of participation in PA (Ashford et al., 2010) and purpose in life was positively associated with self-efficacy (DeWitz et al., 2009). A recent study provided strong support for this hypothesis showing that the positive relationship between purpose in life and PA in middle-aged adults was fully mediated by self-efficacy (Rush et al., 2019). Nevertheless, the mechanisms linking purpose in life to PA still remain largely unexplored and await further elucidation.

The findings from the current study have both theoretical and practical implications for health promotion in the older population. As the link between PA and better health is well-documented in older adults, efforts to encourage their pursuit of a meaningful life may result in health benefits through active engagement of PA. Several recent interventions using meaning-centered therapy (Breitbart et al., 2012) and cognitive behavioral therapy (Ruini & Fava, 2012) have shown to be promising in raising the sense of purpose in life. In addition, such purpose-in-life enhancement interventions could be incorporated into PA interventions to improve intervention outcomes.

Several limitations of the current study should be considered. First, the current study was limited by the self-reported measure of PA. Moreover, PA of each intensity was measured using a single item assessing the frequency, which did not reflect other parameters of PA, such as duration. The single-item measures may also lead to the relatively low stability in PA over time, and thus, the results must be interpreted with caution. However, using single-item measures of PA in the HRS enabled us to examine the associations between purpose in life and PA with different intensities. Future assessment of PA in the HRS may consider adding objective measures such as accelerometers or using more detailed self-reported questionnaires, although doing so may increase the length and cost of the survey. Second, although the cross-lagged panel analysis provided useful information about prospective relationships between PA and purpose in life over time, it should be kept in mind that the conclusive causal relationship between them could not be inferred due to the observational nature of the study. Third, the missing cases at T2 and T3 might cause potential bias of the findings as the independent t-tests revealed significant differences in several baseline characteristics between participants who had complete data and those who had incomplete data across the three waves. Nevertheless, we did not delete any cases and used the full information maximum likelihood estimation method to minimize the potential bias. The sensitivity analysis involving only individuals without missing data also did not change the results. Another limitation with the missing data was that participants without available data of PA and purpose in life at T1 were excluded in the current study. It was possible that participants who did not complete or return the psychosocial questionnaire may have different (possibly lower) levels of purpose in life at baseline compared with the analytic sample included in our study. Fourth, the current study excluded participants who were younger than 65 years at baseline. Examining the association between PA and purpose in life among middle adulthood may help us understand the changes in the association across the transition to retirement. Fifth, purpose in life is only one component of eudaimonic well-being, and thus, the relationship between PA and purpose in life may not be generalizable to other components of eudaimonic well-being. Last, although a wide range of covariates were controlled for in the current study, some other variables that may potentially influence the results were not included such as cognitive function and pain, which could be addressed in future studies.

Despite these limitations, the current study has several considerable strengths. To our knowledge, the current study is the first study to assess the reciprocal relationship between PA and purpose in life, both of which have important health implications for the older population. The use of cross-lagged panel analysis allowed simultaneous investigation of each direction of the associations between PA and purpose in life while controlling their autoregressive effects. In addition, the longitudinal study design with three waves across 8 years enabled us to study changes in PA and purpose in life over a long period. A further strength was the large sample size of the HRS, which assured adequate statistical power to detect the associations between PA and purpose in life over time. Moreover, a wide range of covariates were included in the current study to minimize potential confounding effects on the relationship between PA and purpose in life in older adults.

Conclusion

In conclusion, the findings of the current study suggested a nonreciprocal relationship between PA and purpose in life in older adults. Higher levels of purpose in life predict greater future engagement in vigorous-intensity PA, moderate-intensity PA, and light-intensity PA, but not vice versa. Future research is needed to replicate these findings using objective measures of PA and disentangle the mechanisms linking purpose in life to PA in older adults. Interventions to enhance the sense of purpose in life may improve older adults’ health through its positive influence on PA.

Supplemental Material

sj-pdf-1-jah-10.1177_08982643211019508 – Supplemental Material for Longitudinal Associations Between Physical Activity and Purpose in Life Among Older Adults: A Cross-Lagged Panel Analysis

Supplemental Material, sj-pdf-1-jah-10.1177_08982643211019508 for Longitudinal Associations Between Physical Activity and Purpose in Life Among Older Adults: A Cross-Lagged Panel Analysis by Zhanjia Zhang and Weiyun Chen in Journal of Aging and Health

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Zhanjia Zhang

Supplemental Material

Supplemental material for this article is available online.

