| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Glennan Center for Geriatrics and Gerontology, Eastern Virginia Medical School, Norfolk.
2 Polisher Research Institute (formerly Philadelphia Geriatric Center), North Wales, Pennslyvania.
Address correspondence to Maximiliane Szinovacz, Glennan Center for Geriatrics and Gerontology, Eastern Virginia Medical School, 825 Fairfax Ave., Hofheimer Hall, Rm. 201, Norfolk, VA 23507-1912. E-mail: maxres{at}visi.net
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods: The analyses rely on Waves 1–4 of the Health and Retirement Survey (N = 2,649). Generalized estimating equations models with bootstrapped standard errors and adjustment for survey design and non-independence of dyad members estimate effects of retirement, type of retirement, and spouse's disability on depressive symptoms, controlling for relevant covariates.
Results: The results suggest that depressive symptoms increase when retirement is abrupt and perceived as too early or forced. Women retirees who stopped employment and were either forced into retirement or perceived their retirement as too early report significantly more depressive symptoms with increasing spouse activities of daily living (ADLs) limitations. There is no similar effect for men. In contrast, for working retirees who retired on time, depressive symptoms decrease with increasing spouse ADLs.
Discussion: These results highlight the importance of retirement context on postretirement well-being. They suggest that both type of retirement transition and marital contexts such as spouse's disability influence postretirement well-being, and these effects differ by gender.
Theories pertaining to effects of retirement on well-being have focused on the extent and type of postretirement activities and life styles (Kosloski, Ginsburg, & Backman, 1984
). We argue that the circumstances of the retirement transition itself are also important for well-being, an assumption derived from the life course perspective (Elder, 1995; Moen, 1996). According to the life course perspective, life experiences are linked over time and across spheres. Consequently, past experiences guide current actions as well as the perception of their outcomes, and experiences in one realm (e.g., work/retirement) are tied to experiences in other life domains (e.g., family). For example, some research has shown that care obligations can impinge on retirement decisions (Blau, 1998; Hayward, Friedman, & Chen, 1998; Szinovacz & DeViney, 2000). This implies that the specific circumstances under which retirement occurs (including family circumstances such as care obligations) will help determine how retirement affects well-being. We address this assumption by investigating whether one specific context, namely, spouse's disability, moderates the influence of retirement on depressive symptoms.
Another crucial assumption of life course theory pertains to the timing of life transitions. Transitions can be on-time or off-time either in relation to social norms (e.g., norms about appropriate retirement age) or in terms of individuals' life plans. Life transitions that are off-time tend to undermine well-being (Hagestad, 1990). Although norms about retirement age have become more flexible (Henretta, 1997), many individuals still hold expectations about when retirement should occur (Settersten & Hagestad, 1996). We pay tribute to this assumption by exploring whether perceptions of too early retirement per se or in conjunction with spouse's disability moderate the effect of retirement on depressive symptoms.
Other assumptions guiding our analyses are grounded in control and stress theories. The association between control and well-being has been well established. Evidence suggests that individuals who are able to exert control over their environment and/or derive a sense of control from their actions enjoy enhanced well-being, whereas lack of control reduces well-being (Schulz & Heckhausen, 1997). When life transitions such as retirement occur under circumstances that leave individuals little choice over the transition, they may lead to reduced well-being. For example, earlier studies showed that mandatory retirement is associated with lower retirement satisfaction (Hardy & Quadagno, 1995). Retirement due to spouse's disability may be viewed as beyond individuals' control and thus reduces well-being. We pursue this issue by assessing whether perceptions of forced versus wanted retirement per se, and in conjunction with spouse's disability, moderate the influence of retirement on depressive symptoms.
Both caregiving and retirement can be viewed as potentially stressful life events. How stressful transitions out of a role (such as retirement) are is contingent on whether or not the role was salient and identity enhancing or whether it constituted a burden (Thoits, 1991; Wheaton, 1990). For example, employment can provide caregivers with resources, social supports, and increased opportunities for tension relief, but it can also enhance stresses associated with role overload and burden (Dautzenberg, Diedriks, & Philipsen, 2000; Reid & Hardy, 1999; Waldron, 1998). In addition, gradual exits from roles may be less identity threatening and thus less stressful than abrupt transitions. Lacking data on perceptions of stress, we address this topic indirectly by investigating whether gradual (some work after retirement) versus abrupt (no employment after retirement) retirement per se, and in conjunction with spouse's disability, moderates the influence of retirement on depressive symptoms.
