| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a Department of Clinical and Social Sciences in Psychology, University of Rochester, New York
b Department of Human Development, Cornell University, Ithaca, New York
Jungmeen E. Kim, University of Rochester, 187 Edinburgh Street, Rochester, NY 14608 E-mail: Jungmeen{at}psych.rochester.edu.
Decision Editor: Margie E. Lachman, PhD
 |
Abstract |
|---|
 TOP Abstract Theoretical BackgroundA Life-Course, Ecological ModelMethodsResultsDiscussionReferences |
|---|
Although retirement has been considered one of the most important later life status transitions (Szinovacz 1980
), our knowledge of its psychological consequences is fragmentary. For most older persons, retiring from their primary career jobs is a milestone, marking passage into the later stages of adulthood. It is not simply an objective life course transition, but is also a subjective developmental and social–psychological transformation that may be related to physical and psychological well-being (Moen 2001). On the one hand, the retirement experience may promote a sense of well-being, as workers move out of demanding and/or stressful career jobs. On the other hand, the retirement passage itself may lead to diminished well-being, as individuals lose their occupational attachments, their social network of coworkers, and a major anchor for their identities.
The empirical evidence thus far is inconsistent. Some researchers have identified a significant negative association between retirement and life satisfaction or morale (Elwell and Maltbie-Crannell 1981; Walker, Kimmel, and Price 1981) and a positive association with psychological distress (Bosse, Aldwin, Levenson, and Ekerdt 1987; Bosse, Aldwin, Levenson, and Workman-Daniels 1990). Other researchers have found no deleterious psychological effects associated with retirement, reporting mixed findings or no relationship between retirement and life satisfaction (Gall, Evans, and Howard 1997; Lee 1978; Palmore, Fillenbaum, and George 1984; Stull 1988) or physical health (Palmore et al. 1984), no relationship between retirement and psychological distress (Ross and Drentea 1998; Wright 1990), and even a positive effect of retirement on health or reduced stress level (e.g., Ekerdt, Bosse, and LoCastro 1983; Kasl 1980; Midanik, Soghikian, Ransom, and Tekawa 1995).
Earlier studies suffer from two general weaknesses. First, many focused almost exclusively on men's retirement, with little investigation of women's experiences or the fact that couples are increasingly undergoing two retirements: his and hers. Recent demographic trends—including increases in women's workforce participation, along with increased longevity and decreased retirement age—point to the importance of considering retirement as a "couple phenomenon" (see Henretta, O'Rand, and Chan 1993; Szinovacz, Ekerdt, and Vinick 1992). Many spouses do not retire simultaneously, resulting in various combinations of employment and retirement within couples (e.g., both spouses are retired, both spouses are still in their career jobs, one is still in primary career job while the other is retired). This points to the importance of incorporating the diversity of both spouses' circumstances into models of retirement and psychological well-being.
Second, most prior studies of retirement have been cross-sectional in nature. But differences in psychological well-being by retirement status at one point in time may or may not reflect changes in psychological well-being as individuals move from employment to retirement over their life course. Dynamic, longitudinal analyses are essential to assess whether snapshot distinctions between groups (of retirees and nonretirees) capture any effects of the actual retirement transition. In this article, we draw on longitudinal data of late-midlife workers and retirees (and their spouses) to examine the dynamic relationship between couples' work or retirement status and psychological well-being.
|
Theoretical Background |
|---|
|
TOPAbstractTheoretical BackgroundA Life-Course, Ecological ModelMethodsResultsDiscussionReferences |
|---|
Role theory.
Role theory (e.g., George 1993), in conjunction with a life-course perspective (e.g., Elder 1995; Moen 2001), provides one of the most commonly used explanations of adjustment to retirement. But, in point of fact, role theory lends itself to several life-course–related formulations. From the role enhancement perspective, men and women who retire from their career jobs are vulnerable to feelings of role loss, which can lead to psychological distress. This vantage point postulates that employment is central to one's identity; loss of this critically important work role, and/or the environmental loss accompanying retirement, produces concomitant decline in morale and an increase in depressive symptoms. Alternatively, retirement from the demands of one's primary career job may be a major life-course–role exit that serves to reduce role strain and overload, thereby enhancing psychological well-being. This role-strain reduction theoretical perspective also points to the importance of role quality (e.g., Vandewater, Ostrove, and Stewart 1997), suggesting that circumstances around employment, retirement, and the transition from one to the other should matter.
