Introduction
What predicts aging well? According to the successful aging model proposed by Rowe and Kahn (Reference Rowe and Kahn1998), aging well requires physical and mental health, active engagement in society through volunteer or paid work, and social involvement. Although empirically supported (George, Reference George2010), this model has been criticized for neglecting the role of spirituality (Crowther et al., Reference Crowther, Parker, Achenbaum, Larimore and Koenig2002) and the life course (Stowe and Cooney, Reference Stowe and Cooney2015), for defining successful aging as simply extending middle age into old age, and for calling people unsuccessful agers if they suffer from ill health or physical disability (Dillaway and Byrnes, Reference Dillaway and Byrnes2009).
By contrast, social stratification theory and cumulative advantage and disadvantage theory explain aging well through socioeconomic life course factors and social differences early in life that place individuals on divergent trajectories with increasing differential allocation of resources and assets, so that the rich get richer and the poor get poorer with respective consequences for physical health and well-being (George, Reference George2010; O’Rand et al., Reference O’Rand, Binstock, George, Binstock, George, Cutler, Hendricks and Schulz2006). For example, socioeconomic status in childhood has long-term positive effects on later life health and adjustment to aging (Landes et al., Reference Landes, Ardelt, Vaillant and Waldinger2014). Similarly, whites tend to report greater subjective well-being than non-whites, explained at least partially by differences in socioeconomic status and health (Barger et al., Reference Barger, Donoho and Wayment2009).
However, all these models ignore the role of internal strengths in aging well. Although objective life conditions, such as physical health, socioeconomic status, and demographics, clearly contribute to subjective well-being (George, Reference George2010), internal strengths of older adults—such as wisdom, religiosity, mastery, and purpose in life—seem equally important (Ardelt et al., Reference Ardelt, Landes, Gerlach, Fox and Sinnott2013), particularly when physical health declines (Ardelt and Edwards, Reference Ardelt and Edwards2016; Montross-Thomas et al., Reference Montross-Thomas, Joseph, Edmonds, Palinkas and Jestein press; Peteet et al., Reference Peteet, Zaben and Koenigin press). Because wisdom and religiosity tend to be uncorrelated (Ardelt and Wingard, in press; Le, Reference Le2008; Wink and Dillon, Reference Wink and Dillon2003), they might be alternative paths toward subjective well-being in old age. Whereas a sense of mastery and purpose in life might at least partially explain the relation between wisdom and aging well (Ardelt and Edwards, Reference Ardelt and Edwards2016; Etezadi and Pushkar, Reference Etezadi and Pushkar2013), purpose in life might mediate the association between religiosity and well-being in old age (Krause, Reference Krause2003; Peteet et al., Reference Peteet, Zaben and Koenigin press; Steger and Frazier, Reference Steger and Frazier2005).
The goal of the current study was to examine whether wisdom and religiosity provided two alternative pathways to subjective well-being (SWB) among older and also young adults from Canada and the United States, and whether mastery and purpose in life mediated these associations. We also explored whether age cohort and/or nation of residency moderated any of the effects. Confidence in the generalizability and reliability of the results for North America improves if the same associations are present for both young and older adults from Canada and the United States.
Definition of wisdom, religiosity, and subjective well-being
While there is no single, unified definition of wisdom, many wisdom researchers and lay people conceptualize wisdom as uniting cognitive, reflective, and emotional components (Bangen et al., Reference Bangen, Meeks and Jeste2013; Bluck and Glück, Reference Bluck, Glück, Sternberg and Jordan2005; Staudinger and Glück, Reference Staudinger and Glück2011), although emphasis on individual components varies. In this study, we used the Three-Dimensional Wisdom Model (3D-WM) to define and operationalize personal wisdom as an integration of cognitive, reflective, and compassionate (affective) dimensions (Ardelt, Reference Ardelt2003). The cognitive dimension refers to a desire for deep and profound knowledge and insight about the intrapersonal and interpersonal aspects of life (including the meaning of life and death) and an awareness and acceptance of the complexity of human nature, the inherent limits of human knowledge, and of life’s uncertainty and unpredictability. Reaching such an understanding requires the reflective dimension, defined as perceiving events and phenomena (including oneself) from multiple perspectives, thereby transcending ego-centric subjectivity and projection. Reflection and self-reflection allow for deep insight (including self-insight) that weaken ego-centric self-centeredness and increase understanding, tolerance, and empathy. This process leads to greater compassion and sympathy toward oneself and others and the motivation to foster universal well-being, which describes the compassionate dimension.