Another assumption integral to life course theory is the acknowledgment of diversity in life experiences among diverse population subgroups (Calasanti, 1996). We focus on gender differences because past research suggests that both retirement and caregiving experiences differ by gender. Several studies suggest that women are more stressed by caregiving than men (Yee & Schulz, 2000). Findings concerning the overall impact of retirement on men's and women's well-being have been inconsistent. Some studies indicated that either men or women experience more problems, whereas others found no gender differences in overall adjustment (Carp, 1997; Richardson & Kilty, 1995). These inconsistencies suggest that the main effect of gender on retirement adaptation is weak. Most likely this is because selected retirement contexts moderate or mediate gender differences in retirement experiences. For example, marriage seems to benefit male retirees more than female retirees (Calasanti, 1996). We pursue this line of investigation by addressing the linkages among types of retirement transitions (forced, early, abrupt) and spouse disability in their effects on postretirement depressive symptoms and by exploring whether these effects vary by gender. Previous research sheds little light on this issue. Studies addressing the impact of caregiving on retirement yield contradictory results. Some indicate that men are more prone to retire if their wives are ill or require care (Hayward et al., 1998; Szinovacz & DeViney, 2000), whereas others show a stronger effect of care responsibilities on women's retirement (Dentinger & Clarkberg, 2002). Research assessing caregiving effects on postretirement well-being is also limited. Vinick and Ekerdt (1991) noted that caregiving "spoils" retirement but did not differentiate by gender. Others found lower retirement adaptation among women retiring for caregiving purposes or whose spouses are ill (Haug, Belgrave, & Jones, 1992; Szinovacz, 1989), but no similar effect for husbands (Haug et al., 1992; Szinovacz, 1992).
We suggest that one reason for inconsistencies in previous research on gender, retirement, and caregiving derives from the fact that both the circumstances and the meaning of retirement transitions in the context of spouse disability differ by gender. Because of women's socialization as kin keepers (Rossi & Rossi, 1990), they may feel more obligated than men to retire if their husbands are ill and thus perceive less control over the retirement transition. Indeed, other analyses on the same HRS subpopulations (Szinovacz & Davey, in press) indicate that retired men rarely left the labor force for caregiving purposes and that women, but not men, were more prone to perceive their retirement as forced if their spouses were disabled. Furthermore, many women approach retirement with disrupted work histories due to earlier family obligations (Moen, Robison, & Fields, 1994; Pienta, 1999). Consequently, women may be more likely than men to perceive retirement in conjunction with husband's disability as off-time, an assumption corroborated by preliminary bivariate analyses indicating that spouse's disability is significantly related to women's but not men's perceptions of too early retirement. Last, the caregiving experience per se varies by gender. Men tend to view caregiving for spouses as an extension of their marital role and generally provide less hands-on care than women. Women, on the other hand, tend to view caregiving as an obligatory role within the context of kin care and tend to spend more time with hands-on caregiving tasks (Allen, Goldscheider, & Ciambrone, 1999; Miller & Guo, 2000; Spitze & Ward, 2000). Care experiences that are perceived as obligatory and involve long care hours have been linked to heightened burden or stress (Chappell & Reid, 2002; Cicirelli, 1993). In addition, women more than men exhibit psychological and emotional reactivity to caregiving experiences in general (Atienza, Henderson, Wilcox, & King, 2001; Stoller & Cutler, 1992) and more negative feeling toward the disabled spouse and the changed marital reality (Stoller & Cutler, 1992; Wallsten, 2000). Such reactions can also lead to heightened caregiver stress.
Assumptions and Hypotheses
The arguments presented above lead to the following assumptions:
From these assumptions, we derive our main hypotheses, which link the effects of retirement transition types (forced, early, abrupt) with effects of spouse's disability.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). Analyses pool persons who retired between Waves 1 and 4 to achieve a sufficient number of cases for our complex analyses. Thus, we use individuals who retired either between Waves 1 and 2 or between Waves 2 and 3 or between Waves 3 and 4. In each case, the earlier wave data serve as baseline and the later wave data as basis for the construction of change and outcome variables. Thus, for individuals who retired between Waves 1 and 2, Wave 1 was used for baseline and Wave 2 for outcome measures, whereas for individuals retiring between Waves 2 and 3, Wave 2 measures serve as baseline and Wave 3 data as outcome. Individuals who did not retire between Waves 1 and 4 serve as reference. These individuals were randomly assigned to a baseline/outcome wave (Wave 1/2 or Wave 2/3 or Wave 3/4) to reduce potential wave effects. Specifically, we drew three random samples of continuously employed individuals proportionally to the distribution of retirees from Waves 2, 3, and 4. The group corresponding proportionally to Wave 2 retirees was then assigned baseline measures from Wave 1 and outcome measures from Wave 2; the group corresponding proportionally to Wave 3 retirees was assigned baseline measures from Wave 2 and outcome measures from Wave 3, etc. We also control for wave in the analyses.
