| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Human Care Research Team, Tokyo Metropolitan Institute of Gerontology, Japan.
2 Area of Gerontology, Graduate School of Obirin University, Tokyo, Japan.
3 Department of Humanities and Social Sciences, Jissen Women's University, Tokyo, Japan.
Address correspondence to Yoko Sugihara, PhD, Human Care Research Team, Tokyo Metropolitan Institute of Gerontology, 35-2 Sakae-cho, Itabashi-ku, Tokyo 173-0015, Japan. E-mail: sugihara{at}tmig.or.jp
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Depressive symptoms • Productive roles • Volunteer work
Recently, engagement in productive roles that produce goods or services, whether paid or not (e.g., paid work, housework, and volunteer work), has been emphasized as one of the main factors leading to successful aging (Herzog & House, 1991
; Rowe & Kahn, 1997) as well as the contribution of older adults to society (Bass, Caro, & Chen, 1993; Butler & Gleason, 1985). Previous studies using longitudinal data have found that paid work (Luoh & Herzog, 2002; Rushing, Ritter, & Burton, 1992) and unpaid housework (Menec, 2003) have protective effects against subsequent functional decline and death; that volunteer work is related to greater longevity (Lum & Lightfoot, 2005; Luoh & Herzog; Musick, Herzog, & House, 1999; Oman, Thoresen, & McMahon, 1999; Sabin, 1993), slower functional decline (Lum & Lightfoot; Luoh & Herzog; Menec; Moen, Dempster-McClain, & Williams, 1992; Morrow-Howell, Hinterlong, Rozario, & Tang, 2003), and increased psychological well-being (Lum & Lightfoot; Morrow-Howell et al.; Musick & Wilson, 2003; Thoits & Hewitt, 2001; Van Willigen, 2000); and that engagement in particular types of productive roles is associated with lower risks of mortality (Glass, Mendes de Leon, Marottoli, & Berkman, 1999) and dementia (Wang, Karp, Winblad, & Fratiglioni, 2002).
Although previous research generally has indicated that involvement in productive roles is beneficial for older adults' well-being, there are still some issues remaining. First, much of the work on productive roles is North American, and little is known about non-Western people such as the Japanese. Because the context in Japan concerning productive roles such as paid and volunteer work differs from that in the West (as subsequently described), the relationships between productive roles and well-being among older adults in Japan may differ from the results of previous studies carried out in North America.
Second, most prior studies have been confined to investigating only the independent effect that a specific productive role might have on older adults' well-being, and few studies have investigated any joint effects or possible interactions of various productive roles. Researchers often use concepts from role theory such as "role strain" and "role accumulation" to explain the impact that playing a productive role might have on health and well-being. The role-strain hypothesis suggests that most people engage in several roles and that the different types of role demands cause conflict, overload, and strain, resulting in poor well-being (Goode, 1960; Merton, 1957). In contrast, the role-accumulation hypothesis states that accumulated roles enhance status security, social privileges, valuable resources, and ego gratification and that multiple roles attenuate the stressful impact of any single role, which in turn benefits well-being (Marks, 1977; Sieber, 1974). These theories refer to the relationships between multiple roles and well-being, essentially how multiple roles interact with each other. Although prior studies on productive roles and well-being often refer to these multiple-role theories, little research has investigated the joint effects of particular role combinations. Nevertheless, there is a large body of research attempting to explain the relationships between involvement in multiple roles and well-being. However, earlier studies of multiple roles generally have focused on earlier adulthood, especially the effect that occupational, marital, and parental roles might have on women's well-being in their early to middle years of adulthood (Stephens, Franks, & Townsend, 1994). Relatively little is known about how specific role combinations might affect the well-being of late-middle-aged to older adults and whether there is any gender difference there.