Continuity theory.
Continuity theory (e.g., Atchley 1989; Richardson and Kilty 1991) proposes that people tend to maintain earlier lifestyle patterns, self-esteem, and values, even as they exit their primary career jobs. Therefore, retirement need not lead to maladjustment and distress. Consistent with this perspective, one early panel study by George and Maddox 1977 found remarkable stability in life satisfaction among 57 older men over a 5-year period. The sample, however, combined both those who were continuously retired and those who retired between Time 1 and Time 2. Consequently, the results may not precisely reflect the impact of actual retirement transitions on life-satisfaction changes.
|
A Life-Course, Ecological Model |
|---|
|
TOPAbstractTheoretical BackgroundA Life-Course, Ecological ModelMethodsResultsDiscussionReferences |
|---|
, Kim and Moen 2001b). The life-course approach (Elder 1995) highlights the dynamic processes of development and change over the life span. The ecology of human development (Bronfenbrenner 1995) and role context (e.g., Moen 2001; Musick, Herzog, and House 1999) perspectives suggest locating transitions in the social contexts of other roles, relations, and developmental processes. In viewing retirement as a life-course, ecological transition, we focus on process, the interdependency of linked lives, and context.
Process.
Process connotes both the dynamics of moving into retirement and possible mechanisms through which retirement status might affect psychological well-being. We developed a life-course model of the retirement transition, examining Time 2 psychological well-being, controlling for the same measure 2 years earlier, and considering three groups: the stably employed in their primary career jobs, the stably retired, and those who moved from their career jobs to retirement between two waves of the survey.
Process also suggests that particular mechanisms may facilitate the psychological well-being of those either in or moving into retirement. Research evidence points to three possible mechanisms linking retirement status and psychological well-being: (a) economic resources, (b) personal resources, and (c) social–relational resources. Prior studies have found that inadequate incomes and financial problems predict dissatisfaction and maladjustment in retirement (Atchley 1976b; Gallo, Bradley, Siegel, and Kasl 2000; Hendrick, Wells, and Faletti 1982; Richardson and Kilty 1991). Perceptions of having an adequate income are also related to retirement adjustment (Kim and Moen 1999).
Personal resources include not only sociodemographic status but also health and personality variables. A substantial body of research has shown that health is positively correlated to retirement adjustment (e.g., Shaw, Patterson, Semple, and Grant 1998). Personality characteristics also play a crucial role in retirement adaptation (e.g., Robbins, Lee, and Wan 1994). The most important individual difference mediating or moderating the retirement–well-being linkage may well be a sense of personal control. As people move through late midlife they may experience a decline in personal control, given that the environmental events accompanying aging limit the range of outcomes that are actually attainable (Rodin 1990; Weisz 1983). The relationship between retirement and a sense of personal control, however, is not at all clear. Some scholars presume that retirement may mean the loss of one of the most meaningful sources of instrumental control in life, the work role (e.g., Ross and Drentea 1998; Schulz 1976). Others suggest that retirement may offer more time and opportunities for realizing one's goals and carrying out plans, thus increasing personal efficacy (e.g., Lachman 1986).
Evidence from a range of studies suggests that a sense of personal control has significant influences on retirement adjustment (Fretz, Kluge, Ossana, Jones, and Merikangas 1989; Hendrick et al. 1982; Mutran, Reitzes, and Fernandez 1997; Taylor-Carter and Cook 1995). For example, in a study of the relationships between self-efficacy and anxiety and depression about retirement, a low sense of self-efficacy was one of the best predictors of preretirement worry, in addition to concerns about money or health (Fretz et al. 1989). In their survey of individuals ages 50 to 70, Herzog, House, and Morgan 1991 found that those who stopped work and felt they had little or no choice reported lower levels of health and psychological well-being compared with those who voluntarily retired and those working the amount they would like. The third set of resources is social–relational, which we discuss in the next section.
Linked lives.
A key tenet of a life-course, ecological perspective is that lives are interdependent. Developmental processes always take place in the context of ongoing social relations, including developmental shifts that occur with retirement. Individuals frequently decide to retire on the basis of changes in others' health and/or retirement exits or plans. Furthermore, the retirement experience is played out in a network of shifting social relations. It is crucial to consider the work or retirement statuses of both spouses, given that each spouse's retirement transition represents an important life event for the couple, requiring adjustment on the part of both spouses.
