| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Department of Applied Social Studies, City University of Hong Kong.
2 Asia-Pacific Institute of Ageing Studies, Lingnan University, Hong Kong.
Address correspondence to Sheung-Tak Cheng, Department of Applied Social Studies, City University of Hong Kong, 83 Tat Chee Avenue, Kowloon, Hong Kong. E-mail: tak.cheng{at}cityu.edu.hk
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
LIFE satisfaction (LS) is arguably one of the most commonly conjured images of successful aging (Tate, Lah, & Cuddy, 2003
); it is a key indicator of well-being (Diener, Suh, Lucas, & Smith, 1999) in which affective and cognitive evaluations converge (see Cheng, 2004a, 2004b). Research has shown that one's relationship with others, in terms of the quality of social interactions and support exchanges, is an important determinant of LS (Diener et al., 1999; Larson, 1978). Positive relationships satisfy the basic human needs for belongingness (Baumeister & Leary, 1995) and provide the necessary assistance when difficulties arise. Scholars working from different theoretical perspectives have converged on the view that maintaining positive relations with others is essential to optimal functioning (Ryan & Deci, 2001; Ryff, 1989). It has been further argued that, as time becomes increasingly limited in old age, close relationships become more important because they provide immediate emotional gratification rather than the attainment of some distant goals (Carstensen, Isaacowitz, & Charles, 1999). Nonetheless, to our knowledge, whether the importance of relationships differs by gender has not been examined. Two observations led us to believe that social relationship would not relate to LS in the same way in men as in women.
First, the same condition does not predict LS in the same way for every person. Value moderates the effects of predictors in such a way that experiencing a positive outcome in something important to self would result in a more powerful uplift of LS (Oishi, Diener, Suh, & Lucas, 1999). For example, progress toward communion goals enhanced well-being only in those individuals high in communion motive (Brunstein, Schultheiss, & Grässmann, 1998). Thus, what predicts LS reflects to a certain extent the priorities people place in their lives, and the priority given to connectedness is different for men and for women.
Second, successful aging depends on an assessment of one's priorities vis-à-vis available resources, and the selective pursuit of goals that are both important and attainable so as to maximize gains and minimize losses against a declining trajectory (Baltes & Baltes, 1990). From this perspective, disengagement from unattainable goals is adaptive (see also Brandtstädter & Greve, 1994; Heckhausen & Schulz, 1995).
In the following paragraphs, we argue that not only do women value relatedness more than do men, but they are also more resilient than men to maintaining a positive social network despite the challenges of aging, and are therefore expected to keep seeing relationship goals as important and contributory to their well-being. In contrast, certain conditions in later life present a greater challenge for men than for women to satisfy their relationship goals, and eventually they might find it necessary to disengage from such goals, such as when a major disruption to their network (e.g., death of spouse) has occurred.
Gender and Relatedness
It is well accepted that women value relatedness more than men, that they are socialized to play the role of nurturing and caretaking persons as well as kinkeepers in the family, and that their self-concept is defined more in terms of their relations to important people in their lives (Cross & Madson, 1997; Gallagher & Gerstel, 1993; Saragovi, Koestner, Dio, & Aubé, 1997). Whether in collectivistic or individualistic societies, a gender difference in relatedness is evident (Kashima et al., 1995). Although there is a tendency for social blurring as people age (i.e., the crossover of gender stereotypic cognitions and behaviors; see Gutmann, 1987), such long-standing differences tend to continue into old age (Canetto, Kaminski, & Felicio, 1995; Gallagher & Gerstel). Older women are also more likely than older men to use their friends as confidants, to invest more (such as in the form of support exchanges) in order to maintain closeness with network members over time, and to maintain contact with extended family members as well as with friends (see Chappell, 1989; Field, 1999; Gurung, Taylor, & Seeman, 2003; Wu & Pollard, 1998). As a result, women are popular members even in men's networks (Akiyama, Elliott, & Antonucci, 1996; Antonucci & Akiyama, 1987). Women's higher tendency to engage in support exchanges can be understood from the point of view of an interdependent self-construal, as it is through contributing to the well-being of others that they engender positive feelings about themselves (Cross & Madson). Given this information, it is reasonable to assume that despite dwindling resources in old age, women continue to invest heavily in relationships. This stronger allocation of resources to relationships by women than by men makes us believe that the quality of relationships is a stronger determinant of LS in women than in men.
