Self and Health: Factors That Encourage Self-Esteem and Functional Health (2024)

Search for other works by this author on:

Article history

Abstract

TODAY, as earlier distinctions between old age and middle age are becoming less clear and age-related stereotypes are becoming unstable, later life is emerging as a time for activity and social engagement rather than decline and dependency (Biggs, 2005; Lennartsson & Silverstein, 2001). This is, in part, a result of increased life expectancy and higher rates of social participation among older Americans, as well as increased labor force participation of women and the trend for many adults to become grandparents while still engaged in occupational careers (Kim & Moen, 2001). Similarly, as the percentage of the American population over the age of 65 continues to grow and the length of time that a person can expect to live increases, issues of health and well-being become increasingly more important (Schneider & Davidson, 2003). Therefore, it is not surprising that there has been greater recognition of the connections linking social psychological well-being and physical well-being. “Successful aging” has been described as the intersection of avoidance of disease and disability, maintenance of high cognitive and physical function, and engagement with life (Rowe & Kahn, 1998). We are interested in the dynamic relationship between self and health by focusing specifically on self-esteem, a sense of self-worth, and functional health, an assessment of the ease and absence of discomfort as one engages in physical activities.

The relationship between self and health, generally, and self-esteem and functional health, in particular, has been studied in at least three different ways. First, self-concept and physical health have been understood as independent indicators of a larger sense of well-being. For example, Rowe and Kahn (1998) suggest that self-efficacy and functional capacity independently influence productive activity, and Thoits and Hewitt (2001) found that self-esteem and subjective assessment of health both influence hours of volunteer work. Second, and more common, have been models which propose that health influences self processes such as self-esteem. Thus, Lee and Shehan (1989) found that subjective assessments of health influenced self-esteem for both older men and women, and Reitzes and Mutran (2002) report that poor functional health negatively impacted self-esteem among middle-aged working men and women. Third, self-concepts have also been modeled to influence health. Thoits (2003) and Schieman (2002) found that self-esteem influenced self-rated health; Wickrama, Conger, Lorenz, and Matthews (1995) reported that identity salience and role satisfaction influenced subjective assessments of health.

In this research, we will explore the complex relationship between self-concept and physical health. The data were collected in 1992 and 1994 from men and women, 58 to 64 years old, initially employed full-time and residing in the Raleigh-Durham-Chapel Hill metropolitan area. Our investigation is framed by symbolic interaction theory (Stryker & Burke, 2000) and its core assumptions that: (i) self develops through social interaction and a host of social factors that influence and support self-concepts, and (ii) self processes and concepts actively influence an individual's behaviors and actions. Thus, we are interested in investigating whether respondents' functional health in 1992 changes self-esteem from 1992 to 1994, and whether self-esteem in 1992 changes functional health between 1992 and 1994. The data set also allows us to probe three other issues. Symbolic interaction theory suggests that self includes not only global or overall self conceptions, such as self-esteem and self-efficacy, but also identities, situated self-meanings in social roles and statuses (Owens, 2003). We will probe whether worker, parent, and friend identities change self-esteem and functional health. Further, social background characteristics and social structure are expected to influence changes in both self-esteem and functional health (Keyes & Waterman, 2003). Finally, we are interested in whether gender and retirement impact self-concept and health. Too often in the past, studies of older workers and their retirement have focused only on the experiences of men, and used retrospective accounts of the transition into retirement (Szinovacz, 1998). Therefore, it will be interesting to consider the impact of gender and retirement on changes in self-esteem and functional health.

Research Expectations

Theoretical Background

Symbolic interaction theory (Stryker & Burke, 2000) provides a broad theoretical framework for exploring the relationship between self and health processes. The theory rests on two assumptions. First, symbolic interaction theory begins with Mead's (1934) insight that an individual is not born with a sense of self, but that self, as a social object, develops through social interaction and the experiences of a person throughout his or her lifetime. Thus, symbolic interactionists have focused on a broad set of interpersonal, small group, and larger cultural or societal factors, as well as on social background characteristics and physical factors that influence self-concept and especially self-esteem (Rosenberg, 1979). Further, symbolic interactionists suggest that the organization and structure of self reflects the social organization of society. James (1890) proposed that a person has as many selves as he or she gets recognition from others as a result of occupying multiple roles, and Park (1926) noted that as a result of multiple role occupancy, a person wears different “masks” in different social situations or in the presence of different sets of others. More recently, symbolic interaction theory has formally recognized that social structure influences and changes self and society. Stryker (1980) argues that social background factors such as age, class, and ethnicity create different opportunities for social interaction that can support and change self-concepts and identities.