Research on retirement adjustment suggests that marriage and family relationships serve as social–relational resources in retirement adjustment (George and Maddox 1977; Gove, Style, and Hughes 1990; Hendrick et al. 1982; Kessler and Essex 1982; Kilty and Behling 1985; Mutran et al. 1997; Vinick and Ekerdt 1989). Strong emotional support from a spouse may well help individuals overcome stressful life events, thereby sustaining their psychological well-being. For instance, Lee 1978 reported significant contributions of marital satisfaction to morale for older men and women (aged 60 and older), but the positive effect of marital satisfaction on morale was notably stronger for women than for men. Other research findings similarly indicate that women's psychological well-being is more heavily influenced by marital quality than is men's (Acitelli and Antonucci 1994; Dehle and Weiss 1998; Ross 1995).
Context.
To fully understand the links between retirement and psychological well-being, one must also consider the contextual factors in which retirement transitions are embedded. A key contextual consideration is gender, given that the retirement experience may well be a different process for women than it is for men. Because of their different work histories, employment opportunities, and general life experiences, women appear to adjust to retirement differently than men do. Specifically, there is evidence that women have more negative attitudes toward retirement than men do, and retirement is more disruptive and more likely to be linked with greater depression and loneliness for women than it is for men (Atchley 1976a; Streib and Schneider 1971). Some researchers have found gender differences in the predictors of retirement attitudes and satisfaction (e.g., Atchley 1982; Calasanti 1996; George, Fillenbaum, and Palmore 1984; Matthews and Brown 1987; Quick and Moen 1998; Seccombe and Lee 1986).
Prior levels (baseline) of psychological well-being are another key contextual consideration, with two possible models of their moderating effects. The first is the cumulation of advantage/disadvantage model (closely associated with continuity theory), with those already high on measure of psychological well-being experiencing the same level or even an increase following a recent transition to retirement. Correspondingly, the most disadvantaged would be the least likely to experience a positive change in psychological well-being concomitant a transition to retirement. Conversely, the role strain reduction perspective (Moen, Kim, and Hofmeister 2001) presumes that the least advantaged on a particular measure may well experience heightened psychological well-being in light of a movement out of (presumably stressful) career jobs and into retirement. From this point of view, becoming retired should reduce psychological distress and promote psychological well-being.
We examined possible moderating effects of gender and prior levels of psychological well-being on changes in psychological well-being for three groups of late-midlife individuals: those not yet retired from their career jobs, those who moved into retirement over the 2-year study period, and those who were stably retired. On the basis of the theoretical arguments offered, we proposed the following hypotheses:
|
Methods |
|---|
|
TOPAbstractTheoretical BackgroundA Life-Course, Ecological ModelMethodsResultsDiscussionReferences |
|---|
). The original sample consisted of 762 randomly selected individuals aged 50 to 72 years, including both retirees and workers who were not yet retired. The participants were recruited from six major upstate New York employers. Participants were randomly selected from lists provided by their employers and contacted by letter and telephone to request their participation and arrange for an interview. A total of 1,206 men and women were contacted, and 762 (63%) agreed to participate and completed the interview process.
The present study involves a subsample of married participants: 282 men (60%) and 176 women (38%). The average age of the subsample was 60 years (SD = 5.72) at the first interview (1994–1995), with a range from 50 to 72 years. Because of the demographics of the region, our sample consists of mostly White respondents (95%). Although retirement can be defined in many different ways (see Elder and Pavalko 1993), in the present study, retirement is defined as receiving a pension (or early retirement package) from a career employer and/or Social Security benefits. Our operationalization captures the exit from full-time, career employment. (Another way of defining retirement is as the final exit from the workforce. However, as significant numbers of retirees are taking on part-time jobs, it is difficult to define when an exit is indeed "final.")
The first wave of data was collected from 1994 to 1995. The majority of participants were interviewed in person with a structured interview, which took from 1 to 2H hr. Participants also independently completed a self-administered questionnaire. The survey instruments tapped issues relevant to employment history, retirement, health, activities, and psychological and attitudinal domains. The questionnaires in booklet form were adopted from various sources, including the Health and Retirement Survey (Juster 1992), the Quality of Employment Survey (Quinn and Staines 1979), and the Women's Roles and Well-Being Study (Moen, Dempster-McClain, and Williams 1992). The second wave of data was collected about 2 years later, from 1996 to 1997 (Time 2), and the third wave was collected from 1998 to 1999 (Time 3). At the second and the third waves, 94% and 91%, respectively, of the surviving adults in the original sample were reinterviewed. We primarily draw on Waves 1 and 2, but include as well those who retired between Waves 2 and 3 in our category of newly retired. Thus some newly retired respondents left their primary career jobs between 1994 to 1995 and 1996 to 1997 (26 men and 15 women); others did so between 1996 to 1997 and 1998 to 1999 (21 men and 18 women). Because analyses revealed no difference in effects between the two groups, we combined them as newly retired between Time 1 and Time 2.