For the subsample used in our analyses, we first identified all HRS individuals aged 50 or over (main respondents or main respondent spouses) who were employed 
10 hr/week at baseline and did not define themselves as retirees. From this subsample, we then selected persons who were either continuously employed throughout all four waves or who self-defined as completely or partly retired at the outcome wave (the HRS retirement measures are discussed below). The exclusion of Time 1 nonemployed/retired persons was necessary because we would lack baseline data for this population. We also excluded individuals working fewer than 10 hr/week at Time 1. This population may be underemployed or already in postretirement bridge jobs. Sensitivity analyses indicate that the results remain unaltered if only persons working 20 hr are used. Because our analyses focus on spouses' disability, the sample was further restricted to persons who were married at baseline and reported no change in marital status or partners at Time 2 and whose partners did participate in the surveys. All analyses were restricted to persons aged 51 and older (respondents and spouses) who had valid individual weights. This procedure ensures that only respondents from the representative HRS sample are included in our subsample. The final sample consists of 2,649 respondents. It should be noted that current retirees in the HRS tend to have retired early (mean retirement age in our subsample is 60.3 years for men and 58.86 years for women) compared with the estimated national average of 62.0–62.6 years for men and 61.4–62.5 years for women (Gendell, 2001). Thus, our data may not be fully generalizable to all retirees. Missing values were imputed through multiple imputation (Schafer, 1997). HRS or RAND imputed variables were preserved.
Measures
The main dependent variable is depressive symptoms. Depressive symptoms were measured with eight items from the Center for Epidemiological Studies—Depression (CES-D) Scale (see Radloff, 1977). The CES-D answer categories differed between the first and subsequent waves (Steffeck, 2000). In response to the item stem "Much of the time during the past week you felt ... (depressed, blue, etc.)," answers categories in wave 1 were 0 = rarely/none of the time, 1 = some of the time, 2 = most of the time, 3 = all/almost all of the time, whereas in subsequent waves, responses were coded as yes = 1 or no = 0. Because a subsample of Wave 2 respondents received both coding versions of the CES-D scale, we were able to construct a rescaled Wave 1 measure (for a detailed description of this procedure, see Szinovacz & Davey, 2004). Scores could range from 0 to 8. We include CES-D scores at baseline as control in the analyses, so that results reflect influences on depressive symptoms at Time 2 net of Time 1 depressive symptoms.
The independent variables for the analyses are retirement, type of retirement (forced, early, abrupt), and spouse's disability. The retirement measure was based on the question: "Do you consider yourself partly retired, completely retired, or not retired at all?" Because continuously employed individuals have, by definition, 0 values on all retirement-type variables, we created dummy variables for different types of retirement transitions. We first distinguish between working and nonworking retirees based on the question: "Are you currently working for pay (yes/no)?" These two groups were further differentiated by forced versus wanted and separately by early versus on-time retirement. Early versus on-time retirement was determined by contrasting respondents' planned and actual retirement age. Those retiring earlier than planned were coded 1, and all others were coded 0. The forced-versus-wanted distinction was derived from the question of whether retirement "was something you wanted to do or something you felt you were forced into." The answer categories were "wanted to," "forced into," and "part wanted, part forced." Because the last category was not implied in the question, relatively few individuals chose that answer. We thus created one dummy variable where those reporting exclusive "forced" retirement were coded 1 and the other two groups were coded as 0. For the analyses, we combine employment status and type of retirement into a set of dummy variables. We thus have four dummy variables: fully (nonworking) retired with forced retirement, fully retired with wanted retirement, working retiree with forced retirement, and working retiree with wanted retirement. The same procedure was used for early versus on-time retirement. Continuously employed persons who self-defined as not retired serve as reference in the analyses. Thus, our analyses demonstrate whether individuals retiring under specific circumstances fared better or worse than those who did not retire.