References

Amireault

Baier

J. M.

Spencer

J. R.

(2018). Physical activity preferences among older adults: A systematic review. Journal of Aging and Physical Activity, 27(1), 128-139.

Arbuckle

J. L.

(1996). Full information estimation in the presence of incomplete data. In G. A. Marcoulides & R. E. Schumacker (Eds.), Advanced structural equation modeling: Issues and techniques (pp. 243-277). Lawrence Erlbaum.

Ashford

Edmunds

French

D. P.

(2010). What is the best way to change self-efficacy to promote lifestyle and recreational physical activity? A systematic review with meta-analysis. British Journal of Health Psychology, 15(2), 265-288. doi:10.1348/135910709X461752

Bandura

(1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84(2), 191-215.

Bollen

K. A.

(1989). A new incremental fit index for general structural equation models. Sociological Methods & Research, 17(3), 303-316.

Boyle

P. A.

Buchman

A. S.

Barnes

L. L.

Bennett

D. A.

(2010). Effect of a purpose in life on risk of incident Alzheimer disease and mild cognitive impairment in community-dwelling older persons. Archives of General Psychiatry, 67(3), 304-310.

Brassai

Piko

B. F.

Steger

M. F.

(2015). A reason to stay healthy: The role of meaning in life in relation to physical activity and healthy eating among adolescents. Journal of Health Psychology, 20(5), 473-482.

Breitbart

Poppito

Rosenfeld

Vickers

A. J.

Abbey

Olden

Pessin

Lichtenthal

Sjoberg

Cassileth

B. R.

(2012). Pilot randomized controlled trial of individual meaning-centered psychotherapy for patients with advanced cancer. Journal of Clinical Oncology, 30(12), 1304-1309.

Brown

T. A.

(2015). Confirmatory factor analysis for applied research. Guilford publications.

10.

Chen

D. S.

Betz

Yaffe

Ayonayon

H. N.

Kritchevsky

Martin

K. R.

Martin

K. R.

Harris

T. B.

Purchase-Helzner

Satterfield

Xue

Q.-L.

Pratt

Simonsick

E. M.

Lin

F. R.

(2015). Association of hearing impairment with declines in physical functioning and the risk of disability in older adults. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences, 70(5), 654-661.

11.

Choi

McDonough

I. M.

Kim

(2018). The association between the number of chronic health conditions and advance care planning varies by race/ethnicity. Aging & Mental Health, 24(3), 453-463.

12.

Crosswell

A. D.

Suresh

Puterman

Gruenewald

T. L.

Lee

Epel

E. S.

(2018). Advancing research on psychosocial stress and aging with the health and retirement study: Looking back to launch the field forward. The Journals of Gerontology: Series B, 75(5), 970-980.

13.

Delextrat

A. A.

Warner

Graham

Neupert

(2016). An 8-week exercise intervention based on zumba improves aerobic fitness and psychological well-being in healthy women. Journal of Physical Activity & Health, 13(2), 131-139.

14.

DeWitz

S. J.

Woolsey

M. L.

Walsh

W. B.

(2009). College student retention: An exploration of the relationship between self-efficacy beliefs and purpose in life among college students. Journal of College Student Development, 50(1), 19-34.

15.

Diener

(1984). Subjective well-being. Psychological Bulletin, 95(3), 542-575.

16.

Friedman

E. M.

Hayney

Love

G. D.

Singer

B. H.

Ryff

C. D.

(2007). Plasma interleukin-6 and soluble IL-6 receptors are associated with psychological well-being in aging women. Health Psychology, 26(3), 305-313.

17.

Holahan

C. K.

Suzuki

(2006). Motivational factors in health promoting behavior in later aging. Activities, Adaptation & Aging, 30(1), 47-60.

18.

Holahan

C. K.

Holahan

C. J.

Suzuki

(2008). Purposiveness, physical activity, and perceived health in cardiac patients. Disability and Rehabilitation, 30(23), 1772-1778.

19.

Holahan

C. K.

Holahan

C. J.

Velasquez

K. E.

Jung

North

R. J.

Pahl

S. A.

(2011). Purposiveness and leisure-time physical activity in women in early midlife. Women & Health, 51(7), 661-675.