The average age of retirement from their primary career jobs was 58 years for men and 59 years for women. At Time 2, there were 106 not-yet-retired respondents and 352 retirees. Eighty respondents (23% of the retirees) moved into retirement between Time 1 and Time 2. The men reported an average household income of $45,315 (mdn = $55,000) and the women reported a mean household income of $40,400 (mdn = $45,000). The health status of the participants was typically very good (as indicated by self-rating of health averaging 8.08 [SD = 1.91] for men and 7.93 [SD = 2.01] for women on a rating scale ranging from 0 [very serious health problems] to 10 [very best health]). Overall, the average length of marriage was about 31 years at the time of the first interview for both men and women. Note that this sample of married individuals may be more likely than the general population to be satisfied with their marriages because they represent the marital "survivors," those in late midlife whose marriages are longer in duration. That is, those most dissatisfied may have already left unhappy marriages and are, consequently, not in the married sample.
Measures
Primary independent variables: work/retirement status and transition.
The respondents were divided into three work/retirement-transition groups: (a) long-term retirees, who left their primary career employment before the first interview and thus were continuously retired at both Time 1 and Time 2 (181 men and 91 women), (b) newly retired individuals, who left their career jobs at some time between Time 1 and Time 2 (47 men and 33 women), and (c) not-yet-retired individuals, who had not left their career jobs and were continuously employed in these career jobs at both Time 1 and Time 2 (54 men and 52 women). The spouses were simply divided into those who were employed and those who were not employed at Time 2. This is because finer divisions according to spouses' retirement transitions created very small cell sizes in the couples' joint distribution.
Dependent variables: morale and depressive symptoms.
We used two measures of psychological well-being, at two points in time. First, the Philadelphia Geriatric Center (PGC) Morale scale (Lawton 1975) assesses a positive aspect of psychological well-being in late midlife adults. The scale was developed to measure the subjective psychological well-being of individuals of retirement age or older. We used a subset of seven items from the PGC Morale scale, which measures attitude toward one's own aging (e.g., "I am as happy now as I was when I was younger."). Items were answered "yes" or "no." The average correlation coefficient among the seven PGC Morale scale items was .23, with a range from .16 to .40, and all correlations were statistically significant (p < .01) at Time 1. Internal consistency (Cronbach 
) of the scale was .68 at Time 1. The test–retest reliability was .60 (p < .001).
Second, depressive symptoms were assessed with 12 items taken from the 20-item Center for Epidemiologic Studies–Depression scale (CES-D; Radloff 1977). The validity and utility of this scale for use with community-dwelling older adults has been demonstrated in previous studies (Berkman et al. 1986; Radloff 1977). The items of the CES-D scale ask about the ways the respondent has felt or behaved during the past week (e.g., "I felt bothered by things that usually don't bother me."). The average correlation coefficient among the 12 CES-D scale items was .39, with a range from .17 to .72, and all correlations were statistically significant (p < .01) at Time 1. The alpha reliability for this scale was .88 at Time 1. The test–retest reliability over 2 years was .58 (p < .001).
Contextual (mediating) variables: changes in income adequacy, subjective health, marital quality, and personal control.
We assessed changes in various resources from Time 1 to Time 2. Income adequacy is respondents' rank of the adequacy of their current incomes relative to their income needs, on a scale of 0 to 100, on which 0 represents completely inadequate, and 100 represents more than adequate. The subjective health measure is respondents' rating of their health status on a scale ranging from 0 (very serious health problems) to 10 (very best health). We measured marital quality using the sum of two items of marital satisfaction and marital-conflict frequency. First, marital satisfaction was assessed using the response to a single question, "Taking all things together, how satisfied are you with your marriage?" Respondents had a choice of five answers ranging from completely satisfied (5) to not at all satisfied (1). The measure of marital conflict was derived from the question, "How often would you say the two of you typically have serious disagreements or conflicts?" The choices were more than once a week (5), about once a week (4), 1–3 times a month (3), less than once a month (2), or never (1). To create composites of marital quality with higher scores indicating higher marital quality, we reversed marital conflict scores. The two measures of marital quality are significantly and negatively correlated with each other, r = -.41 (average of Time 1 and Time 2 correlations, p < .01).