Second, symbolic interaction theory recognizes that self is not only an outcome or product but also an important independent influence on individual actions and outcomes (Gecas & Burke, 1995). Cooley's (1902) “looking glass self” and Goffman's (1959) “presentation of self” describe the ongoing process of affirming one's self and the active process by which individuals initiate lines of interaction to gain self-confirming responses from others. Biggs (2005) invokes other theoretical perspectives to make the similar point that elders actively use identity management skills and strategies in negotiating intergenerational exchanges and interactions. Gecas (1986) proposed three underlying self processes that lead to self-initiated actions and behaviors: self-esteem, a preference to maintain or increase a favorable view of self; self-efficacy, the desire and motive to perceive oneself as a causal agent and to view oneself as competent and effective; and self-authenticity, the desire to be true or consistent with one's core meanings and values. Thus, individuals are motivated to maintain and enhance their sense of self-worth; therefore, the desire to preserve and protect one's self-esteem may influence and enhance a diverse set of outcomes, including physical health (Thoits, 2003).

Factors Changing Self-Esteem

Symbolic interaction theory suggests that self as a socially constructed object is created, maintained, and changed by a variety of social and physical factors. We expect that functional health encourages self-esteem. Good health provides a resource that facilitates behaviors that support self-esteem. Poor health challenges a sense of self-worth and also indirectly changes self-esteem through its negative effects on self-efficacy and autonomy. The literature suggests several different processes that may contribute to the relationship. First, health difficulties, as manifested by illness and health problems, may create a major source of stress that may lower self-esteem. Coleman, Antonucci, and Adelmann (1987) report that health interference, a summary score of 13 doctor-diagnosed health problems, had a negative effect on the self-esteem of middle-aged black women. Poor health and illness, with their pain, confinement, uncertainty, and high cost of treatment may also negatively affect a sense of self-worth. Further, poor functional health may limit or restrict opportunities for social support and activities that encourage self-esteem. Reitzes, Mutran, and Pope (1991) found that poor functional health both had a direct, negative impact on psychological well-being among retired men, and indirectly effected well-being by reducing participation in social activities. Lee and Shehan (1989) noted that poor subjective health was one of the strongest factors that influenced self-esteem for older men and women. They suggest that poor health may raise the specter of dependency, which would depress self-esteem. Thus, we expect that functional health will increase self-esteem over the course of the 2 years of the study.

One of the significant contributions of symbolic interaction theory has been its recognition that individuals do not just form global self-concepts, such as self-esteem, but also create self-meaning in specific social roles and statuses (Stryker & Burke, 2000). Identities as situated self-meanings are responsive to variations in individual behaviors, as well as the actual and perceived responses of others, whereas self-esteem tends to be more stable and resilient to change. Therefore, it is expected that it would take an accumulation of positive identities or positive identities over time to change self-esteem. Further, Cast and Burke (2002) suggest that self-esteem is an important outcome of the active process by which individuals strive to validate and enhance positive identities. Through reflected appraisals and social comparison processes, individuals seek to confirm their identities with a positive sense of self-esteem. They found that spousal identities that were confirmed by a partner had a positive effect on self-esteem and self-efficacy 3 years later.

Reitzes and Mutran (2002) report that worker and spouse identities had a positive effect on self-esteem for middle-aged, working men and women; and the grandparent and parent identities exerted positive effects on self-esteem for middle-aged grandparents (Reitzes & Mutran, 2004). In this study, we focus on three important adult identities—worker, parent, and friend—and the self-meanings of competence, confidence, and sociability for each identity (Mortimer, Finch, & Kumka, 1982). Thus, a positive worker identity reflects a person's image of himself or herself as a successful, relaxed, and sociable worker. Identities reflect cognitive self-meanings in situations, whereas self-esteem taps overall self-evaluation. Our next expectation is that identities will encourage positive changes in self-esteem.