For spouse's disability, we rely on a measure of activity-of-daily-living (ADL) limitations. ADL limitations were assessed for 16 tasks such as walking several blocks, walking across a room, getting up from a chair, and getting in and out of bed and were coded as 1 = any difficulty, 0 = no difficulty. Scores could range from 0 to 16. Changes in ADLs were assessed with change scores where positive scores reflect increases in limitations. Because we do not know when spouses became disabled relative to respondent's retirement and in order to reduce collinearity between level of and change in functional limitations, level of spouse's disability was averaged over the two waves. In addition, both level of spouse's ADLs and change in spouse's ADLs were centered for inclusion in interaction terms (Aiken & West, 1991).
Controls
We attempted to include variables that have been shown most consistently to influence postretirement well-being in the analyses, provided data were available. The first set of controls comprised demographic background variables, namely, gender (1 = female, 0 = male), age (in years), education (in years), race (White is reference), and whether respondents have dependent children in the household (1 = yes, 0 = no).
Foremost among other predictors of postretirement well-being are health and economic situation (Bossé, Aldwin, Levenson, & Workman-Daniels, 1991; Calasanti, 1996; Hardy & Quadagno, 1995). For respondents' health, we use five indicators. Self-rated health was measured with a single item: "Would you say your health is ...?" Answer categories ranged from "poor" (1) to "excellent" (5). Limitations in ADLs were assessed the same way as for spouse's disability. Changes in health were assessed with simple change scores, that is, the difference between Time 1 and Time 2 scores. Positive change scores indicate an increase in self-rated health. We also included whether respondents ever had psychiatric problems (1 = yes, 0 = no) measured at baseline.
Four measures were used to capture financial status. Change in couples' income between waves was based on RAND constructed household income variables and includes income of both spouses from all income sources. Net worth reflects couples' assets minus their debts, measured at baseline. Household income at baseline and changes in net worth had no effect on depressive symptoms and were dropped from the final models. Both income and assets were recoded into $1,000 groups and truncated to reduce skewness.
As postretirement well-being may be linked to individuals' work commitment and job stress, we included indicators for this concept, namely, number of years in the current or last job (this is the best work history variable available in the HRS) and perception of the importance of work. The latter variable was derived from the question of whether individuals considered work mostly important as a source of income or valued their work in itself (3 = work itself is important, 2 = both are important, 1 = money important). Job stress is based on two items, namely, whether the job has become more difficult and whether the job involves stress (1 = strongly disagree, 4 = strongly agree; Cronbach's 
for the full sample ranges from.93 to.95 over the four waves). Jobs requiring much physical effort may be stressful to older workers and thus undermine their well-being. Respondents were asked whether their jobs required lots of physical effort, lifting heavy loads, or stooping, kneeling, or crouching. Answer categories for these items ranged from 4 = almost all of the time to 1 = none or almost none of the time. Scores for the three items were combined ( ranges from.93 to.94 across waves).
Several studies further indicate that spouse's employment status influences retirees' well-being (Moen, Kim, & Hofmeister, 2001; Myers & Booth, 1996; Szinovacz, 1996; Szinovacz & Davey, 2004). We thus control for spouse's employment status (1 = employed at time 2, 0 = not employed). Nonemployed spouses include retirees as well as the unemployed and homemakers. The wave from which individuals were drawn was also controlled to adjust for potential cohort effects. Means and standard deviations of all variables are shown in Table 1.
|
) for each imputed data set. Results for each imputation were then combined through the standard process of multiple imputation inference (Schafer, 1997). Bias in parameter estimates and standard errors was generally quite small, and plots and tests indicated that the bootstrapped estimates were essentially normally distributed and that the results were robust to violations of this assumption. Population-based weights could not be included in the current analyses due to software limitations; however, the main variables used in constructing the weights (i.e., age, gender, etc.) were included in all models, and so little bias in parameter estimates is expected as a result. Multicollinearity appeared to pose little problem in these data, with the variance inflation factor < 1.40 overall for the baseline models and < 2.00 for any single variable. We first estimated baseline models that included all independent and control variables but no interaction terms. Additional models were then estimated to assess the moderating effect of type of retirement in conjunction with spouse's disability. The final models exclude nonessential (i.e., not needed as part of the specification of higher-order interactions) nonsignificant interaction terms. |
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Our two main hypotheses addressed the combined effects of type of retirement and spouse's disability on depressive symptoms by gender. For full retirees, we find significant positive three-way interaction effects of Female x Type of retirement x Spouse's disability on depressive symptoms. These interactions are shown in Figures 1 (forced retirement) and 2 (early retirement). As indicated in the figures, women full retirees (those who stopped employment) who were either forced into retirement or perceived their retirement as too early report significantly more depressive symptoms with increasing spouse ADLs. Although a similar trend occurs for women who retired fully but not early, this effect is much less pronounced and not significant (see Figure 2 and Table 1, second column). In contrast, spouse's ADLs are not related to depressive symptoms among men whose full retirement was forced or early nor among employees of either gender.