A sense of personal control was measured with seven items from the Mastery Scale (Pearlin and Schooler 1978; e.g., "What happens to me in the future mostly depends on me."). Each item was followed with a four-level response format ranging from strongly agree (1) to strongly disagree (4). We calculated the scale score as the mean of the items, with higher scores indicating higher levels of personal control. The reliability of the scale (Cronbach ) was .84 at Time 1. The test–retest reliability (between Time 1 and Time 2) was .40 (p < .001).
Multiple Imputation (MI) of Missing Data
Regarding missing data (2–23 cases for men and 1–5 cases for women) for some of predictor variables, we performed MI of missing data by means of the NORM program developed by Schafer 1999. First we ran the expectation-maximization (EM) algorithm step, whereby unknown parameters of complete data structure are estimated on the basis of assumed values for the parameters. This process repeats until the estimates converge to maximum-likelihood estimates that implicitly average over the distribution of the missing values, with those estimated parameters serving as starting values for subsequent runs of data augmentation (DA). The DA is an iterative process that alternately fills in the missing data and makes inferences about the unknown parameters. The DA step first performs a random imputation of missing data on the basis of parameter estimates generated by the EM as starting values and then draws new parameters on the basis of the observed and imputed data. In the present study, we performed a single-imputation inference. Then we compared single imputation findings with MI inference results. For MI inferences, m = 5 (where m is the number of imputations) imputed data sets were generated through DA, and the resulting five sets of parameters and standard errors were subjected to an iteration procedure for MI inference (Rubin 1987) using the NORM program. The results regarding the estimates of the model coefficients based on the single imputation inference were approximately identical with the results based on MI inferences. For the regression models specified in the Results section, we report findings of the regression analyses that were based on the single-imputation inference data to evaluate overall model fits (i.e., F statistic and R2) and statistical comparisons of the regression models (i.e., 
R2).
Analytic Strategy
Because we believe that both one's work/retirement status and the transition to retirement are gendered processes, contributing to psychological well-being in distinctive ways for men and women, we analyzed all data separately by gender. We conducted analyses with ordinary least squares (OLS) regression. We first examined the relationship between retirement status/transitions and psychological well-being. In Model 1, the two psychological well-being measures, morale and depressive symptoms, were separately regressed on sets of variables reflecting respondents' retirement status and transitions. Age was included as a covariate and Time 1 psychological well-being score was included to allow us to examine the effect of initial levels of psychological well-being on later level of psychological well-being (or continuity of psychological well-being over time). The interactions between Time 1 psychological well-being and retirement transitions test whether the relationship between retirement status/transitions and subsequent psychological well-being depends on the level of Time 1 psychological well-being. To decrease the likelihood of multicolinearity between the interactive term and its components, we centered the Time 1 psychological well-being scores before taking their cross-products with retirement status/transitions by subtracting the mean value for each of these variables from individual scores (e.g., Jaccard, Turrisi, and Wan 1990). Next, in Model 2, we added spouses' employment status and the interaction between respondents' retirement status/transitions and spouses' employment status. The interactions test whether respondents' retirement status/transitions have a differential impact on changes in psychological well-being depending on spouses' employment status. Finally, in Model 3 we added both Time 1 measures and change scores of personal control, marital quality, subjective health, and income adequacy to test whether changes in contextual variables mediate any effects of retirement status/transitions on changes in psychological well-being after controlling for the baseline (Time 1) scores of these mechanisms. We created measures of change from Time 1 to Time 2 by subtracting Time 1 scores from Time 2 scores.