Religiosity in this study was self-identified. Unlike spirituality, which can refer to non-institutionalized or non-tradition-centered spiritual beliefs and practices, religiosity refers to how strongly individuals identify with an established religious faith (Wink and Dillon, Reference Wink and Dillon2003).
Subjective well-being in this study combines cognitive evaluation of one’s whole life—expressed through life satisfaction (Diener et al., Reference Diener, Emmons, Larsen and Griffin1985)—and positive affective states, such as cheerfulness and happiness (George, Reference George2010).
Pathways leading from wisdom and religiosity to subjective well-being
While personal wisdom tends to be moderately related to greater SWB at all ages (Ardelt, in press), the association between religiosity and well-being tends to be weaker (Ardelt and Wingard, in press; Green and Elliott, Reference Green and Elliott2010; Witter et al., Reference Witter, Stock and Okun1985). Some individuals might only turn to religion when in distress (Ai et al., Reference Ai, Peterson, Bolling and Rodgers2006; Ellison and Taylor, Reference Ellison and Taylor1996), whereas others find strength, contentment, and happiness through their religion (Koenig et al., Reference Koenig, Kvale and Ferrel1988; Lewis and Cruise, Reference Lewis and Cruise2006; Peteet et al., Reference Peteet, Zaben and Koenigin press). If so, the positive effect of religiosity on well-being that some people experience and the increased religiosity of some individuals who suffer acute distress might cancel each other out.
The mechanisms that link wisdom and religiosity to well-being might also differ. The association between wisdom and well-being appears to be mediated by indicators of psychological well-being, such as mastery and purpose in life. Wise individuals who perceive phenomena and events from multiple perspectives learn from their mistakes and life experiences, accept reality as it is, and have developed the necessary equanimity and confidence to master life’s vicissitudes (Ardelt, Reference Ardelt2005). Because wise persons have attained a deep and compassionate understanding of the intrapersonal and interpersonal aspects of life, they typically comprehend the deeper meaning and purpose of life (Levenson et al., Reference Levenson, Aldwin, Cupertino and Neri2001). Indeed, wisdom tends to be moderately to strongly correlated with greater mastery and purpose in life (Ardelt, in press), and both mastery and life purpose tend to be moderately to strongly associated with greater SWB (Ryff and Keyes, Reference Ryff and Keyes1995; Steger et al., Reference Steger, Kawabata, Shimai and Otake2008). Hence, mastery and purpose in life are likely to at least partially mediate the association between wisdom and well-being, as documented in previous studies of older adults (Ardelt and Edwards, Reference Ardelt and Edwards2016; Etezadi and Pushkar, Reference Etezadi and Pushkar2013).
By contrast, the association between religiosity and well-being appears to be mediated by purpose in life (Krause, Reference Krause2003; Steger and Frazier, Reference Steger and Frazier2005). That religiosity is related to a sense of purpose in life is not surprising, since religion has been described as a meaning-making system that creates “an all-embracing sacred order” (Berger, Reference Berger1969) and provides meaning even when confronted with suffering, illness, and death (Ardelt et al., Reference Ardelt, Ai and Eichenberger2008; Peteet et al., Reference Peteet, Zaben and Koenigin press). Religiosity also gives meaning to people’s lives if they feel that they are following their religious “calling” (Koenig, Reference Koenig, Wong, Wong, McDonald and Klaassen2007).
Like the association between religiosity and well-being, however, the relation between religiosity and mastery might either be positive or negative. Some individuals might feel a greater sense of mastery through their religious faith and connection to a higher power, while others might concede control to the will of God or might try to find solace in religion if they feel that their life is out of control (Ardelt et al., Reference Ardelt, Ai and Eichenberger2008; Schieman et al., Reference Schieman, Nguyen and Elliot2003). These opposing effects might explain why the empirical evidence is mixed and the association tends to be weak. The correlation between religiosity and mastery has been found to be weakly positive (Jackson and Bergeman, Reference Jackson and Bergeman2011), weakly negative (Schieman et al., Reference Schieman, Nguyen and Elliot2003), or non-significant (Ardelt and Wingard, in press). Hence, it is unclear whether mastery mediates the relation between religiosity and well-being.