|
|
|
Although we presented no hypotheses for the controls, a brief note on other predictors of depressive symptoms is warranted. We find expected relationships between depressive symptoms and other indicators of respondents' physical and mental health. Hispanic individuals report more symptoms than non-Hispanic Whites. Individuals whose spouses are employed and those who still have dependent children in the household score higher on depressive symptoms. Respondents with long job tenure and those working more for nonfinancial reasons tend to report fewer depressive symptoms, whereas those in physically demanding jobs report more depressive symptoms. The wave from which individuals were drawn is negatively related to depressive symptoms, suggesting that individuals drawn from the latest cohort (Wave 4) have more depressive symptoms.
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
Interpretation of these findings is rendered difficult because of data limitations. Most importantly, we had to rely on spouse disability as an indicator for potential caregiving responsibilities. However, spouse disability may have other ramifications that could explain some of our findings. For example, disabled spouses may provide less help around the house, or they may have left the labor force unexpectedly and their discontent with such transitions or with their illness itself may impinge on the well-being of the partner (Haug et al., 1992). Later waves of the HRS (Wave 4 and thereafter) provide some information on spouse care, so that it will be possible to revisit our findings with caregiving data.
Another limitation pertains to the implied causality between spouse's disability and perceptions of forced and early retirement. Although we could confirm positive associations between spouse's disability and perceptions of forced (Szinovacz & Davey, in press) or early retirement for women, we lack direct data on the reasons for early or forced retirement. Only additional research that includes reasons for specific retirement transitions can shed light on this issue.
As is true for previous research, it is also not clear whether the gender disparity in the influence of spouse's disability reflects differences in men's and women's perceptions of and adjustment to caregiving, in the caregiving situation itself (e.g., extent and difficulty of care), or in the expression of emotions (Lutzky & Knight, 1994). We lack measures pertaining to the caregiving situation (beyond the number of spouse's ADLs used as the independent variable). However, it would be possible to explore our hypotheses with other health measures available in the HRS.
At the most general level, our findings confirm the assumption that the effect of retirement on well-being depends on contextual features surrounding the retirement transition. This finding is in line with earlier studies (Hardy & Quadagno, 1995; Myers & Booth, 1996; Szinovacz & Davey, 2004) that also showed that such contextual influences (e.g., involuntary retirement or marital contexts) can moderate retirement effects on well-being. It casts doubt on research findings that merely contrast workers with retirees (Charles, 2002). Such studies neglect heterogeneity among retirees and are thus misleading. In light of our results, more research should be targeted at exploring other contextual features that can enhance or reduce postretirement well-being. The integration of assumptions from life course, control, and stress theories may be particularly useful for such ventures.
The specific theoretical arguments underlying our analyses receive mixed support. The timing hypothesis derived from the life course perspective (Assumption 1 and Hypothesis 1) receives some support. Early retirement is associated with increased depressive symptoms. In addition, abrupt early retirement moderates the effect of spouse's disability on women's but not men's depressive symptoms. We contend that this gender difference derives from the fact that wives are more prone than husbands to be pushed into premature retirement for caregiving reasons. They may resent both the premature retirement as well as their caregiving obligations, and this situation may be rendered even more depressing by impingement of spouse's disability on enjoyable retirement endeavors (spoiled retirement). Men, on the other hand, rarely retire for family care reasons. For men who retire early for other reasons (e.g., their own health limitations or job displacement), care for disabled wives may provide a sense of purpose and a source of self-esteem. Indeed, earlier studies suggest that men often view caregiving responsibilities as a labor of love and thus a worthwhile postretirement endeavor (Szinovacz, 1992; Thompson, 2002).
The control hypothesis (Assumption 2 and Hypothesis 1) receives similar support as the timing assumption. Both men and women who perceive their retirement as forced report more depressive symptoms. Although earlier research also supports the control argument (Hardy & Quadagno, 1995), it is important to note that our measure of forced versus wanted retirement is retrospective. It is conceivable that unhappy or depressed retirees come to define their retirement as involuntary. Thus, future studies need to address the causality of these relationships. Forced retirement also moderates the impact of spouse's disability among women, most likely due to the same mechanisms as those applicable to the timing hypothesis above.