20.

Homan

K. J.

(2016). Self-compassion and psychological well-being in older adults. Journal of Adult Development, 23(2), 111-119.

21.

Homan

K. J.

Boyatzis

C. J.

(2010). Religiosity, sense of meaning, and health behavior in older adults. International Journal for the Psychology of Religion, 20(3), 173-186.

22.

Hooker

S. A.

Masters

K. S.

(2016). Purpose in life is associated with physical activity measured by accelerometer. Journal of Health Psychology, 21(6), 962-971.

23.

Jacobs

T. L.

Epel

E. S.

Lin

Blackburn

E. H.

Wolkowitz

O. M.

Bridwell

D. A.

Zanesco

A. P.

Aichele

S. R.

Sahdra

B. K.

MacLean

K. A.

King

B. G.

Shaver

P. R.

Rosenberg

E. L.

Ferrer

Wallace

B. A.

Saron

C. D.

(2011). Intensive meditation training, immune cell telomerase activity, and psychological mediators. Psychoneuroendocrinology, 36(5), 664-681.

24.

Johnson

B. T.

Acabchuk

R. L.

(2018). What are the keys to a longer, happier life? Answers from five decades of health psychology research. Social Science & Medicine, 196, 218-226.

25.

(2017). The relationship between physical activity, meaning in life, and subjective vitality in community-dwelling older adults. Archives of Gerontology and Geriatrics, 73, 120-124.

26.

Kearney

M. W.

(2017). The SAGE encyclopedia of communication research methods. SAGE Publications, Inc.

27.

Kenny

D. A.

(1975). Cross-lagged panel correlation: A test for spuriousness. Psychological Bulletin, 82(6), 887-903.

28.

Kim

E. S.

Sun

J. K.

Park

Peterson

(2013). Purpose in life and reduced incidence of stroke in older adults: 'The health and retirement study'. Journal of Psychosomatic Research, 74(5), 427-432.

29.

Kim

Lee

Chun

Han

Heo

(2017). The effects of leisure-time physical activity for optimism, life satisfaction, psychological well-being, and positive affect among older adults with loneliness. Annals of Leisure Research, 20(4), 406-415.

30.

Kim

E. S.

Shiba

Boehm

J. K.

Kubzansky

L. D.

(2020). Sense of purpose in life and five health behaviors in older adults. Preventive Medicine, 139, 106172.

31.

Kline

R. B.

(2015). Principles and practice of structural equation modeling. Guilford Publications.

32.

Leonhardt

Lehr

Chenot

Keller

Luckmann

Basler

Baum

Donner-Banzhoff

Pfingsten

Hildebrandt

Kochen

Becker

(2009). Are fear-avoidance beliefs in low back pain patients a risk factor for low physical activity or vice versa? A cross-lagged panel analysis. Psycho-Social Medicine, 6, Doc01.

33.

Mack

D. E.

Wilson

P. M.

Gunnell

K. E.

Gilchrist

J. D.

Kowalski

K. C.

Crocker

P. R. E.

(2012). Health-enhancing physical activity: Associations with markers of well-being. Applied Psychology: Health and Well-Being, 4(2), 127-150.

34.

McMahan

E. A.

Estes

(2011). Hedonic versus eudaimonic conceptions of well-being: Evidence of differential associations with self-reported well-being. Social Indicators Research, 103(1), 93-108.

35.

Meeks

T. W.

Vahia

I. V.

Lavretsky

Kulkarni

Jeste

D. V.

(2011). A tune in “a minor” can “b major”: a review of epidemiology, illness course, and public health implications of subthreshold depression in older adults. Journal of Affective Disorders, 129(1-3), 126-142.

36.

Milanovic

Jorgić

Trajković

Sporiš

Pantelić

James

(2013). Age-related decrease in physical activity and functional fitness among elderly men and women. Clinical Interventions in Aging, 8, 549-556.

37.

Milton

Bull

F. C.

Bauman

(2011). Reliability and validity testing of a single-item physical activity measure. British Journal of Sports Medicine, 45(3), 203-208.

38.

Milton

Clemes

Bull

(2013). Can a single question provide an accurate measure of physical activity? British Journal of Sports Medicine, 47(1), 44-48.

39.