|
Results |
|---|
|
TOPAbstractTheoretical BackgroundA Life-Course, Ecological ModelMethodsResultsDiscussionReferences |
|---|
|
Retirement and Changes in Psychological Well-Being
Table 2 and Table 3 summarize findings of the hierarchical regression analyses predicting changes in psychological well-being by respondents' retirement status/transitions, their spouses' employment status, and other contextual variables for men and women, respectively. When we compared the estimated means of men's morale at Time 2 (after controlling for age and Time 1 morale), we found that newly retired men reported the highest morale score (estimated M = 5.92, standard error [SE] = .19) as opposed to those who were continuously retired (estimated M = 5.70, SE = .10) or those who had not yet retired from their primary-career jobs (estimated M = 5.55, SE = .19). An interaction between Time 1 morale and retirement status/transitions tested whether or not those with lower morale experienced the greatest change following retirement. We found that baseline morale indeed served as a moderator in the relationship between men's retirement status/transitions and changes in their morale (see Table 2 ). Fig. 1 depicts the predicted regression lines for the significant interactions found in men's psychological well-being. Specifically, among men who moved into retirement by Time 2 (the newly retired), those who were lower in morale at Time 1 experienced greater increases in their morale compared with those who were higher in morale.
|
|
|
Turning to findings for women, there was no strong evidence of significant main effects of retirement transitions or of interaction effects between retirement transitions and Time 1 psychological well-being on women's psychological well-being at Time 2 (see Table 3 ).
Linked Lives: Spouse Employment Status
As can be seen in Table 2 , there were no statistically significant differences in men's and women's well-being depending on their spouses' employment status. We further tested whether changes in spouses' employment status (rather than static status of spouses' employment) mattered for changes in psychological well-being of men and women (results not shown). The regression model included age, Time 1 psychological well-being, respondents' retirement transitions, interactions between Time 1 psychological well-being and respondents' retirement transitions, spouses' employment changes, and the interactions between respondents' retirement transitions and spouses' employment changes. We categorized spouses' employment status/change as follows: (a) spouses continuously out of the workforce at both Time 1 and Time 2, (b) spouses change of employment status between Time 1 and Time 2, and (c) spouses continuously employed at both Time 1 and Time 2. A significant interaction indicated that men who are continuously retired whose wives are also continuously out of the workforce tend to report lower levels of depressive symptoms at Time 2 compared with their counterparts whose wives are continuously employed or have changed employment status (ß = -.44, p < .01). The finding suggests a positive impact (in terms of reduced depressive symptoms) of having a nonemployed wife for those men who are continuously retired.
Resources as Mediating Mechanisms
For men's morale and depressive symptoms, adding contextual factors significantly increases the variance accounted for (R2 = .12, p < .05, for morale and R2 = .12, p < .05, for depressive symptoms; see Table 2 ). As expected, various resources seem to serve as mechanisms through which retirement circumstances shape men's psychological well-being. Specifically, increases (between waves) in perceptions of income adequacy, subjective health, and personal control were associated with concomitant increase in men's morale after controlling for Time 1 levels of income adequacy, subjective health, and personal control. As can be seen in Model 3 of Table 2 , for men's morale, the interaction between being newly retired and Time 1 morale is attenuated when the contextual variables are added to the model (standardized estimate ß decreases from -.20 to -.15). Regarding men's depressive symptoms, decreased health and decreased personal control were related to increased depressive symptoms after controlling for subjective health and personal control at Time 1. Note that the significant interaction between being continuously retired and Time 1 CES-D scores was no longer significant once these contextual variables were added to the model, suggesting that they do indeed serve as mediating mechanisms between retirement circumstances and men's depressive symptoms (standardized estimate ß decreases from .26 to .16).
Turning to women's psychological well-being, adding contextual variables to the model significantly increases the variance accounted for (R2 = .17, p < .05, for morale and R2 = .17, p < .05, for depressive symptoms; see Table 3 ). Increases in subjective health and personal control were associated with increased morale after we controlled for initial levels of subjective health and personal control. Declines in women's subjective health, marital quality, and personal control over a 2-year period were related to increased depressive symptoms, after we controlled for baseline levels of subjective health and personal control.
|
Discussion |
|---|
|
TOPAbstractTheoretical BackgroundA Life-Course, Ecological ModelMethodsResultsDiscussionReferences |
|---|
The Retirement Process in Context
Our findings indicate that retirement is not simply a state but a complex process, embedded in prior psychological resources as well as gendered experiences. Men tend to experience increases in their morale as they undergo actual transitions into retirement. This is consistent with previous findings of positive changes in life satisfaction during the 1st year postretirement among male retirees (Gall et al. 1997). It may well reflect the enjoyable "honeymoon" phase proposed by Atchley 1976b. He suggested that right after the event of retirement there often is a euphoric, busy, honeymoon phase during which retirees may feel more energetic, healthy, and satisfied as they pursue desired plans or experiment with new activities and roles. These findings seem to support the reduced role strain hypothesis: Retired men feel released from the pressures of their career jobs, thus retirement is beneficial for their psychological well-being. Other researchers also have shown that retirement may provide freedom from the time demands and daily structure of work life to pursue other interests and activities at a relaxed pace (e.g., Kelly and Westcott 1991; Nelson 1980). In that sense, retirement may represent a gain in the resource of time and a reduction in role strain. Consequently, when late-midlife men move into retirement, their morale or general satisfaction with life appears to increase.