In addition, the interactions between roles can be observed not only when individuals adopt multiple roles simultaneously but also when their roles change; for example, a particular role could moderate the potential negative impact caused by the loss of another role. Late middle age is a time of life when major roles such as carrying out paid work reach their term. Although there is a lot of research on the psychological consequences of retirement in the United States, the empirical evidence is inconsistent. Several studies suggest that how retirement affects psychological well-being depends on the specific contexts in which retirement occurs (e.g., Kim & Moen, 2002; Szinovacz & Davey, 2004). For example, an early study shows that mandatory retirement is associated with lower retirement satisfaction because the retirement transitions generally occur under circumstances that leave individuals little choice (Hardy & Quadagno, 1995). Because most Japanese companies adopt a mandatory retirement system, it is therefore quite possible that retirement may lead to reduced well-being for retirement-aged Japanese, though so far, to our knowledge, little research has investigated the effects of retirement on well-being in Japan.
In contrast, the literature suggests that unpaid work, especially volunteer work, may substitute for role loss such as that which exists in retirement (Chambré, 1984). An existing study shows that volunteering moderates the negative impact of having more major role-identity absences on older adults' psychological well-being (Greenfield & Marks, 2004). Because volunteerism tends to increase among older adults in Japan, volunteering may provide alternatives to paid work for retirement-aged adults, and it may moderate the potential negative impact on well-being caused by the loss of the paid work role.
In this study we examine the relationships between productive roles and depressive symptoms by using data from a national longitudinal study of late-middle-aged Japanese. Specifically, we aim to advance the understanding of productive roles and depressive symptoms by addressing the following research questions: First, what relationships exist between the various types of productive roles (i.e., paid work, volunteer work, and unpaid work at home) and depressive symptoms among late-middle-aged Japanese? Second, what are the relationships between specific productive role combinations and depressive symptoms? Third, what are the relationships between transitions in the paid work role and depressive symptoms? Fourth and finally, how does carrying out volunteer work moderate the relationship between the loss of a paid work role and depressive symptoms? In this study we also focus on possible gender differences in these relationships. Because men and women generally have different role opportunities and role experiences, the meanings of roles may vary with gender.
With regard to model specification, we referred to the theoretical model of productive aging by Sherraden, Morrow-Howell, Hinterlong, and Rozario (2001), which consists of the following categories: sociodemographics, individual capacity, public policy, institutional capacity, productive behavior, and outcome. Although those researchers emphasize the importance of public policy and institutional capacity (i.e., the ability of social institutions such as businesses, social civic clubs, and families to offer productive roles for older adults), in our study we did not investigate these factors. Therefore, we analyzed the models by including only sociodemographic variables (age, marital status, and education), individual capacity (income, physical health, and previous scores on the Center for Epidemiologic Studies–Depression scale), productive roles, and outcome.
Contexts of Productive Roles Among Late-Middle-Aged Japanese
Most companies in Japan adopt a mandatory retirement (teinen taisyoku) system, which requires most wage employees to retire by about the age of 60 years. However, many Japanese continue to work beyond that age. The participation of this later-life labor force is higher in Japan than in other industrialized countries. The labor force participation rate of men aged between 60 and 64 in 2004 was 70.7% in Japan, compared with 57.0% in the United States and 37.7% in Germany. Even men aged between 65 and 69 had a participation rate of 45.6% in Japan, compared with 32.6% in the United States and 7.2% in Germany (International Labor Organization, 2004). Especially, elderly Japanese men are highly motivated to work. A cross-national research shows that Japanese men aged 60 and older desire to work longer than their counterparts in the United States and Germany (Cabinet Office, 2002). Even though workers' pay and status after teinen is normally reduced, the higher work participation levels among elderly men in Japan are seen as a reflection not only of economic factors (Seike & Yamada, 2004) but of cultural values (Bass, 1996; McCallum, 1988; Shibata, 1998). Bass (1996) and Shibata (1998, 2002) suggest the importance of the Japanese concept of ikigai (purpose or meaning in life) for understanding the value of work and productivity in Japan. They suggest that many Japanese men find their ikigai in their work roles. From these high levels of later-life labor force participation and work ethics in Japanese men, it appears possible that paid work may have an important effect on Japanese men's psychological well-being. Despite this, because they have to leave their job when they reach a certain age, retirement may lead to even more psychological distress among elderly men in Japan than among those in Western countries.