Gender and Adaptation to Widowhood
We further asked if the gender effect might differ by widowhood status because men put "more of their intimacy ‘eggs’ into the marriage ‘basket’" (Cross & Madson, 1997, p. 24). Consequently, studies have found that men reported greater decline in mood and LS, especially when they have remained widowed for a long period of time (Chipperfield & Havens, 2001; Lee, DeMaris, Bavin, & Sullivan, 2001; Van Grootheest, Beekman, Van Groenou, & Deeg, 1999). Cross and Madson argued that intimate disclosure and seeking social support with a same-sex friend are seen as threatening to men's autonomy (see also Nagurney, Reich, & Newsom, 2004). However, intimacy with another person high in interdependent, rather than independent, self-construal (typically a woman) would be less threatening. Indeed, research shows that older men confide almost exclusively in their spouse (Chappell, 1989; Gurung et al., 2003), and once they are widowed they are less likely than women to make up for network loss (Lamme, Dykstra, & Broese Van Groenou, 1996; Pihlblad & Adams, 1972). In fact, being freed from caregiving responsibilities, widows often increase social participation, rekindle old ties, intensify existing ties, and develop new ones (Gallagher & Gerstel, 1993; Seltzer & Li, 2000; Zettel & Rook, 2004). Scholars have suggested that men play a more passive role in initiating social contacts (even with children) and in intimate behaviors (Cross & Madson; Lawton, Silverstein, & Bengtson, 1994), and thus they are less skilled in finding substitutes for emotional closeness when the spouse is gone. They also have fewer same-sex friends who are likewise widowed with whom to share their feelings. In Hong Kong, women 65 years or older are 3.5 times as likely as men to lose their spouse (51.0% vs 14.1%; Census and Statistics Department, 2002); this is similar to the situation in the United States, where the proportion of women and men widowed are 44.3% and 14.3%, respectively (U.S. Census Bureau, 2004). Although proportionally more older men (23.3%) in China are widowed, the basic fact that widowhood is a more common experience for women (50.5%; National Bureau of Statistics of China, 2002) remains unchanged.
When the spouse is gone, children become an important source of support, especially for Asians. However, whether in societies that are individualistic or collectivistic (in which one's obligations to parents are well defined), adults feel closer to older mothers than to older fathers (Akiyama et al., 1996; Antonucci, Akiyama, & Takahashi, 2004; Lawton et al., 1994), and are expected to interact with their widowed parent accordingly (e.g., provide more assistance to the mother when needed; see Chappell, 1989; Gurung et al., 2003; Ingersoll-Dayton, Starrels, & Dowler, 1996). These observations led us to speculate that widowers would tend to distance themselves from affiliation or relationship goals, but the same would not apply to widows. Hence we further hypothesized that the gender difference in the importance attached to relationships is even more pronounced among widowed persons (i.e., a Gender x Widowhood x Relatedness effect on LS).
The Study
Using structural equation modeling, we tested the moderating effect of gender and widowhood, after controlling for other major predictors of LS, namely health, formal social participation, financial strain, and quality of housing. We selected these factors on the basis of prior research on the LS of older adults. For instance, Larson's (1978) early review showed that, besides social contact and the quality of interactions, only physical health and socioeconomic factors consistently emerged as strong predictors of LS in older adults across studies (see also Diener et al., 1999). In studies conducted in Chinese societies (including China, Taiwan, and Hong Kong) where elderly people are not protected by any pension system, financial strain consistently emerged as a strong predictor of LS (e.g., Chen, 2000; Chou & Chi, 1999; Zhang & Yu, 1998). Moreover, Larson's early review also supported the contribution of living conditions (e.g., quality of housing) and formal social participation (e.g., organizational and church activities; see also Okun, Stock, Haring, Witter, 1984) to LS.