Musich

Wang

S. S.

Kraemer

Hawkins

Wicker

(2018). Purpose in life and positive health outcomes among older adults. Population Health Management, 21(2), 139-147.

40.

Muthén

(2019). The comprehensive modelling program for applied researchers: user’s guide. Muthén & Muthén.

41.

Pinquart

(2002). Creating and maintaining purpose in life in old age: A meta-analysis. Ageing International, 27(2), 90-114.

42.

Roepke

A. M.

Jayawickreme

Riffle

O. M.

(2014). Meaning and health: A systematic review. Applied Research in Quality of Life, 9(4), 1055-1079.

43.

Ruini

Fava

G. A.

(2012). Role of well-being therapy in achieving a balanced and individualized path to optimal functioning. Clinical Psychology & Psychotherapy, 19(4), 291-304.

44.

Rush

Hooker

Ross

Frers

Peters

Masters

(2019). Brief report: Meaning in life is mediated by self-efficacy in the prediction of physical activity. Journal of Health Psychology, 26(5), 753-757.

45.

Ruuskanen

J. M.

Ruoppila

(1995). Physical activity and psychological well-being among people aged 65 to 84 years. Age and Ageing, 24(4), 292-296.

46.

Ryan

R. M.

Deci

E. L.

(2001). On happiness and human potentials: A review of research on hedonic and eudaimonic well-being. Annual Review of Psychology, 52(1), 141-166.

47.

Ryan

R. M.

Huta

Deci

E. L.

(2008). Living well: A self-determination theory perspective on eudaimonia. Journal of Happiness Studies, 9(1), 139-170.

48.

Ryff

C. D.

(1989). Happiness is everything, or is it? Explorations on the meaning of psychological well-being. Journal of Personality and Social Psychology, 57(6), 1069-1081.

49.

Ryff

C. D.

(2014). Psychological well-being revisited: Advances in the science and practice of eudaimonia. Psychotherapy and Psychosomatics, 83(1), 10-28.

50.

Ryff

C. D.

Boylan

J. M.

(2016). Linking happiness to health: Comparisons between hedonic and eudaimonic well-being. In L. Bruni & P. L. Porta (Eds.), Handbook of research methods and applications in happiness and quality of life. Edward Elgar Publishing.

51.

Ryff

C. D.

Singer

(1998). The contours of positive human health. Psychological Inquiry, 9(1), 1-28.

52.

Schmidt

C. K.

Raque-Bogdan

T. L.

Piontkowski

Schaefer

K. L.

(2011). Putting the positive in health psychology: A content analysis of three journals. Journal of Health Psychology, 16(4), 607-620.

53.

Seligman

M. E. P.

Csikszentmihalyi

(2000). Positive psychology: An introduction. American Psychologist, 55(1), 5-14.

54.

Smith

Marshall

Huang

(2005). Screening for physical activity in family practice: Evaluation of two brief assessment tools. American Journal of Preventive Medicine, 29(4), 256-264.

55.

Sonnega

Smith

(2015). Health and retirement study: A longitudinal data resource for psychologists. In N. Pachana (Ed.), Encyclopedia of geropsychology (pp. 1-13). Springer.

56.

Sonnega

Faul

J. D.

Ofstedal

M. B.

Langa

K. M.

Phillips

J. W.

Weir

D. R.

(2014). Cohort profile: The health and retirement study (HRS). International Journal of Epidemiology, 43(2), 576-585.

57.

Stathokostas

Jones

G. R.

(2016). Exercise modality choices one year after intervention in previously inactive older men and women. Journal of Aging and Physical Activity, 24(3), 435-443. doi:10.1123/japa.2015-0167

58.

Steger

M. F.

Frazier

Oishi

Kaler

(2006). The meaning in life questionnaire: Assessing the presence of and search for meaning in life. Journal of Counseling Psychology, 53(1), 80-93.

59.

Vella-Brodrick

D. A.

Stanley

(2013). The significance of transport mobility in predicting well-being. Transport Policy, 29, 236-242.

60.

Wiesmann

Hannich

H.-J.

(2011). Salutogenic perspectives on health maintenance. GeroPsych, 24(3), 127-135.

61.

World Health Organization (2014). Mental health: A state of well-being. Retrieved from http://www.who.int/features/factfiles/mental_health/en/.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.15 MB