Finally, our theory suggests that social background factors may change self-esteem. First, socioeconomic status reflects the financial resources, social and intellectual skills, and prestige that may directly enhance a person's self-evaluation. We expect that individuals with higher levels of income, education, and occupational status will serve to bolster and increase self-esteem. In addition, social roles provide important social references as well as behavioral guidance and opportunities for social support and identity confirmation, so we expect that being a parent and being married (independent of parent and spouse identities) will increase self-esteem (Thoits, 2003). Third, although age variation in this study is limited (all respondents were in the 58- to 64-year-old age range), age may be perceived as hindering self-esteem. In addition, Whites face less racial stereotyping and racial barriers to social and economic success than do blacks, and therefore may experience social and social psychological advantages that contribute to heightened self-esteem not available to blacks (Rosenberg, 1979).

Factors Changing Functional Health

The investigation of self-esteem emanates from the understanding that self is a social object and that self-esteem is influenced by a diverse set of factors. We now turn to the investigation of functional health and the recognition among symbolic interactionists that self processes provide a source of agency encouraging intrinsic meanings and motives for individual behaviors and actions. Gecas (1986) suggests that individuals desire to think well of themselves. Therefore, they are often motivated to establish positive self-esteem and to seek ways of confirming and maintaining their self-esteem. Thoits (2003) argues that people with high self-esteem are motivated to maintain their health. Good health serves to validate and confirm a sense of self-esteem the same way that an expensive new car or high fashion wardrobe announces to others, and affirms to oneself, that he or she is successful and deserving of high regard. Another interpretation is suggested by Schieman (2002), who proposes that self-esteem serves as a personal resource upon which a person can draw to avoid or cope with the deleterious effects of stressors on one's physical health. His data analysis revealed that self-esteem does indeed have a positive effect on self-rated health. We expect that self-esteem will encourage changes in functional health over the 2 years under investigation.

Identities also may stimulate improvements in functional health. Cast and Burke (2002) argue that individuals seek behaviors and support from others who confirm their identities. Positive worker, parent, and friend identity meanings may encourage functional health. In this sense, individuals with positive identities may interpret good health as a physical confirmation and behavioral support for one's identities. This may be especially true of the worker and parent identities, where good health can support the time, energy, and identity-related activities that affirm being a confident and competent worker and parent. Wickrama and colleagues (1995) proposed that success and satisfaction in highly salient roles will promote health, whereas dissatisfaction in a highly salient role will provoke illness, especially for roles important to one's identity. They found that marital satisfaction decreased poor health over time for both men and women, parental satisfaction decreased poor health for women but not men, and job satisfaction decreased poor health for men but not women. We propose that worker, parent, and friend identities will increase functional health over time.

Social background factors, in addition, are expected to encourage good functional health. In a fascinating study of the health consequences of social roles, Moen, Dempster-McClain, and Williams (1992) found that multiple role occupancy in 1956 had a positive effect on changes in women's health 20 years later. We will explore the impact of the parent and spouse roles on health changes. Turning to other social background factors, Schieman (2002) found that neither education nor income influenced self-rated health, but that age had a negative health effect, and being a parent had a positive effect on health. As part of their study of the volunteer role and well-being, Thoits and Hewitt (2001) considered factors that changed subjective assessments of health over a 3-year period and found that: age decreased health; education and being employed increased health; and race, family income, and having children at home did not influence changes in health. In this study, we expect that being married and being a parent will increase functional health and that the socioeconomic resources associated with education, income, occupation, as well as being White and younger will encourage functional health.

Gender and Retirement

One of the hallmarks of symbolic interaction theory has been its interest in roles and how roles structure self-concept and influence behaviors (Stryker & Burke, 2000). Roles are sets of shared experiences and meanings that identify the common characteristics of occupants of a social position or status. In this way, roles reflect the social opportunities as well as constraints that confront individuals as a member of a group, institution, or society. We will investigate two roles, gender and retirement, that past research suggests may change self-esteem and functional health.

The impact of gender on self-esteem and health is interesting. It is often assumed that, given men's advantages over women in terms of educational attainment, income, and occupational status, there will be comparable gender differences in self-esteem. However, the majority of studies do not find differences in self-esteem scores of men and women (Mackie, 1983; Schieman, 2002). Women may compare themselves to other women and not necessarily to men, or women may use different standards to assess themselves. The gender socialization literature suggests that women may use interpersonal ties and identity meanings to affirm self-esteem, whereas men use external social status to affirm their self-esteem (Wood, Rhodes, & Whelan, 1989). More than gender socialization may be at work. Neff and Harter (2002) found that power inequality in role relationships enables men to emphasize autonomy and independence, whereas women's subordinate positions lead to concerns about maintaining role relationships. We will explore whether there are different factors that change self-esteem for men and women. Differences in work experiences may change the functional health of men and women. As a consequence of interrupted work histories, lower seniority, or restricted occupational mobility, working women have fewer financial resources than do working men. Because working women may be less able to retire early than men, there may be more women with health problems remaining in the labor force (Perkins, 1992). Further, as a result of differential gender socialization, women's health may be influenced by family-related roles and identities, whereas men's health may be influenced by work-related roles and identities (Wickrama et al., 1995).