Our longitudinal analysis suggests the importance of viewing retirement not only as a process (with differences in well-being, for example, between the newly retired, who have only recently made this transition, and those who have been retired for a longer period of time) but also as one occurring in particular contexts. Specifically, the impact of moving into retirement on subsequent changes in morale depends on initial levels of morale: Men retiring from their primary career jobs who have low levels of morale tend to experience greater improvements in morale following retirement, congruent with the role strain reduction theoretical perspective. Thus the retirement transition may well reduce the role strain and overload associated with the career jobs of those with low morale, thereby promoting their psychological well-being. Our findings also demonstrate that continuously retired men tend to report greater increases in depressive symptoms (compared with both those who are newly retired and those who are not yet retired), suggesting that being retired (as opposed to moving into retirement) may be a significant contributor to depression in later adulthood (e.g., Bosse et al. 1987; Wright 1990). The relationship between being retired and depressive symptoms, however, varies somewhat depending on prior psychological well-being. Consistent with the cumulation of disadvantage model, those retired men who are relatively higher in depressive symptoms are vulnerable to experiencing even more depressive symptoms as they move through the retirement years.
Gender is also a key factor in shaping the ways in which individuals experience the transition to retirement. Both one's own and one's spouse's work/retirement status and transitions are more strongly related to men's psychological well-being than women's. Although we found no evidence in support of earlier findings linking retirement with greater depression for women than for men (e.g., Atchley 1976a; Streib and Schneider 1971), these prior findings may reflect women's initially disadvantaged position. Specifically, our study shows that women come to retirement with higher initial levels of depressive symptoms but lower levels of morale, personal control, and perceived income adequacy (see Table 1 ).
Retirement as a Relational Transition
We focused on the interdependence of linked lives regarding the psychological well-being of married men and women around what are increasingly two retirement transitions, his and hers (e.g., Kim and Moen 2001a, Kim and Moen 2001b). Accordingly, we conceptualized retirement as a relational transition and considered both spouses' conjoint circumstances. On the basis of previous findings of significant spouse influences in accounting for variations in postretirement marital quality (e.g., Moen et al. 2001; Myers & Booth, 1996; Szinovacz 1996), we expected that the effects of respondents' retirement status or transitions on their psychological well-being would be contingent on their spouses' employment status. Our data, however, provide no strong evidence that spouses' employment status conditions the effect of respondents' own retirement transitions on subsequent psychological well-being. Although it is not statistically significant (p = .10), there is a tendency for newly retired men whose wives are still employed to report the fewest depressive symptoms at Time 2 (after controlling for the initial level of depressive symptoms), whereas men who are not yet retired from their primary career jobs but whose wives are no longer in the workforce report the greatest depressive symptoms. Such interesting patterns warrant further attention in future research. It may well be that men do better in terms of psychological well-being when they retire first and their wives subsequently follow them in to retirement.
Retirement and Resources
We find that resources—financial, personal, and social–relational—serve as mechanisms for explaining the link between retirement status and changes in psychological well-being. For both men and women, the relationship between retirement and psychological well-being seems to be partially mediated by changes in these resources, which, in turn, predict changes in psychological well-being. First, changes in income adequacy matter for men's morale but not for women's. Second, changes in subjective assessments of physical health are consistent predictors of changes in psychological well-being for both men and women. Third, sense of personal control is a key personal resource: Increases in one's sense of control from Time 1 to Time 2 predict increased morale and decreased depressive symptoms for men and decreased depressive symptoms for women. In fact, the personal-control measure is the strongest predictor of changes in men's morale and changes in depressive symptoms for both men and women, and the second strongest predictor of changes in women's morale (following subjective health). These findings reinforce prior evidence linking personal control and psychological well-being (e.g., Lachman and Burack 1993) by showing the salutary effects of a sense of personal control around the retirement transition. Finally, the hypothesis that marital quality matters more for women's well-being than for men's is supported by our data. For women, but not for men, declines in marital quality predict increases in depressive symptoms. Women appear to be more sensitive to changes in the quality of the marital relationship than men are. The stronger effects of marital quality on women's psychological well-being seem to be consistent with the typically greater involvement of women in marital and family roles over the life course, as well as the greater focus on interpersonal relationships for women than for men (Helgeson, 1984).