The present study
Based on the literature review and prior research, this study tested the mediated path model depicted in Figure 1 and explored moderating effects by age cohort and nation of residency. The dashed arrows in Figure 1 signify possibly insignificant or weak associations. As in prior studies of older adults (Ardelt and Edwards, Reference Ardelt and Edwards2016; Etezadi and Pushkar, Reference Etezadi and Pushkar2013), the relationship between wisdom and SWB was predicted to be at least partially mediated by mastery and purpose in life. Research has also shown that mastery and purpose in life are moderately to strongly correlated (Ryff and Keyes, Reference Ryff and Keyes1995; Scheier et al., Reference Scheier2006). Mastery was expected to have a positive impact on purpose in life, because a sense of purpose and meaning in life might depend on feeling in control over one’s destiny to follow one’s calling, whereas a loss of control might be accompanied by a loss of life meaning (Ardelt and Edwards, Reference Ardelt and Edwards2016; Bierman et al., Reference Bierman, Fazio and Milkie2006). The association between religiosity and SWB was assumed to be primarily mediated by purpose in life as documented in prior studies (Krause, Reference Krause2003; Steger and Frazier, Reference Steger and Frazier2005). Yet no prior study has simultaneously analyzed the paths from wisdom and religiosity to well-being mediated by mastery and purpose in life and compared whether the relations are significant and equally strong for young and older adults alike.
Figure 1. Theoretical mediated path model.
Finally, we also tested whether the relations of education and race on SWB were moderated by age cohort and nation of residency. Based on the theory of cumulative advantage and disadvantage (George, Reference George2010), and the fact that the United States is a more individualistic society than Canada, which provides better access to community resources such as health care and welfare (Lasser et al., Reference Lasser, Himmelstein and Woolhandler2006), we expected a stronger effect of education and white race on SWB among older adults in the U.S.A. than among older adults in Canada or young adults.
To summarize, the following hypotheses were tested:
Hypothesis 1: Wisdom and religiosity are significantly related to greater SWB in young adulthood and old age, even after controlling for subjective health and demographics.
Hypothesis 2: The relation between wisdom and SWB is at least partially mediated by a sense of mastery and purpose in life. Mastery mediates the association between wisdom and SWB both directly and indirectly through a positive relation with purpose in life.
Hypothesis 3: The association between religiosity and SWB is primarily mediated by a sense of purpose in life.
Hypothesis 4: Education and race (white vs. non-white) have stronger effects on SWB among older adults in the U.S.A. than among young adults or older adults in Canada.
Methods
Sample
The sample comprised 211 respondents: 60 older adults from Canada (age range 62–99 years, M = 80.68, SD = 9.08), 51 older adults from the U.S.A. (age range 64–90 years, M = 73.10, SD = 6.97), 50 young adults from Canada (age range 21–30 years, M = 24.78, SD = 2.72), and 50 young adults from the U.S.A. (age range 21–30 years, M = 23.33, SD = 2.48). Six young adults from Canada, 20 older adults from Canada, and three older adults from the U.S.A. were foreign born.
Procedure
Between April 2008 and June 2011, a quota sampling technique was used to obtain a sample with approximately equal representation by age group, nation, and gender. Young adults were primarily recruited at a university campus in Canada and the U.S.A. and older adults were recruited through social groups or personal connections of the research team. Interviews were conducted face-to-face in participants’ homes or at a place of their choice, and all respondents gave informed consent. A semi-structured interview about participants’ most important personal memories and their descriptions of wisdom exemplars and experiences related to wisdom was conducted first, followed by structured survey questions. However, the present study only used the quantitative survey data of the interview portion. The qualitative interviews lasted between 14 and 150 minutes (M = 55, SD = 30, median = 45) with another 10–20 minutes for the quantitative part of the interview.