Retirement, as a major life course transition, may be expected to change both self-esteem and health. Reitzes, Mutran, and Fernandez (1996) found that self-esteem did not change as men and women moved from preretirement to retirement, just as it did not change for men and women who continued to work full-time during the same 2-year period. Unlike widowhood, retirement for most workers is an anticipated transition, and less likely to have a dramatic negative impact on self-esteem (Ekerdt & DeViney, 1993). With the onset of retirement, the loss of the worker role may have negative consequences for health. However, there is little evidence for this proposition (Kim & Moen, 2001). Instead, retirement may signal reduced stress and the absence of job-related anxieties (Midanik, Soghikian, Ransom, & Tekawa, 1995), and enable retirees to invest their time and energy in activities that support valued roles and identities (Atchley, 1999). We will investigate whether gender and retirement change self-esteem and functional health, as well as whether there are differences by gender or retirement status in the impact of other factors on changes in self-esteem and functional health.

Methods

Data

The two waves of data were collected in 1992 and 1994 as part of the ongoing study of the transition into retirement (CHATS, Carolina Health and Transitions Study). Beginning with the earlier data set, the sampling procedures were designed to identify approximately 400 men and 400 women aged 58 to 64 years who were working at least 35 hr a week and residing in the Raleigh-Durham-Chapel Hill, North Carolina metropolitan area. To gather a representative sample of older, working men and women, driver history files were obtained from the North Carolina Department of Motor Vehicles. The file is estimated to include more than 80% of the entire population in the desired age group. From the list, which records age, home address, and gender, but not the telephone number or work status of applicants, names were randomly selected in proportion to the size of the three counties included in the study area. Following Dillman (1978), introductory letters were sent out, as well as screening postcards (three weeks apart) and follow-up telephone calls (up to 9 tries) to verify telephone numbers and to identify full-time working prospective participants living in the area. Of the people identified by the screening procedures as eligible, 62% (826) consented and in 1992 participated in two 20-minute telephone interviews (60% of the men, 64% of the women). The response rate may have lowered due to the length of the interviews, but they are similar to the response rates for other telephone surveys (Thoits, 2003).

Each of the 826 workers was tracked at six-month intervals over the next 2 years. The attrition rate of 8.2% reflects the loss of 68 cases (14 died and 54 dropped out). By July 1994, 758 of the respondents had either retired or remained working full-time. An additional 21 people provided incomplete information and were dropped from this investigation. Of the 737 respondents who provided useable follow-up interviews, 438 were still employed full-time and 299 had retired. Overall, the sample contains a diverse set of workers and retirees with a variety of social background characteristics, including 73% married, 83% White, and 52% female. The diverse sample enables the analysis to include exploration of the relationship between self-esteem and functional health.

Variables

Self-esteem 1 and Self-esteem 2 are derived from Rosenberg's (1965) self-esteem scale, which has proven to be a durable measure of a person's assessment of self. Our theory suggests that individuals desire to think well of themselves, and that good health may be understood as a sign of well-being. Self-esteem measures self-worth, the evaluative dimensions of self-concept, and therefore is well-suited for this investigation. The items used to construct the scales come from the 1992 survey for Self-esteem 1 (α =.88) and the 1994 survey for Self-esteem 2 (α =.89). They include ten items such as “I feel that I'm a person of worth, at least on an equal basis with others,” “I am able to do things as well as most other people,” and “I wish I could have more respect for myself” (recoded). Responses ranged from “strongly agree” (4) to “strongly disagree” (1).

Health 1 (α =.79) and Health 2 (α =.82) are our functional health measures taken from the 1992 and 1994 surveys, respectively. The scales are based on 7 functional activities items that asked if respondents had any difficulties with: (1) walking about three blocks; (2) using stairs or inclines; (3) standing for long periods of time; (4) sitting for long periods of time; (5) stooping, kneeling, crouching, or bending; (6) lifting or carrying weights up to 10 pounds; and (7) reaching above your head. Responses included “Never” (3), “Sometimes” (2), or “Often” (1).