Recently, Japanese policy makers have focused on volunteering to help retain ikigai in later life after retirement (Cabinet Office, 2005). Since the time when many volunteers from all over the country came to help victims of the Great Hanshin Earthquake in 1995, there has been a growing interest in volunteerism in Japan (Economic Planning Agency, 2000). According to a time-use survey, the overall rate of volunteering in Japan was 28.9% (of people aged 10 years and older) in the period from 2000 to 2001, up 3.6 percentage points from the previous survey 5 years earlier (Ministry of Internal Affairs and Communications [MIAC], 2001). Although volunteering had recently become popular in Japan, the volunteerism rate was still lower than that in the United States (44% of people over the age of 21 volunteered; see Independent Sector, 2001). Therefore, the beneficial effect of volunteer work on the psychological well-being of older adults may not be as great in Japan as in the Unites States. However, volunteer work tends to increase with age among middle-aged and older men in Japan (MIAC). This increase in volunteering with age among men may be an indication that volunteer work will become increasingly important to elderly Japanese men. In contrast, volunteer work among Japanese women tends to decrease with age because women often volunteer in or through their children's school; once their children mature, they no longer have the opportunity to volunteer as much as before.
Women carry out more unpaid work at home, such as housework and caregiving, than men do (MIAC, 2001). Although household labor is productive, it can be a burden because it is more routine, is less likely to be appreciated, and provides lower levels of work fulfillment than do other productive roles (Bird, 1999). In addition, women in their fifties are most likely to care for frail family members, and about 10% of them are caregivers in Japan (MIAC, 2001). The literature suggests that caring for elderly relatives compounds emotional exhaustion in female caregivers (Sugihara, Sugisawa, Nakatani, & Hougham, 2004). Although late-middle-aged women make an important contribution by doing unpaid work at home, this may not lead to increased psychological well-being.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
), of late-middle-aged adults in Japan. Using a stratified two-stage random-sampling procedure, researchers identified 4,000 men and 2,000 women aged 55 to 64 years from the nationwide resident registration, and they interviewed them in 1999. The baseline response rate was 63.3% (n = 2,533) among the male sample and 72.0% (n = 1,440) among the female sample. The follow-up interviews were conducted in 2001 (2,074 men and 1,214 women participated) and in 2003 (1,896 men and 1,113 women participated). All interviews were conducted face to face in participants' homes. This study did not obtain data from married couples.
Measures
Productive roles
Although there are significant differences in the definition of productive roles in the existing literature (Hinterlong, Morrow-Howell, & Sherraden, 2001), many empirical studies have defined a productive role as engagement in any activity that produces goods or services, whether paid or not, including paid work, unpaid work at home, or volunteer work (e.g., Glass, Seeman, Herzog, Kahn, & Berkman, 1995; Herzog, Kahn, Morgan, Jackson, & Antonucci, 1989; Morrow-Howell, Hinterlong, Sherraden, & Rozario, 2001). We adopted this operational definition.
Paid work
Participants were first asked if they were doing any paid work at the present time. Participants who answered in the affirmative were then asked how many hours per week they had worked on average during the past month.
Unpaid work at home
Participants were asked if they had carried out any unpaid housework, home maintenance, yard work, or taken care of children or frail family members, in the past year. Participants who responded in the affirmative were then asked about the average frequency in which they engaged in those activities during the past year, based on the following categories: every day, 4 to 6 days a week, 1 to 3 days a week, 1 to 3 days a month, or a few times a year. Participants who reported 1 day or more per month were then asked about the average number of hours spent on the activity per day. We converted the frequency category into the number of days per week (i.e., 7 days a week, 5 days a week, 2 days a week, 0.5 days a week), and we obtained the average number of hours spent on the activity per week by multiplying the number of days a week and the number of hours a day.