Strength and Limitations
This study has several distinct methodological and substantive strengths. First, we focused on gender differences in the impact of retirement status/transitions on subsequent adjustment (rather than statistically controlling for gender differences), which may contribute to redrawing what has been a "male model" of retirement (Calasanti 1996; Moen 1996). We found both similarities and differences between men and women in the relationships linking couples' retirement status/transitions and various resources to psychological well-being. Second, we considered retirement transitions as a couple-level phenomenon. This is particularly important, as most retiring individuals are married, and an increasing number of couples experience (or will experience) the retirement of both spouses. Finally, our use of longitudinal data permitted us to estimate the amount and direction of changes in psychological well-being, revealing important distinctions between being retired versus moving into retirement.
The most obvious limitation comes from constraints on generalization of the findings. Our sample included mostly White, relatively healthy men and women in late-midlife recruited from large organizations in upstate New York. The effect of retirement status and transitions on psychological well-being may differ among couples from different ethnic groups and for socially and economically disadvantaged couples (e.g., Szinovacz and DeViney 1999). In addition, the couples in our data belong to the cohorts born between 1922 and 1944. As with any age cohort, their employment and retirement experiences, as well as their perceptions of psychological well-being and marriage, are inevitably influenced by the historical context in which they live (Elder and Caspi 1992). Caution should be exercised in generalizing our findings to future cohorts. Further research on retirement transitions and psychological well-being could also consider more contextual variables, such as pre-retirement job characteristics, the timing of retirement, voluntariness of retirement, and retirement planning. Additional relational contexts should be probed, using a variety of couple-level or partner-related variables (e.g., caregiving activities, the division of housework, and spouse's health).
The conceptual and practical utility of our life-course ecological perspective is demonstrated by the evidence that psychological adaptation processes shift over time and vary by context for three groups of married men and women: those not yet retired, those retiring during the 2-year study period, and those already retired at Time 1. The findings reported here underscore the complexity of adaptation processes as late midlife men and women navigate and negotiate life transitions such as retirement, suggesting the importance of personal (such as personal control and health) and social–relational (such as marital quality and spouses' employment status) resources, even apart from a sense of financial adequacy. Future research should acknowledge and examine the dynamics and the heterogeneity of well-being around the retirement transition, including its embeddedness in gender, and other multilayered contexts.
|
Acknowledgments |
|---|
We thank John R. Nesselroade and Elaine Wethington for their helpful comments. We also gratefully acknowledge the assistance of the Cornell Employment and Family Careers Institute staff and especially Sarah Demo.
Received for publication January 19, 2000. Accepted for publication September 24, 2001.
|
References |
|---|
|
TOPAbstractTheoretical BackgroundA Life-Course, Ecological ModelMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  D. Pushkar, J. Chaikelson, M. Conway, J. Etezadi, C. Giannopoulus, K. Li, and C. Wrosch Testing Continuity and Activity Variables as Predictors of Positive and Negative Affect in Retirement J Gerontol B Psychol Sci Soc Sci, October 29, 2009; (2009) gbp079v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Kleinspehn-Ammerlahn, D. Kotter-Gruhn, and J. Smith Self-Perceptions of Aging: Do Subjective Age and Satisfaction With Aging Change During Old Age? J. Gerontol. B. Psychol. Sci. Soc. Sci., November 1, 2008; 63(6): P377 - P385. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
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]
|
|
|
|
|
|
Y. Hao Productive Activities and Psychological Well-Being Among Older Adults J. Gerontol. B. Psychol. Sci. Soc. Sci., March 1, 2008; 63(2): S64 - S72. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Mo Wang and T. E. Bodner Growth Mixture Modeling: Identifying and Predicting Unobserved Subpopulations With Longitudinal Data Organizational Research Methods, October 1, 2007; 10(4): 635 - 656. [Abstract] [PDF]
|
|
|
|
|
|
H. van Solinge and K. Henkens Involuntary Retirement: The Role of Restrictive Circumstances, Timing, and Social Embeddedness J. Gerontol. B. Psychol. Sci. Soc. Sci., September 1, 2007; 62(5): S295 - S303. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