The identity variables—worker identity, parent identity, and friend identity—are adapted from the study by Mortimer and colleagues (1982) to measure meanings of competence, confidence, and sociability in each identity. After the leading phrase “As a worker, I am …,” “As a parent I am …,” or “As a friend, I am …,” adjective pairs were organized in a semantic differential five-point format (Osgood, Succi, & Tannenbaum, 1957). The adjective pairs include: active-inactive; successful-unsuccessful; competent-not competent; relaxed-tense; happy-sad; confident-not confident; warm-cold; open-closed; interested in others-interested in self; and sociable-solitary. For each adjective pair, responses nearer to the positive adjective meaning (e.g., happy, confident, warm) receive higher scores (4 or 5), and responses closer to the negative or less positive adjective meaning (sad, not confident, or cold) receive lower scores (1, or 2, with middle responses of 3). Scores for each adjective pair were summed to create a single measure that captures positive identity meanings. The alpha reliability scores for worker, parent, and friend identity meanings are.77,.85, and.89.

Table 1 displays correlations between self-esteem and identity meanings. Looking first at the identities, we find correlations ranging from.58 to.67. These moderately strong correlations are consistent with our interpretation that self-meanings in different roles are related but independent constructs. As expected, the correlations between Self-esteem 1 and the identities are in the.27 to.32 range. These correlations are consistent with our expectation that identities, as sets of cognitive self-meanings, are different from self-esteem, which taps general or global, evaluative self-assessments. Finally, the correlation between the two self-esteem measures suggests that self evaluations are basically stable, but there were some changes that occurred between 1992 and 1994.

Also of interest are the effects of gender and retirement in changing self-esteem and functional health. Female will serve as the gender dummy variable. Respondents are considered “retired” if in 1994 they reported that they were no longer working or working less than 35 hours a week in their primary job, and if they agreed that they were “retired.” If employment at their primary job dropped below 35 hours a week, the retirement screening protocol was initiated. The retirement literature suggests that a combination of self-definitions that asks respondents if they consider themselves retired and departed from full-time employment (working less than 35 hours a week) is the most appropriate way to operationally define retirement (Hayward, Hardy, & Grady, 1989). Thus, retired, is a dummy variable (1 = retired, 0 = working) taken from the 1994 survey.

Social background characteristics include six indicators. Parent and married are dummy variables coded to indicate role occupancy. Income comes from a question that asks for the total 1991 household income with 10 response categories ranging from “$7,500 or less” (1), “$35,001 to $50,000” (5), and “$200,001 and over” (10). Education is based on the highest grade completed in school, and coded in years. Occupation refers to preretirement occupation, and is measured by a 100-point occupational prestige scale using 1980 U.S. census classifications and 1989 National Opinion Research Center (NORC) prestige scores. Scores range from 86 for physicians and 75 for lawyers to 09 for shoe shiners and 19 for news vendors (N.O.R.C., 1991). Finally, age is measured in years, and race is defined by a dummy variable (White), with Whites coded 1 and non-Whites coded 0. The non-White category is almost entirely composed of African-Americans (less than 1% of the non-Whites self-identified as Asian American, Hispanic, American Indian, or another racial group).

Results

We begin with a descriptive overview. Table 2 reveals that our respondents are in good mental and physical health. They exhibit high self-esteem at both times, which is consistent with responses from other adults (Rosenberg, 1979). Most do not have difficulties with their functional health. Indeed, 28% stated that they never have any difficulties with the seven health activities in 1992, and 29% reported no difficulties in 1994. They also exhibit and maintain positive worker, parent, and friend identities. The overwhelming majority (93%) are parents, and 73% are married. Further, slightly more than half are female (52%), and 41% retired by 1994. Their average income is a little above the $35,001 to $50,000 range, with 43% in managerial or professional occupations, and 12% in blue collar craft or laborer positions. On average they have 2 years of education beyond high school. Finally, most of our respondents were White (83%) and were between 60 and 66 years old in 1992.