Measures
Subjective well-being was assessed by six items of the General Well-Being Schedule (Fazio, Reference Fazio1977) and the Satisfaction with Life Scale (Diener et al., Reference Diener, Emmons, Larsen and Griffin1985). The six items of the General Well-Being Schedule consisted of two items of the life satisfaction subscale (e.g., How happy, satisfied, or pleased have you been with your personal life during the past month?; 1 = very dissatisfied and 6 = extremely happy) and four items of the cheerfulness subscale (e.g., On a scale from 0 to 10, how DEPRESSED ( = 0) or CHEERFUL ( = 10) have you been during the past month?). The scales were converted into 0–5 scales before the average was taken. Cronbach’s alpha was .82 for young adults and .84 for older adults. The average of the 5-item Satisfaction with Life Scale (e.g., In most ways my life is close to my ideal; 1 = strongly disagree and 7 = strongly agree) was also converted into a 0–5 scale. Cronbach’s alpha was .84 for young adults and .83 for older adults. SWB was the average of the two subscales, which correlated highly (r = .66, p < .001 for young adults and r = .68, p < .001 for older adults).
Wisdom was measured by the cognitive, reflective, and compassionate (affective) dimensions of the 39-item Three-Dimensional Wisdom Scale (3D-WS) on 5-point scales (1 = strongly agree and 5 = strongly disagree or 1 = definitely true of myself and 5 = not true of myself) scored in the direction of greater wisdom (Ardelt, Reference Ardelt2003). The 14 items of the cognitive dimension (e.g., Ignorance is bliss—reversed) measure an understanding of life or the desire to know the truth. The 12 items of the reflective dimension (e.g., I always try to look at all sides of a problem) capture the ability to look at phenomena and events from multiple perspectives and to overcome subjectivity and projections. The 13 items of the compassionate dimension (e.g., Sometimes I feel a real compassion for everyone) gauge sympathy and compassion for others, the motivation to nurture others’ well-being, and the absence of indifferent or negative emotions and behavior toward others. All items were averaged for each dimension, resulting in Cronbach’s α for the cognitive, reflective, and compassionate wisdom dimensions, respectively, of .67, .70, and .68 and for young adults and .82, .77, and .72 for older adults. Cronbach’s α for composite three-dimensional wisdom, computed as the average of the three dimensions was .57 for young adults and .75 for older adults. Although Cronbach’s α for combined wisdom was somewhat lower than desirable for young adults, given that its calculation was based on the three dimensions, it is still acceptable. (Cronbach’s α was .79 for young adults and .89 for older adults for the 39 items.)
Religiosity was measured by asking respondents how religious they were on a scale ranging from 0 (not at all religious) to 10 (very religious).
Mastery was assessed as the average of Pearlin and Schooler’s (Reference Pearlin and Schooler1978) 7-item Mastery Scale (e.g., What happens to me in the future mostly depends on me). Purpose in life was the average of three “pure” purpose and meaning items (e.g., I have discovered satisfying goals and a clear purpose in life) from Crumbaugh and Maholick’s (Reference Crumbaugh and Maholick1964) Purpose in Life Test. All items were assessed on 5-point scales (1 = strongly agree and 5 = strongly disagree) and scored in the direction of greater mastery or purpose in life before the average was taken. Cronbach’s alpha for mastery and purpose in life, respectively, was .70 and .74 for young adults and .76 and .77 for older adults.
Control variables were subjective health, education, race (0 = non-white, 1 = white), gender (0 = male, 1 = female), and nation of residency (0 = United States, 1 = Canada). Subjective health was the average of one item that asked respondent to rate their overall health at the present time (1 = excellent and 5 = very bad) and two items from the General Well-Being Schedule (Fazio, Reference Fazio1977) that assessed health concern, worries, or distress (e.g., How concerned or worried about your health have you been?) on an 11-point scale (0 = not concerned at all and 10 = very concerned) and a 6-point scale (1 = all the time and 6 = none of the time). All scales were transformed into 1–5 scales and scored in the direction of greater subjective health. Cronbach’s alpha was .73 for young adults and .84 for older adults. Education was measured by educational degree (1 = no high school, 2 = High school degree or equivalent, 3 = Some college, 4 = Bachelor’s Degree or equivalent, 5 = Master’s Degree or equivalent, and 6 = Doctoral degree or equivalent).
To examine whether nation of residency moderated the effects, interaction terms between all independent variables and nation were created after centering all continuous variables at the mean to reduce the effect of multicollinearity (Aiken and West, Reference Aiken and West1991).
Analytic strategy
As a first step, we examined how similar or different the four subsamples were regarding the study variables through two-way analyses of variance (ANOVA) with nation of residency and age group as the between-subjects factors. Then we analyzed the bivariate correlations between all study variables separately for each age group, using Pearson’s r. Preliminary regression analyses were conducted to check for outliers and whether any control variables or additional interaction effects between nation and the independent variables should be included in the model, besides the hypothesized interaction effects between nation and education or nation and race on SWB.