Volunteering
Participants were asked if they had given informal support to friends or neighbors or carried out any of the following volunteer work in the past year: engaging in an environmental arrangement, participating in social welfare activities, supporting community activities, and engaging in other volunteer work or helping others. We obtained the average number of hours spent in volunteer work per week through the same calculation process as we used for unpaid work at home.
Depressive symptoms
We used the Japanese version of the Center for Epidemiological Studies–Depression scale (CES-D; Radloff, 1977) in this study (Shima, Shikano, Kitamura, & Asai, 1985). A previous study using confirmatory factor analysis has confirmed that the original four-factor structure of the CES-D was successfully replicated for Japanese elderly persons (Yatomi, Liang, Krause, & Akiyama, 1993). The CES-D is a 20-item self-report measure used to examine the frequency of mood and behavioral symptoms that occurred during the previous week. Frequency of behavior is rated on 4-point scales ranging from 0 (rarely or none of the time) to 3 (most of the time). Positive items are reverse coded. Example items include "I felt depressed," "I felt that people dislike me," and "I felt hopeful about the future." Total scores could range from 0 to 60, with higher scores indicating more depressive symptoms. Internal consistency for this measure throughout the research ranged from.73 to.76.
Control variables
Control variables include age (in years), marital status (coded 1 if currently married, 0 if not married), education, income, physical health, and previous levels of depressive symptoms. We coded educational attainment as 1 = junior high school, 2 = high school, and 3 = college degree or more. Annual income was the total gross income of the participants and their spouses, which we coded as 1 = less than 3 million yen (1 Japanese yen = approximately 0.01 U.S. dollar), 2 = 3 to 5 million yen, 3 = 5 to 10 million yen, and 4 = 10 million yen or more. Physical health is captured by the number of serious or chronic diseases as follows: hypertension, stroke, angina pectoris or myocardial infarction, bronchitis or pneumonia, asthma, stomach or duodenal ulcer, hepatitis or liver cirrhosis, kidney disease, diabetes, arthritis or rheumatism, and cancer.
Statistical Analysis
In this study we used the generalized estimating equations (GEE) approach to control for within-subject correlated error across waves (Diggle, Heagerty, Liang, & Zeger, 2002). We specified an ordinary least squares approach with bootstrapped standard errors and an unstructured form for the working correlation matrix in all analyses. We constructed a data set with individual person-time observations that incorporated time-dependent variables. In this study, we regressed the dependent variable (depressive symptoms at Waves 2 or 3) on levels of productive roles taken from the same wave as the dependent variable, controlling for level of depressive symptoms taken from the previous wave of data. (If depressive symptoms at Wave 2 were used for outcome, we included depressive symptoms at Wave 1 as control in the analyses. If depressive symptoms at Wave 3 served as outcome, we controlled for depressive symptoms at Wave 2.) The other control variables apart from depressive symptoms were taken from the same wave as the dependent variable. Pooling the observations in this manner resulted in a sample size of 3,970 for men and 2,327 for women.
Regarding missing values of some study variables, we used multiple imputation methods through the Markov chain Monte Carlo method (Schafer, 1997) to create five data sets with no missing value. Among the imputed variables, income had the highest percentage of missing values (11.1%). We analyzed each imputed data set separately by GEE, and we combined the results to arrive at final estimates of parameters and their standard errors. We executed GEE analyses by using the GENMOD procedure and multiple imputation methods through the MI and the MIANALYZE procedures of SAS version 9.1.
We tested our hypotheses in four models separately by gender. In Model 1, we regressed depressive symptoms on the activity time of each productive role. In Model 2, we entered specific combinations of three productive roles. We operationalized the role combinations as follows. First, we coded respondents as paid workers if they reported devoting some time per week to paid work, and we applied the same coding manner to both volunteer work and unpaid work at home. Next, we categorized the specific combinations of roles held by respondents (e.g., paid work only, paid work and volunteer work, and all three roles). In Model 3, we examined the associations between changes in paid work status and depressive symptoms. For changes in paid work status, we pooled the information both between Waves 1 and 2 and between Waves 2 and 3, and we classified respondents according to whether they moved from working to nonworking, nonworking to working, nonworking at both waves, and working at both waves. Finally, in Model 4, we added the interactions between volunteer status and changes in paid work status to test whether volunteering moderated any effects of changes in paid work status on depressive symptoms.