Before we could test the targeted interaction effect, a proper specification of the structural relationships among the variables was necessary. We tested four models in order to ascertain which one provided the best fit to the data. Because two major demographic variables, gender and marital status, have already been included as moderators and thus excluded from the predictor list, and because previous studies have shown that demographic variables contribute negligibly to the explanation of LS (Diener et al., 1999; Larson, 1978), we did not include these variables in model construction; that way, we could construct more parsimonious models. The four models are depicted in Figure 1.
|
) and on social activities. They differed in the way interpersonal relations, health, and social participation relate to each other. Our guiding theoretical model is one that recognizes the fact that participation is enabled by physical health, which in turn is enhanced by positive relations with others (Seeman, 2000). In the Alternate A model, we added a path from participation to interpersonal relations to take into account the possible influence of participation on relatedness, as well as the effect of health on relatedness by means of participation (see Van Tilburg & Van Groenou, 2002). Participation in social activities might enhance one's social network and contribute to the maintenance of relationships in the long run. In Alternate B, however, both interpersonal relations and participation are partial products of physical health (i.e., health is assumed to affect relatedness directly rather than indirectly through participation). For example, chronic or serious health problems might exert a strain on the supportive network and adversely affect the elderly person's relationship with close network members (e.g., Lawrence, Tennstedt, & Assmann, 1998). Finally, Alternate C is a nonrecursive model that contains bidirectional paths between health and interpersonal relations, and between health and participation. Besides acknowledging the mutual influence of health and relatedness, this model takes into account the effect of activities on health (e.g., Glass, Mendes de Leon, Marottoli, & Berkman, 1999). In sum, the alternate models are meant to exhaust probable interrelationships among relatedness, health, and participation as known in the literature and to see which of these models provide the most parsimonious, yet powerful, description of the data. Once we identify the most fitting model, our primary interest is on the path leading from interpersonal relations to LS, to see how this path varied by gender and widowhood status. |
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). The original study recruited a representative sample of older persons (aged 60 or older) in Hong Kong and contained measures of the variables needed for this study. The sampling procedure, details of sample characteristics, and the items used in this study are described elsewhere (Cheng, Chan, & Phillips, 2004). However, in order to test our hypotheses in the current study, we found that regrouping of items was necessary to form the measure of social participation.
Participants
The sample represents an exhaustion of households with one or more elderly persons in the General Household Survey conducted in the year 1999. The General Household Survey is conducted by the Census and Statistics Department of the Hong Kong Special Administrative Region in order to update the population's labor force characteristics on a quarterly basis. The sample used for the General Household Survey was composed of a stratified random sample of all households in Hong Kong. Within this General Household Survey's sample, 3,000 households with elderly persons were approached for an interview. However, because of various reasons such as geographical mobility, housing redevelopment, transfer to institutions, mortality, and so on, only 2,190 households were successfully contacted. Researchers randomly picked an elderly person from each household, and, after clearing person, time, and place orientation (items taken from Mini-Mental State Examination; see Folstein, Folstein, & McHugh, 1975), invited the person for an interview (typically in his or her own home). The interview lasted for approximately 45 min. Altogether, 1,616 community-dwelling elderly persons participated in the study (response rate = 73.8%), and 10% of these cases were randomly selected for a quality check interview 1 week after the original interview to ensure data accuracy. Missing data were rare and were filled in by follow-up telephone interviews. This sample was equally balanced between women (50.7%) and men (49.3%) and had a mean age of 70.64 years (SD = 7.12, range = 60–99). There was no age difference between women and men, M = 70.63 versus 70.66, t(1604) = 0.08, ns. Thirty-two percent of the sample was widowed, and another 60.9% were married.
Measures
As will be evident, the constructs to be measured are multifaceted, and hence we formed indicators for the latent constructs in such a way so as to make explicit the multifaceted nature of the constructs (Little, Cunningham, Shahar, & Widaman, 2002). Where possible, we formed a facet indicator by averaging multiple items because these item parcels are more reliable than the individual items themselves, are less likely to suffer from distribution problems, and improve the ratio of sample size to indicators (Little et al.; Nasser & Takahashi, 2003). Note that even if the reliabilities of the parcels are not most desirable, the reliabilities of the indicators, and hence the corresponding latent construct, are still improved.
Satisfaction with current life
We measured life satisfaction with two items, which assessed the degree to which one is satisfied with one's current life as a whole, and is leading a way of life with which one is happy. Although our measure consisted of just two items, research has shown that such short measures are as good as more lengthy ones in assessing global well-being such as LS (Larsen, Diener, & Emmons, 1985). The alpha coefficient was 0.73 in the present sample.