Because four variables had between one and eight missing values, we used the Markov Chain Monte Carlo (MCMC) multiple imputation method in PRELIS to impute missing values. The imputed data set was then divided into a group of 100 young adults and 111 older adults to compare whether the associations between the variables differed significantly by age group. Because the variables did not follow a multivariate normal distribution, covariance and asymptotic covariance matrices were computed for each group to estimate corrected standard errors of the coefficient estimates and corrected chi-square statistics. However, due to the relatively small sample size, robust maximum likelihood (ML) estimation (rather than weighted least squares estimation) was used to obtain the coefficient estimates (Jöreskog et al., Reference Jöreskog, Sörbom, du Toit and du Toit1999).
Nested multi-group path models were compared and analyzed in LISREL 9.3 to find the best fitting parsimonious model, starting with equality of form. This unconstrained model allows coefficients estimates to vary freely between the two groups. The next model tested the equality of all direct effects, with the covariance matrix between the independent variables and the error variances of the dependent variables allowed to vary freely between the two groups. The chi-square difference test was used to determine whether the difference in fit between the two models was statistically significant. If the chi-square difference test between the unconstrained and the more constrained model is non-significant, there is no evidence that the two models differ in model fit and the more parsimonious (constrained) model is preferred. Modification indices were consulted to check whether specific coefficient estimates significantly varied between the two age groups. The coefficient estimates were significantly different for young and older adults if the model that allowed coefficient estimates to vary between the two groups produced a significantly better model fit, as shown by a significant chi-square difference test between the constrained and less constrained model.
Overall model fit was determined by a non-significant chi-square test, indicating that relative to the sample size, the difference between the estimated covariance matrix and the sample covariance matrix was small enough to be non-significant, and overall fit indices (Kline, Reference Kline2005). For small samples with ML estimation, Hu and Bentler (Reference Hu and Bentler1998) recommended the standardized root mean square residual (SRMR) and the comparative fit index (CFI) as overall fit indices. Values higher than .95 for CFI and lower than .08 for SRMR signify a good model fit (Hu and Bentler, Reference Hu and Bentler1999; Kline, Reference Kline2005).
Results
Comparisons of means by age group and nation and bivariate correlations
Table S1 (see supplementary material) reveals that the four nation and age groups differed statistically for all study variables, except gender distribution, with significant interaction effects for nation and age group, except for religiosity and purpose in life.
The bivariate correlations between all study variables are shown in Table 1 for young adults (above the diagonal) and older adults (below the diagonal) separately. As predicted by Hypothesis 1, wisdom was positively correlated with SWB for both age groups, yet contrary to Hypothesis 1, religiosity was uncorrelated to well-being. In both age groups, SWB and wisdom were associated with greater mastery and purpose in life, and mastery was positively related to purpose in life, as stated in Hypothesis 2. However, religiosity was unrelated to these variables, seemingly rejecting Hypothesis 3. In opposition to Hypothesis 4, education was only positively related to SWB among young adults rather than older adults, and race was unrelated to SWB.
Table 1. Correlations between subjective well-being, wisdom, religiosity, mastery, purpose in life, and demographics, pearson’s r
Note. Correlations for young adults above the diagonal and for older adults below the diagonal. SWB = subjective well-being.
**p < .01, *p < .05.
Multi-group path analyses
Preliminary regression analyses revealed one outlier with a standardized residual of −3.95 when purpose in life was the dependent variable, but a removal of this outlier did not substantially change the results. Hence, the case was retained. Preliminary analyses also showed that gender was unrelated to the three dependent variables in the model after controlling for the other variables. Because the groups were relatively small, the analyses could not be performed if too many variables or too many paths were included in the models. Therefore, gender was not included in the final path model.