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
10 hours/week, 10–40 hours/week, 40–48 hours/week, or 
48 hours/week) and examined whether any beneficial amount of work hours on depressive symptoms could be observed. The results using dummy variables were similar (i.e., linear relationship) to those using continuous variables of working hours (data not shown).
|
Next, we examined whether changes in PW status might have an impact on depressive symptoms (Table 3, Model 3). The distribution of changes in PW status was as follows: from working to nonworking (12.6% of men, 10.3% of women), nonworking at both waves (18.6% of men, 48.3% of women), nonworking to working (4.3% of men, 4.0% of women), or working at both waves (64.5% of men, 37.5% of women). Those men who shifted from working to nonworking (B = 1.023, p <.001) or those men who were continuously not working over both waves of the survey (B = 1.006, p <.001) reported more depressive symptoms than did those men who were continuously working. For women, however, none of variables for paid work transitions exerted a significant main effect on depressive symptoms.
|
|
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Several studies have identified gender as the critical context that determines whether roles are beneficial or not (Danigelis & McIntosh, 1993; Herzog & Morgan, 1992; McIntosh & Danigelis, 1995). Gender is closely related to the norms governing behavior. The literature suggests that people are most likely to feel distress when their role repertoire departs from the norms of their age and gender (Menaghan, 1989). Social norms of productivity generally tend to be more emphasized in men than women. The finding that PW is associated with reduced depressive symptoms in Japanese men but not women makes sense if one considers that Japanese women of that generation are not as strongly oriented toward PW as men.
Although VW does not seem to be as greatly influenced by gender role or social norms as PW, as with PW, our findings show that there was a gender difference in the relationship between VW and depressive symptoms. This gender difference might be more attributable to the fact that men have more to lose when they retire rather than differing social norms. Chambré (1984) suggests that VW may replace loss of role in later life. If Chambré's hypothesis is correct, then the beneficial effect of volunteering would be more pronounced in men than women because late-middle-aged men tend to experience more drastic role transitions, such as retirement, than late-middle-aged women do.
Another possible explanation for the lack of a significant relationship between women's productive role and depressive symptoms might be attributed to the high rates of depression among women. Because the statistical models in this study controlled for prior levels of depressive symptoms, the variables might merely be describing changes in the variance of depressive symptoms during the period between waves. Because women tend to report more depressive symptoms than men do, changes in the variance of depressive symptoms among women may have been relatively small.
Several U.S. studies looking at how much involvement is necessary to benefit well-being suggest that a modest degree of role involvement may be sufficient. For example, researchers have found that the impact of volunteerism is maximally beneficial when it totals 1 to 3 hours per week (Morrow-Howell et al., 2003; Musick et al., 1999; Van Willigen, 2000). Contrary to the findings on older Americans, this study indicates that the relationship between activity level of PW or VW and depressive symptoms is a linear function of the amount of time spent doing it. The degree of involvement in PW tends to decrease after a person reaches his or her late fifties, and there is not much VW in Japan generally. The reason why we found activity level to be linearly inversely related to depressive symptoms in this study might be because only a few people worked sufficiently long hours for it to become problematic.
Second, we examined the relationships between specific productive role combinations and depressive symptoms, based on role-strain and role-accumulation hypotheses. Of the two hypotheses, the role-accumulation hypothesis has received greater support from empirical analyses (e.g., Adelman, 1994a, 1994b; Rushing et al., 1992; Thoits, 1986). The findings of this study generally tend to support the role-accumulation hypothesis for women. Women who were involved in VW, in addition to UWH, and women who held all three productive roles reported fewer depressive symptoms than those who did only UWH. Although PW and VW might not have independent effects on the depressive symptoms of women, engaging in multiple productive roles might be related to fewer depressive symptoms in comparison with doing only UWH.