Interpersonal relations
Three composite measures, that is family relations, intergenerational relations, and friendship, and a single-item measure, companionship, formed the indicators. We measured family relations with two items (
= 0.68) relating to social support and harmony within the family. Intergenerational relations referred to the degree to which one engaged in actions to nurture the younger generation, and was treated in a respectful and filial way in return. We measured it with four items, with an alpha coefficient of 0.62. We measured friendship with two items ( = 0.60) tapping the extent of the friendship network and the social support received. One item measured the extent of time being spent with someone else, which we labeled companionship. The use of such measures, rather than the items themselves, as indicators prevents the overall construct from being dominated by specific qualities having more items, and it maintains the meaningful structure of the construct (Cheng et al., 2004).
Health
Four variables made up this construct. We measured mobility and vitality with three items ( = 0.69) about the ability to move around and do exercise. One item assessed the quality of sleep and another assessed oral health in terms of the ability to chew food (in the Chinese context of "food obsession," chewing ability can be an important determinant of perceived health). General health referred to the frequency of illness and self-rated overall health, and we measured it with two items ( = 0.53). Although the alpha coefficient for general health was not impressive, it was quite acceptable for the purpose of estimating the latent construct given the number of items (Nasser & Takahashi, 2003).
Formal social participation
Earlier factor analyses showed that the frequency of participation clustered together with a sense of talent fulfillment, suggesting that participation in activities was often driven by the opportunity to actualize one's talents (Cheng et al., 2004). Two indicators made up this construct. We assessed formal activity level with four items ( = 0.79) measuring the frequency of an individual's engagement in organized or group activities, volunteer services, and interesting pursuits. One item also asked if the individual participated in organizing activities for others. We measured fulfilling activities with three items ( = 0.71) concerning whether there was sufficient opportunity for the elderly person to utilize his or her talents and to learn new skills.
In addition, we measured financial strain with one item tapping whether the respondent could make ends meet, and we measured housing quality with three items relating to spaciousness and cleanliness of the residence ( = 0.68). Though we measured it with only one item, the financial strain measure correlated at.84 with a three-item measure widely used in research (see Krause, Jay, & Liang, 1991) in a separate community sample of 206 elderly persons in Hong Kong (unpublished data), and it correlated at.53 with household income in the present sample. For the purpose of structural equation modeling, we specified the latent variable of financial strain as equal to the observed item with zero residuals.
Participants rated all of the aforementioned items on a 5-point Likert scale. We reversed the scores where appropriate, so that a higher composite score indicates a stronger quality in the area concerned.
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
; Smith & Baltes, 1998), there was a small tendency, in terms of point-biserial correlations, for women (coded as 1, men coded as 0) to report less mobility and vitality (rpb = –.13), worse sleep quality (rpb = –.11), poorer general health (rpb = –.11), and fewer opportunities to fulfill their talents (rpb = –.08), as well as more financial strain (rpb =.08), all ps <.001. None of the variables correlated with age to a significant extent.
Testing Structural Equation Models
We subjected the covariance matrix of the indicators to structural equation modeling using the maximum likelihood method of estimation by LISREL (version 8.52). We first compared the fitness of the four models, and we then tested the hypothesized moderating effect in the most fitting model in four subsamples: married women, married men, widows, and widowers. As suggested by post hoc modification analysis, we allowed a correlated error between family and intergenerational relations; we justified this correlated error on the basis that intergenerational dynamics are played out in the family context on a day-to-day basis (Cheng et al., 2004).
The goodness-of-fit indices of the four models are shown in Table 2. The theoretical model was well fitting to the data; except for the chi-square statistic, which is sensitive to sample size, the Comparative Fit Index (CFI), standardized root mean square residual (SRMR), and root mean square error of approximation (RMSEA) were well within acceptable limits (Hu & Bentler, 1999; Marsh, Hau, & Grayson, in press). Moreover, this model produced fit indices that were generally superior to those of the other three alternate models, most notably in terms of the ratio of chi square to degrees of freedom (
2/df) and the Akaike Information Criterion (AIC). The AIC is a modification of the chi-square statistic with a penalty for model complexity; an AIC difference 
10 is a strong indication that one model is inferior to the other (Burnham & Anderson, 2002). As we can see, the differences in AIC values between the theoretical and the alternate models were all well above 10. Hence the theoretical model should be the model of choice. The structural path coefficients found for the theoretical model are shown in Figure 2 (ignore the coefficients in the parentheses for the time being).