Table 3 summarizes a structural equation model (Joreskog & Sorbom, 2004) that simultaneously identifies factors that contribute to changing self-esteem and functional health over time. The model is saturated, the fit is perfect with all measures estimated, and the model allows for correlated error between self-esteem and functional health in 1994. Beginning with the analysis of self-esteem, we found that, as expected, although there is considerable stability in self-esteem over the 2 years, functional health encourages positive changes in self-esteem. In addition, worker identity meanings enhanced self-esteem. Our third set of expectations focuses on gender and retirement. We found neither exerted a statistically significant effect on changing self-esteem. Thus, men were not more likely than women to experience positive changes in their self-esteem, despite their favorable social background characteristics. Also, the transition from full-time work to retirement does not negatively change self-esteem. Among the social background factors, being married and age lower self-esteem, whereas higher levels of education and higher income encourage positive changes in self-esteem over the 2 years.

Table 3 suggests that, as expected, self-esteem is a factor that contributes to increases in functional health. However, neither positive worker, parent, or friend identity meanings nor occupying the parent or married roles effect changes in functional health. Self-esteem as a motive for changes in health appears stronger than do motives to enhance identities and role through good health. Further, we find that women's health decreased relative to men's over the 2 years, but retirement status did not effect changes in functional health. Finally, none of our social background factors—education, income, occupation, race, or age—directly influenced change in functional health.

Beyond the direct effects of gender and retirement status, we also were interested in whether there are systematic differences in the way factors may change self-esteem and functional health for men and women and for workers and retirees. Therefore, in two subgroup analyses (Joreskog & Sorbom, 2004) we first compared the general model presented in Table 3 with models generated specifically for men and women, and then compared the general model with models for workers and retirees. The results suggest that separate models for men and women did not fit the data better than did the general model (Minimum Fit Function χ² = 22.69, p =.75; Normed Fix Index [NFI] =.99 and Critical N [CN] = 1,564) and that separate models for workers and retirees also did not fit the data better than did the general model (Minimum Fit Function χ²e = 34.84, p =.21; NFI =.99, and CN = 1,047). The finding is consistent with an earlier ordinary least squares analysis that compared a general equation with separate equations for men and women and for workers and retirees, and found that there were not systematic differences in the way factors influenced change in self-esteem and functional health (Chow, 1960). We also compared individual coefficients (Kleinbaum, Kupper, & Muller, 1988) and found there were not statistically significant differences across groups. For example, worker identity encouraged self-esteem for both men (β =.13, p <.05) and women (β =.08, p <.05), and functional health declined more for women than men among both workers (β = −.10, p <.05) and retirees (β = −.09, p <.05).

Discussion

This study focuses on the relationship between self and health processes. Symbolic interaction theory suggests that self, as a social construct, is influenced by multiple factors, and that maintaining a sense of self-worth is an intrinsic source that can motive individual behaviors and actions (Stryker & Burke, 2000). Beginning with the former, our theory recognizes that individuals are born with a sense of self, but that self conceptions, such as self-esteem, are influenced by a number of factors including functional health. We propose that functional health enables individuals to engage in the behaviors and participate in the activities that foster self-regard as well as to avoid the pain and discomfort of poor health, which could inhibit feelings of self-worth. Further, just as clothing or high status objects are part of the process of self-presentation (Goffman, 1959), so is good health. Good health contributes to the process of affirming and announcing a sense of self-worth. Our first research question is whether functional health encourages change in self-esteem over 2 years. We found that the answer is “yes.” Functional health in 1992 exerted a positive and statistically significant effect on change in self-esteem (β =.16, p <.05), so, in addition to the desirability of maintaining good health in and of itself, we find support for the proposition that good physical health also may contribute to enhancing one's psychological well-being and mental health.

Our theory, as well as psychological identity theories (Biggs, 2005), acknowledge the importance of agency in understanding human behavior. One of the strengths of symbolic interaction theory is its recognition of the ongoing process of affirming one's self and the active manner in which individuals initiate lines of action to maintain and enhance a positive sense of self (Cast & Burke, 2002). Self-esteem, as Gecas (1986) argued, provides a motive, a desire to confirm a positive sense of self that encourages consistent and supportive outcomes. Thoits (2003) suggested that the desire to preserve and enhance one's self-esteem may enhance one's physical health. We found that self-esteem exerted a positive and statistically significant effect on changes in functional health (β =.22, p <. 05). The finding is important, in part, because there has been more research on impact of health and changes in health on self-esteem and well-being than on the impact of processes on physical health.