There is considerable support for the stress hypotheses (Assumption 3 and Hypothesis 1). Continued employment after retirement seems to protect retirees both from the potential negative effects of forced or untimely retirement as well as from those of spouse's disability. Apparently, the benefits of role enhancement (Reid & Hardy, 1999) derived from the worker role outweigh negative aspects of the retirement transition. Furthermore, as retirees' work efforts are considerably diminished (they work about 10 hr less than continuously employed workers), they will experience less overload problems resulting from the combination of work and spouse care while still enjoying the benefits of role enhancement. It is not clear, however, why working retirees should benefit from spouse disability, as our findings suggest. It is conceivable that retirees derive a special sense of purpose and self-efficacy from the combination of work and caregiving without the stresses of overload problems; that is, finding an acceptable solution to the competing demands of work and spouse care requirements may be in itself gratifying. Obviously, more research is needed to disentangle these complex relationships.
Further research is also needed to address other limitations of this study. First, the early waves of the HRS rely on a special subgroup of retirees, namely, those retiring relatively early. Consequently, our results may not be generalizable to individuals retiring closer to the institutionalized retirement age (i.e., age of full Social Security eligibility). It is possible that women who retire for caregiving reasons may experience fewer problems if their retirement is closer to the institutionalized time of retirement. For example, support networks may differ for early and late retirees as many of early retirees' friends may still be in the labor force. In addition, we lack some predictors that may moderate or mediate the joint effects of spouse's health and retirement. Foremost among them are a complete work history (the HRS provides only information on the last job), which seems particularly important for women's retirement decisions and distress (Davies, 1999). Another important predictor of depression missing in our analyses owing to lacking data is social network support (Krause, 2001), but other factors may play a role as well.
This study provides directions for theoretical development in retirement research and for interventions and policy. Theoretically, it demonstrates the importance of contexts and linked lives for understanding retirement adjustment processes. The circumstances surrounding the retirement transition influence postretirement well-being, and these circumstances are not restricted to individuals' personal characteristics but include the retirement transition process as well as the life situation of spouses and, perhaps, of other family members. From a programmatic and policy perspective, we provide evidence that premature or unwanted retirement can undermine well-being. Thus, measures to reduce job displacement and increased attention to preventative health care among older workers will benefit retirees and increase chances of longer labor force participation among older workers (at least until they become eligible for Social Security benefits). On the other hand, gradual retirement seems to protect retirees against the negative impact of forced or off-time retirement. This suggests the need to increase part-time job opportunities for retirees. Our results further indicate that leaving the labor force for caregiving reasons can undermine well-being even among women who are close to retirement age. Health care professionals and counselors may be more inclined to recommend that wives leave the labor force to care for ill spouses if wives are close to retirement age. Our results caution against such recommendations. Instead, more efforts should be directed toward enabling caregiving women of all ages to remain in the labor force if they so desire.
|
Acknowledgments |
|---|
The Health and Retirement Survey is sponsored by the National Institute on Aging and conducted by the University of Michigan. Detailed information on the data set is available from their website (http://www.umich.edu/
hrswww/).
|
Footnotes |
|---|
Received for publication February 18, 2004. Accepted for publication June 15, 2004.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  Y. Sugihara, H. Sugisawa, H. Shibata, and K. Harada Productive Roles, Gender, and Depressive Symptoms: Evidence From a National Longitudinal Study of Late-Middle-Aged Japanese J. Gerontol. B. Psychol. Sci. Soc. Sci., July 1, 2008; 63(4): P227 - P234. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. L. Christ, D. J. Lee, L. E. Fleming, W. G. LeBlanc, K. L. Arheart, K. Chung-Bridges, A. J. Caban, and K. E. McCollister Employment and Occupation Effects on Depressive Symptoms in Older Americans: Does Working Past Age 65 Protect Against Depression? J. Gerontol. B. Psychol. Sci. Soc. Sci., November 1, 2007; 62(6): S399 - S403. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. T. Gallo, E. H. Bradley, J. A. Dubin, R. N. Jones, T. A. Falba, H.-M. Teng, and S. V. Kasl The persistence of depressive symptoms in older workers who experience involuntary job loss: results from the health and retirement survey. J. Gerontol. B. Psychol. Sci. Soc. Sci., July 1, 2006; 61(4): S221 - S228. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