|
|
review, physical health and housing quality contributed considerably to the explanation of LS. Contrary to some Western findings, social participation had a small, though statistically significant, effect on LS after we controlled for the other factors. Furthermore, it should be noted that, besides its direct effect on LS, interpersonal relations had a massive indirect effect (.52 x.21 =.11) through affecting health. The indirect effect of financial strain through health deterioration was much smaller (–.14 x.21 = –.03). Theoretically, relations with others and health and financial strain also make an impact on LS through their effects on participation, but given participation's small effect on LS, these indirect effects were bound to be negligible. Overall, interpersonal relations had an effect size, direct and indirect effects combined, on LS that was equivalent to that of financial strain (both being.40); this conclusion would not have been permitted had we used the multiple regression analysis typically used in this literature.
Testing the Moderating Effect
In structural equation modeling, one can test moderating effects by forcing specific structural path coefficients to be equal across levels of the moderating variable(s), following a demonstration of measurement invariance (Rigdon, Schumacker, & Wothke, 1998). If the path coefficients cannot be forced equal (i.e., variant), then a moderating effect exists. Here, we excluded those who were never married (n = 68), living together (n = 2), and divorced or separated (n = 39), because our interest was to see how the loss of a spouse further interacted with gender on the relationship between relatedness and LS. We split the data set up into four groups, married women (n = 473), married men (n = 498), widows (n = 280), and widowers (n = 235), and we fit the theoretical model separately for them. Results showed that the model fit the data rather well across the four groups, that is, 2/df = 2.09–2.51, CFIs =.93–.97, SRMRs =.05–.07, RMSEAs =.05–.07, and the same factor pattern existed in them, 2(368) = 836.81, CFI =.95, SRMR =.06, RMSEA =.06. We assessed the invariance of factor loadings by forcing all loadings to be equal across the four groups, over and above factor-pattern invariance (i.e., a nested model), resulting in 
2(30) = 31.75 (ns), CFI =.95, SRMR =.06, and RMSEA =.06. These findings suggested that the same set of latent constructs was being measured (i.e., invariance), across married women, married men, widows, and widowers.
With measurement invariance established, we proceeded to test the moderating effect of gender and widowhood in yet another nested model. On top of factor-loading equality, we forced the path coefficient from interpersonal relations to LS to be equal across the four groups. This, however, produced a significant increase in chi-square = 18.99 for 3 more degrees of freedom (p <.001), meaning that the relationship between relatedness and LS varied across the four groups. To illustrate this interaction effect, we give the group-specific coefficients in parentheses next to the path in Figure 2. As we can see, women's interpersonal relations had a much larger effect on LS than did men's, whether in the married or in the widowed condition. Moreover, relations with others did not affect LS, after we controlled for the other variables, in widowers (note that this path was nonsignificant), so that gender's moderational role was more pronounced among the widowed than among the married. These results were consistent with our hypotheses.
However, there was a potential confound in these findings. Because widowed elderly individuals were generally older than the married ones, that is, Mage = 74.17 versus 68.90, t(1474) = 14.48, p <.001, it was not clear if what was found was a Gender x Widowhood interaction, or a Gender x Age interaction. We dealt with this issue by examining if the effect of interpersonal relations varied by age and widowhood status in men (because the earlier result suggested that widowhood had a moderating effect in men only). However, because of sample size constraint, we were not able to perform this test using structural equation modeling (purely because of an insufficient number of men in the old-old category, married or widowed). Therefore, we used a hierarchical regression approach. We regressed LS on, in order of entry, (a) age, widowhood (0 = married, 1 = widowed), and interpersonal relations; (b) Age x Relations, Widowhood x Relations, and Age x Widowhood; and (c) Age x Widowhood x Relations. We formed the measures of LS and interpersonal relations for the regression analysis using composites of the standardized scores of the constituent indicators in order to maintain the measurement structure of these variables as used in structural equation modeling. Of the four interaction terms, only the Widowhood x Interpersonal relations effect was significant (p <.01). Thus, despite possible inconsistencies between the latent and the observed variable level, we feel that the moderating effect modeled in structural equation modeling was best interpreted as the joint effects of gender and widowhood.