Individuals do not just form global self concepts, such as self-esteem, but also form self-meanings in specific social roles and statuses (Stryker & Burke, 2000). We proposed that positive worker, parent, and friend identities would encourage positive changes in self-esteem and functional health. We found only limited support for our expectation. Worker identity meanings exerted a positive and statistically significant effect on changes in self-esteem (β =.11, p <.05). Older adults who viewed themselves as competent, confident, and sociable workers increased their self-esteem over the 2 years of the study. Parent and friend identities may be more closely tied to specific, role-related self-assessments such as role satisfaction or role-related well-being, whereas worker identities may be linked to more general or global self-assessments. Alternatively, the worker role may be more stressful than are most adult roles, and therefore a positive worker identity would have an especially positive impact on self-esteem (Wheaton, 1990).

We also explored the impact of gender roles and retirement status on self processes and health. First, we considered the direct effects of gender on changes in self-esteem and functional health. We found that neither influenced changes in self-esteem. The findings are nevertheless interesting. On one hand, despite substantial differences in work histories and typically fewer financial resources (Perkins, 1992), older women in this study, all of whom were working in 1992, were not less likely than men to change their self-esteem. Women and men may focus on different roles and social supports to maintain and enhance their self-concepts (Neff & Hartner, 2002). On the other hand, women were more likely than men to experience declines in their functional health. Here, disparities in work histories and financial rewards may come into play. Women may be less able than men to afford the medical treatment for nagging pain and restricted physical mobility that is reflected in measures of functional health. Second, retirement, as expected, did not effect changes in self-esteem or functional health. Unlike adults with an unexpected role loss (such as widowhood), workers can anticipate and prepare for retirement so as to minimize some of its negative consequences. Further, many older adults may perceive their work roles as stressful. Retirement may provide new opportunities to engage in desirable and valued nonwork activities which balance some of the intrinsic and extrinsic rewards of full-time employment (Kim & Moen, 2001).

Finally, we were interested in whether the set of health, self, and social background processes influenced changes in self-esteem and functional health differently by gender and/or retirement status. Our subgroup analyses (Joreskog & Sorbom, 2004) and an earlier comparison of regression analyses (Chow, 1960) revealed that factors influenced changes in self-esteem and functional health in a similar manner for men and women and for workers and retirees. We found, for example, that worker identity, education, and income enhanced self-esteem for both men and women. We expected that worker identity and external social status would have stronger effects on self-esteem for men, and that the parent and friend identities would have stronger effects on self-esteem for women. Gender differences in factors that influence self-esteem and health may be more pronounced earlier in the life course, when issues of balancing marital and parental roles with occupational roles have traditionally created different challenges and opportunities for men and women (Wickrama et al., 1995). The finding that similar sets of factors influence changes in self-esteem and functional health was less surprising and consistent with our emerging understanding that retirement is a gradual, continuous process and not an abrupt crisis (Atchley, 1999).

The research has at least three limitations. First, a larger sample would allow us to explore a latent factor measurement model (Joreskog & Sorbom, 2004) where we could estimate coefficients for each of the ten indicators of self-esteem in 1992 and 1994, seven indicators of functional health for each period, as well as ten indicators for each of the identity items. Second, it would be interesting to compare our result with those from a sample that included women of the same age (58 to 64 years old) who were not working full-time in 1992. There may be more pronounced gender differences in factors enhancing self-esteem and functional health when working men are compared to women who are not employed full-time outside the home. Parent and friend identities may exert strong effects on self and health among nonworking women. Finally, a larger and more racially diverse sample would enable us to investigate whether being African American exerts direct effects on self and health and whether the same set of factors influence changes in self-esteem and functional health for African Americans and Whites.

In conclusion, this study is exploratory in nature. It is important to recognize that not only does health influence self-esteem, but self processes also may influence health outcomes. If so, the next step is to systematically address the potential therapeutic effects of self processes. Can programs and interventions be developed that link self processes with health outcomes? For example, efforts to encourage exercise or smoking cessation may be effectively tied to being a parent or grandparent. A good parent or grandparent needs to be active and functionally healthy. Clearly, any such efforts must be sensitive to life-course changes and the different experiences of adult men and women. Further, good health may be more strongly tied to a positive parent identity when children are younger, and more important to a positive identity as a mother than as a father. Good health is not just an objectively desirable outcome, but becomes a way of confirming and enhancing self-esteem and identities.

Decision Editor: Charles F. Longino, Jr., PhD

References

The Gerontological Society of America

Issue Section:

Journal of Gerontology: Social Sciences

Email alerts

Citing articles via

More from Oxford Academic