We first tested the direct effects Model 1 (see Table 2) that regressed SWB on wisdom, religiosity, subjective health, education, race, nation of residency, and the interaction terms between nation and education and nation and race. Model 1a, which tested equality of form, was a saturated model with no remaining degrees of freedom, so this model fit the data perfectly. Model 1b was the constrained model that tested whether the coefficient estimates of the direct effects were equal between the two age groups. This model fit the data well with a non-significant chi-square difference test, a high value for CFI, and low values for SRMR for the two groups. However, a relatively high modification index suggested that the effect of education on SWB differed for the two age groups. Model 1c significantly improved the model fit by allowing this coefficient estimate to vary freely between the two groups, as indicated by a significant chi-square difference test between M1b and M1c. Thus, Model 1c was the best fitting directs effect model. As shown in the first column of Table 3, and as predicted by Hypothesis 1, both wisdom and religiosity were positively related to SWB after controlling for the positive medium size effect of subjective health and demographics. The association between wisdom and SWB was moderately strong, whereas the relation between religiosity and SWB was relatively weak. Whites and Canadian residents tended to score higher on SWB than non-whites and U.S.A. residents. However, the interaction effect between race and nation of residency was not significant. Education was positively related to SWB among young adults (but not among older adults) and was not moderated by nation of residency, rejecting Hypothesis 4. Among young adults and older adults, 35 percent and 53 percent, respectively, of the variation in SWB could be explained by the variables in Model 1c.
Table 2. Goodness-of-fit indices and chi-square difference tests of nested structural equation models testing the relations of wisdom and religiosity on subjective well-being directly and mediated by mastery and purpose in life; multi-group comparisons between young adults and older adults in LISREL 9.3
a Satorra-Bentler scaled χ2 adjusted for nonnormality.
Note: ***p < .001; **p < .01; *p < .05; CFI = comparative fit index; SRMR = standardized root mean square residual; YA = young adults; OA = older adults; SWB = subjective well-being.
Table 3. Comparison of direct effects model and mediated path model of the relations of wisdom and religiosity on subjective well-being
Note: n = 100 for young adults and n = 111 for older adults; Robust maximum likelihood estimation using LISREL 9.30; b = unstandardized coefficient estimate, β = common metric standardized coefficient estimate; t-values corrected for nonnormality; SWB = subjective well-being, 3D-WS = three-dimensional wisdom; *** t-value >3.29 (p < .001); **t-value > 2.58 (p < .01); *t-value > 1.96 (p < .05).
Next, we tested a mediated path model (M2) that introduced mastery and purpose in life as mediators of the paths from wisdom and religiosity to SWB, with mastery predicted to be positively related to purpose in life. Model 2a, which tested equality of form, fit the data well (see Table 2). However, Model 2b, which constrained the coefficient estimates of the direct effects between variables to be equal across the two age groups, did not fit the data well, with an overall chi-square value that was borderline significant and a significant chi-square difference test between M2a and M2b. As in Model 1c, Model 2c allowed the path between education and SWB to differ between the two age groups, which significantly improved the model fit compared to M2b. Based on high modification indices, the following paths were allowed to differ between the two age groups: race on mastery (Model 2d), mastery on purpose in life (Model 2e), and wisdom on mastery (Model 2f). Chi-square difference tests confirmed that freeing up the equality constraints for these paths significantly improved the model fit between subsequent models. We determined that the best fitting model was M2f, which fit the data very well with a non-significant chi-square value (χ2(33) = 22.23, p = .922), a CFI of 1.00, and low SRMR-values for both age groups. The unconstrained model M2a and the more constrained model M2f did not differ in model fit, as indicated by a non-significant chi-square difference test (Δχ2(13) = 15.16, p = .297).
The best fitting mediated path model M2f is shown in Figure 2 and in the last four columns of Table 3. As predicted by Hypotheses 2 and 3, mastery and purpose in life partially mediated the effects of wisdom and religiosity on SWB, with mastery being directly and indirectly related to SWB, mediated by purpose in life. About half of the effect size of wisdom and religiosity on SWB in the direct effects model M1c could be explained by mastery and purpose in life in the mediated path model M2f, although both wisdom and religiosity were still significantly related to SWB. Yet, contrary to Hypothesis 3, purpose in life did not primarily mediate the association between religiosity and SWB. The association between religiosity and purpose in life was similar to the association between religiosity and mastery, which means that both variables equally mediated the relation between religiosity and SWB. However, two differences by age group emerged: Although mastery and purpose in life mediated the effect of wisdom on SWB for both age groups, the relation between wisdom and mastery was significantly stronger for older adults than young adults (indicated by a significant chi-square difference test between Models M2e and M2f), whereas the association between mastery and purpose in life was significantly stronger for young adults than older adults (indicated by a significant chi-square difference test between Models M2d and M2e). The opposing differences in the mediating effect sizes between young adults and older adults balanced each other out to result in similar indirect effects of wisdom on SWB mediated by mastery and purpose in life (unstandardized indirect effect for young adults = .43, p < .001; unstandardized indirect effect for older adults = .46, p < .001). Because wisdom and religiosity were unrelated to each other, they can be considered two alternative pathways to greater SWB, mediated by mastery and purpose in life.