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). Moreover, in the Chinese context, social interactions are often structured around meals. A popular way to show how much you esteem or love a friend or family member is to give him or her a big treat. Children learn from childhood that a customary way to act filial toward parents is to offer good (e.g., delicious, nutritious, and often expensive) food to them. All such interactions, we believe, can engender a sense of physical as well as social well-being, and hence higher LS.
Nevertheless, interpersonal relations acted in different ways for men and women. For older women, interpersonal relations had a direct impact on LS, while also having a moderate effect by means of health. For the older women, it was relations with others that were most important, and this was true even if they had suffered the loss of a close relationship (i.e., husband). For men, however, the loss of spouse made a lot of difference: Interpersonal relations had a moderate direct effect on LS for married men only; for widowers, the effect did not exist at all. We believe that these patterns are consistent with the view that what determines LS is moderated by priorities and goals (e.g., Brunstein et al., 1998; Oishi et al., 1999). Because there was no gender difference in the reported supportiveness of the network, we assumed that the difference in the magnitude of the relatedness-to-LS path represented a difference in the priority attached to relationships in older women's and men's lives. To the extent that LS is a product of the fulfillment of goals important to oneself (Cheng, 2004a), the present findings suggest that affiliation or communion remains a more important goal for women than for men, even in old age.
Having said this, we were still taken aback by the data on widowers: Relations with others did not matter at all as far as LS was concerned, after we controlled for the other factors. It is possible that whereas women are more willing to seek help (Cross & Madson, 1997), men withdraw socially instead as a result of the prolonged depressed state following bereavement (Van Grootheest et al., 1999). Would the loss of a spouse, perhaps the most emotionally meaningful relationship for older men, lead to a certain degree of detachment from other relationships as well, even though such relationships continue to be available? In view of a worldwide trend of declining fertility rate (United Nations Population Division, 2005), more and more older men in future cohorts are expected to be childless, and their social well-being, once they are widowed, may become a social issue. (Hong Kong happens to be one of the places with the lowest fertility rate in the world, with just about one child per woman of childbearing age.) Much is yet to be known about the psychological and social adjustment of widowers across cultures and across time.
A few limitations of the present study have to be mentioned. First, this study was cross-sectional and the causal relationships among the factors were theoretically derived, not demonstrated. The lack of longitudinal data also prevented us from studying how models of LS altered with the transition into widowhood (cf. Chipperfield & Havens, 2001). Second, given the likelihood of further gender-role convergence in the future (Arber, 1996), whether the present findings can apply to future cohorts remains to be seen. Another possible development is the increasing acceptance of remarriage and cohabitation in old age (currently extremely unlikely among Chinese), which might revive men's relationship goals. Third, the issue of longitudinal data aside, our measure of widowhood ignored the length of time of being a widow, which might affect the results obtained, especially when the length of widowhood differed between the widows and the widowers. Fourth, measures of negative social interactions as well as objective health were not obtained. Both of these might produce a different picture about how relatedness and physical health relate to LS, as well as how relatedness relates to health. Fifth, some of the facets forming the latent construct were measured by only one item, and others were measured by more but still a very limited number of items. To further improve the reliability of the estimation and to enhance model fit, researchers should use more items per facet constructed on an a priori basis (Nasser & Takahashi, 2003). Sixth, an issue that is generic to most studies of community-dwelling elderly people, the bias toward the healthier elderly individuals, made it difficult to assess how models of LS might differ for the less healthy ones. Finally, more data are necessary to ascertain the cross-cultural generalizability of the present findings.
To conclude, on the basis of a representative sample of community-dwelling elderly individuals in Hong Kong, we found strong support for the role of interpersonal relations in LS, except among widowers. On the whole, our findings by and large showed cross-cultural similarities in the determinants of LS, but they also confirmed the significance of financial strain in Chinese elderly persons living in this region. We suspect that financial strain might be a similarly important factor in elderly persons living in other parts of the world that do not provide a sufficient safety net for people who are getting old. Finally, this study demonstrates that what determines LS can vary substantially by one's values and priorities, and consequentially patterned by gender and one's place in the life course (e.g., widowhood). Future researchers should pay closer attention to these issues when they develop theories of LS.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication January 11, 2005. Accepted for publication July 25, 2005.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  S.-T. Cheng and A. C. M. Chan Filial piety and psychological well-being in well older chinese. J. Gerontol. B. Psychol. Sci. Soc. Sci., September 1, 2006; 61(5): P262 - P269. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