For men, in contrast, we could not find any evidence to show that the level of depressive symptoms decreased with an increasing number of productive roles. Although engaging in PW was associated with fewer depressive symptoms of men compared with having no role and doing only UWH, engaging in multiple productive roles in addition to PW was not necessarily related to fewer depressive symptoms compared with doing only PW. This finding would be useful information for public health programs and policies designed to maintain well-being in later life, because it tells them that just having a specific productive role such as PW or VW may be sufficient to exert a positive effect on a person's well-being.
Third, we examined the relationship between changes in PW status and depressive symptoms. The findings suggest that a loss of the PW role is related to higher levels of depressive symptoms for late-middle-aged Japanese men but not for late-middle-aged Japanese women. Researchers suggest that many Japanese men tend to find their ikigai in PW (Bass, 1996; Shibata, 1998, 2002). Because of this work ethic in Japanese men, a loss of the PW role might lead to psychological distress. In addition, the mandatory retirement system in Japan may also be associated with higher levels of depressive symptoms among men who lost their PW role, because individuals are forced to quit work and have no say in the matter.
Most prior studies investigating how changes in PW status might impact psychological well-being focused on men's retirement, with little investigation of women's experiences. This study's findings suggest that the relationship between changes in PW status and depressive symptoms may vary by gender in Japan. Because of gender differences in expectations, opportunities, and experiences for PW, women of this generation may not be as severely affected by the loss of PW status as men. However, women's participation in the workforce has been increasing. Such demographic trends warrant further attention in future research. Changes in PW status may soon become more important for women's depressive symptoms, as is presently the case with men.
Finally, we examined whether the relationship between loss of PW status and depressive symptoms might be influenced by taking on VW. The results suggest that VW can protect retirement-aged men from the negative effects of lacking PW status. Using the concept of role context, researchers have suggested that VW might have differential benefits on well-being according to subgroup variations within the older adult population. For example, Musick and colleagues (1999) found that taking on VW provides a greater protective effect on mortality for older adults with lower levels of social contact than for those with higher levels. Because PW tends to be associated with increased social contact, loss of PW status may lead to decreased social contact, especially for men. Taking into account the findings of Musick and colleagues, we find that VW may reduce the negative impacts of decreased social contact that often accompany the transition to retirement among men. Meanwhile, for women, loss of PW status may not be associated with lower social contact because women often tend to have a lot of informal social contact outside their PW sphere. Such gender differences in social contact may explain our findings, in which no association between VW and depressive symptoms was found in retirement-aged women in Japan.
However, there was a limitation in the way VW was analyzed in this study. Although previous studies on VW and well-being in the United States have often investigated the effects of formal VW, this study could not distinguish between formal and informal VW because of the relatively small number of people who engage in formal volunteerism in Japan. Volunteerism in Japan has started to increase over the past decade, so further investigation on the relationship between VW and the well-being of older adults is needed.
This study was also limited by the fact that it does not investigate the context of productive roles and the individual situation in much detail. As the diversity of the older population increases, research on productive roles and well-being is required to specify these relationships within more detailed contexts of the various roles. Furthermore, the study's sample was limited to late-middle-aged adults, that is, those aged between 55 and 64 years at baseline and between 59 and 68 at Wave 3, because the research from which this study obtained data focused on the retirement transition. The relationships between productive roles and well-being may differ for adults aged 70 and older.
Nonetheless, this study contributes to the growing body of knowledge on late-life roles and well-being from the cross-cultural and role-context perspective. This study also has the advantage that it used three waves of a longitudinal data set to control for preexisting depressive symptoms. Consistent with previous studies in the United States, our findings suggest that being involved in either PW or VW can protect retirement-aged men from increased levels of depressive symptoms in Japan. However, the findings also demonstrate that neither all types of productive roles are associated with reduced depressive symptoms, nor are they associated with depressive symptoms for women. Further research is needed to address the limitations of this study and to develop a more precise understanding of gender differences on the impact and meaning of productive roles.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication May 31, 2007. Accepted for publication December 21, 2007.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