Figure 2. Results of path analysis model 2f.
Among the control variables, subjective health was positively related to mastery, and white older adults tended to report a stronger sense of mastery than non-white older adults, whereas race was unrelated to mastery among young adults. The interaction effects between education and nation and between race and nation became statistically significant in the mediated path model. A simple slope test (Dawson, Reference Dawson2014) showed that the effect of education on SWB was only statistically significant for young U.S.A. adults (.25, p = .001) but not young Canadian adults (.09, p = .52), older U.S.A. adults (.07, p = .17), or older Canadian adults (−.09, p = .46). The total effect of race on SWB was significant for young U.S. adults (.37, p = .001) and older U.S. adults (.47, p<.001) but not young Canadian adults (−.04, p = .88) or older Canadian adults (.06, p = .83). Whites tended to report greater SWB than non-whites in the U.S.A. but not in Canada, and this effect was more pronounced among older adults than young adults, partially confirming Hypothesis 4. Among young adults and older adults, respectively, the variables in Model 2f explained 17 percent and 45 percent of the variation in mastery, 26 percent and 30 percent of the variation in purpose in life, and 57 percent and 60 percent of the variation in SWB.
Discussion
The purpose of the current study was to test whether wisdom and religiosity are two alternative pathways to SWB mediated by mastery and purpose in life and whether the mediated path model is similar for older adults and young adults from the U.S.A. and Canada. Although research has shown that mastery and purpose in life partially or fully mediate the association between wisdom and SWB among older adults (Ardelt and Edwards, Reference Ardelt and Edwards2016; Etezadi and Pushkar, Reference Etezadi and Pushkar2013), no prior studies analyzed the simultaneous pathways from wisdom and religiosity to SWB or examined the mediated path model among young adults. The results support Hypothesis 1. Despite significant differences in variable means by nation of residency and age group, wisdom and religiosity were significantly related to SWB both in young adulthood and old age after controlling for subjective health and demographics. Hypothesis 2 was also confirmed. The association between wisdom and SWB was partially mediated by mastery and purpose in life with mastery being positively related to purpose in life, as has been reported in an early study of older adults (Ardelt and Edwards, Reference Ardelt and Edwards2016). The relation between religiosity and SWB was similarly mediated by mastery and purpose in life rather than primarily by purpose, as predicted by Hypothesis 3, but the relation was weaker than the association between wisdom and SWB. Although religiosity was unrelated to SWB, mastery, and purpose in life in bivariate correlations, the effects of religiosity on SWB, mastery, and purpose in life became positive and significant after controlling for subjective health and demographics. This suggests that the results of bivariate correlations can be misleading if differences in subjective health and demographic characteristics are not taken into account.
The relations between the variables in the path model in Figure 2 were significant for young and older adults from the U.S.A. and Canada, although the size of the coefficient estimates varied by age group for two associations. Wisdom had a statistically stronger effect on mastery among older adults than young adults, and mastery was statistically more strongly related to purpose in life among young adults than older adults. Because mastery tended to be significantly lower in this study for older adults than young adults, wisdom might have been more important for older adults to retain a sense of mastery and control over their lives (Ardelt, Reference Ardelt2005; Montross-Thomas et al., Reference Montross-Thomas, Joseph, Edmonds, Palinkas and Jestein press). By contrast, older adults tended to report significantly greater purpose in life than young adults (see Table S1 published online at https://www.cambridge.org/core/journals/international-psychogeriatrics). It is possible that older adults are able to find purpose and meaning in life, even if they feel that control over their life has declined, whereas young adults’ life purpose is more dependent on a secure sense of mastery and threatened by a loss of control (Bierman et al., Reference Bierman, Fazio and Milkie2006).
Hypothesis 4 that the positive effects of education and white race on SWB would be stronger among older adults in the U.S.A. than older adults in Canada or young adults based on cumulative advantage/disadvantage theory (George, Reference George2010) and the differing social systems in Canada and the U.S.A. (Lasser et al., Reference Lasser, Himmelstein and Woolhandler2006) was only partially supported. Consistent with Hypothesis 4, white respondents in the U.S.A. tended to report greater SWB than non-whites, particularly among older adults, whereas race was unrelated to SWB among Canadian residents. One explanation for this finding is that race is less salient to Canadian identity, where multiculturalism is written into the highest law of the land. In fact, Section 27 of the Canadian Charter of Rights and Freedoms states that the charter “shall be interpreted in a manner consistent with the preservation and enhancement of the multicultural heritage of Canadians,” which some legal scholars believe was inserted precisely to shape Canadian identity (Uberoi, Reference Uberoi2009).
Contrary to Hypothesis 4, however, education was only related to greater SWB for young U.S.A. adults but not older U.S.A. adults. This might be due to the highly educated sample that we collected. All of the young adults were either undergraduate or graduate students, and only one older U.S.A. adult had no college education, whereas 17 older Canadian adults were without a college education. Therefore, we did not have enough variation in education, particularly among older U.S.A. adults, to test the cumulative effect of education on SWB in old age. Yet, a higher education still affected SWB among young U.S.A. adults, but not young Canadian adults, which indicates the importance of education in a highly competitive individualistic society that lacks a strong safety net for those without monetary resources, as exists in Canada.
Limitations
The major limitations of this study are the cross-sectional nature of the data, the relatively small sample size, and the non-representativeness of the data. Because the data are cross-sectional, we can only test a possible path model but cannot determine the temporal order of the associations. Yet a short-term longitudinal study of older adults found that after controlling for the respective baseline scores, wisdom predicted mastery, purpose in life, and SWB 10 months later but not vice versa (Ardelt, Reference Ardelt2016). These results lend credence to parts of the suggested path model, although a longitudinal study with a larger, nationally representative sample is needed to test whether the results can be replicated. Moreover, a large number of coefficients were estimated for the mediated path model, which increases the probability of chance findings. If only those coefficient estimates that were significant at p < .001 are accepted as statistically significant, the direct effect of religiosity on mastery and purpose in life and the indirect effect of religiosity on SWB mediated by mastery and purpose in life would no longer be considered statistically significant. Additional limitations are that all measures were self-assessed and therefore might be affected by a social desirability bias (Staudinger and Glück, Reference Staudinger and Glück2011) and that religiosity was measured by a single item. It is possible that other measures of wisdom and religiosity would have resulted in different outcomes. For example, rating measures of wisdom tend to be unrelated or inconsistently related to SWB, whereas scale measures of wisdom are more consistently associated with greater SWB (Ardelt, Reference Ardelt, Wingard, Bengtson and Silversteinin press). Finally, our analysis did not consider the qualitative data provided by the interviews, which might show how participants understand the relationship between, for example, wisdom, mastery, purpose, and SWB. While important, such analysis deserves treatment in a separate paper.
Conclusion
Overall, this study demonstrated that wisdom and religiosity might be alternative paths to SWB, mediated by mastery and purpose in life. The pathways were not significantly different for young and older adults from Canada and the U.S., which increases confidence in the generalizability of these results. Yet, greater mastery and purpose in life is not the only pathway that connects wisdom and religiosity to SWB. Future longitudinal research should investigate other possible pathways, such as coping ability (Etezadi and Pushkar, Reference Etezadi and Pushkar2013), gratitude (König and Glück, Reference König and Glück2014), self-compassion (Neff et al., Reference Neff, Rude and Kirkpatrick2007), and resilience.
Conflict of interest
None.
Source of funding
The study was supported by a grant from the Social Sciences and Humanities Research Council of Canada (SSHRC), Grant Number 861070035. SSHRC did not influence the study design, data collection, data analysis, or decision to publish.
Description of authors’ roles
M. Ardelt was the co-PI of the larger study grant, designed the present study, analyzed the data, and wrote the article. M. Ferrari was the PI of the larger study grant and contributed to the writing of the article. Both authors supervised the collection of the data.
Acknowledgments
We like to thank the many graduate and undergraduate students who assisted in the study as recruiters, interviewers, transcribers, and data enterers and the research respondents for their participation in the study.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1041610218001680.
