Self-esteem and the intersection of age, class, and gender

Self-esteem and the intersection of age, class, and gender

Journal of Aging Studies 18 (2004) 75 – 90 Self-esteem and the intersection of age, class, and gender Julie Ann McMullin a,*, John Cairney b b a Dep...

216KB Sizes 0 Downloads 24 Views

Journal of Aging Studies 18 (2004) 75 – 90

Self-esteem and the intersection of age, class, and gender Julie Ann McMullin a,*, John Cairney b b

a Department of Sociology, University of Western Ontario, London, Ontario, Canada N6A 5C2 Health Systems Research and Consulting Unit, Centre for Addiction and Mental Health and the Department of Psychiatry, University of Toronto, 33 Russell Street, Toronto, Ontario, Canada M5S 2S1

Abstract In this paper, we explore the relationships among self-esteem, age, class, and gender. We ask whether relationships between self-esteem and gender are similar across age and class groups. Contrary to past research, which suggests that age has little influence on self-esteem or that self-esteem increases with age, we find that levels of self-esteem are lower in older age groups for both men and women. Furthermore, in all age groups, women have lower levels of self-esteem than do men. Social class does not influence levels of self-esteem for young men or women but does so for those in middle age and older age groups. From middle age on, social classes diverge in their levels of self-esteem such that men and women from lower social classes experience the lowest levels of selfesteem. Unlike past research that links age differences in self-esteem to either role or development theories, we consider these findings within a theoretical framework that recognizes the structural power relations embedded in class, age, and gender, which in turn influence gendered identities. D 2003 Published by Elsevier Inc. Keywords: Age; Gender; Social class; Self-esteem; Feminist methods

1. Introduction During the early stages of writing this paper, one of the authors (McMullin) attended a fundraiser for Brescia University College, a small women’s school in London, Ontario, Canada. The appeal in attending such an event was tied to the cause—raising money so that disadvantaged women would be given scholarships to attend this university. Less significant (it seemed at first) was that Dini Petty, a famous Canadian journalist, was the guest speaker. By all accounts, Petty has had a successful career. She began her work as a traffic reporter who rode in a helicopter and reported about the trouble spots on Toronto’s busy streets. Unlike anyone else at the * Corresponding author. E-mail addresses: [email protected] (J.A. McMullin), [email protected] (J. Cairney). 0890-4065/$ - see front matter D 2003 Published by Elsevier Inc. doi:10.1016/j.jaging.2003.09.006

76

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

time, Petty decided that she would prefer to both fly the helicopter and report on traffic. Soon she became only the 138th woman in the world to have a helicopter license. Petty held subsequent jobs as a reporter, an anchor for the 6 o’clock news, the host of a Canadian talk show, and most recently author of a very good children’s story called The Queen, the Bear, and the Bumblebee. When Petty took the stage at the gala, she presented herself as a confident, self-assured, articulate, and funny woman. Yet, the focus of her talk was on her lifelong struggle to gain self-esteem. ‘‘If there were medals awarded for lack of self-esteem,’’ said Petty, ‘‘I would have received gold.’’ In her talk, Petty identified two key factors that contributed to her low self-esteem: (1) the fact that girls learn at a very early age that they need to be nice to be liked and (2) the fact that girls are faced with a lot of pressure to be beautiful. Of course, in doing so, Petty concurred with decades of feminist literature that has found the same thing. However, Petty also talked about her epiphany: about the moment at which she looked in the mirror and saw a beautiful person ‘‘both inside and out.’’ She talked about the things she has done over the last few years to gain self-confidence and ultimately self-esteem. Not insignificantly, Petty told the audience that she would soon turn 58, a point that drew applause from some of the listeners, no doubt because she does not ‘‘look her age.’’ As Petty noted, self-esteem for young women is linked to cultural notions of beauty and femininity (Abell & Richards, 1996). Young women who perceive themselves as ugly, fat, too short, too tall, and so on experience lower levels of self-esteem than do those who have more positive assessments of their body (Abell & Richards, 1996). Cultural ideals of female beauty, at least in North America, are also tied to youthful appearances. Women are not considered beautiful if they are wrinkled, gray-haired, or fluffy (Abu-Laban & McDaniel, 2001). Hence, one would expect that, among women, self-esteem would diminish with increasing age. Yet, this hypothesis stands in contrast to Petty’s experience of gaining selfesteem when she was in her 50s. The fact that Petty’s self-esteem did not decline with advancing age but rather increased markedly may be tied, at least in part, to her privileged structural location as a White, economically well-off, professional, woman. Most feminist research on self-esteem focuses on young women. Yet, in light of what we know about the negative implications of ageist attitudes on perceptions of body and self (Calasanti & Slevin, 2001; Hurde Clarke, 2001), the neglect of feminist research on the relationship between aging and self-esteem is unfortunate. Hence, this paper is a modest attempt to explore the relationship among age, gender, social class, and self-esteem using a feminist perspective. In the end, more questions about this relationship are left unanswered than are answered, thus providing fruitful avenues for future research in feminist gerontology.

2. Literature review Self-esteem is a sociopsychological construct that assesses an individual’s attitudes and perceptions of self-worth. Thus, self-esteem is ‘‘an understanding of one’s quality as an object—that is, how good or bad, valuable or worthless, positive or negative, or superior or inferior one is’’ (Thoits, 1999, p. 342). Individual assessments of self-esteem are formed through two interrelated processes. First, individuals compare their social identities, opinions, and abilities with others. To the extent that individuals feel that they are inferior to those with whom they interact, their self-esteem will be negatively affected. Second, individuals assess themselves through their interaction with others. People learn to see themselves as others believe them to be. If significant others do not think highly of an individual, that individual will

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

77

come to think poorly of himself or herself. This is referred to as the ‘‘reflected appraisal’’ of one’s selfworth (Rosenberg & Perlin, 1978). Self-esteem is strongly and negatively correlated with distress and depression (Rosenberg, 1985; Rosenberg, Schooler, & Schoenbach, 1989), and individuals who have high perceptions of self-worth and self-esteem are thought to cope better with stress (Perlin, Lieberman, Menaghan, & Mullan, 1981). Hence, to understand distress and depression, one needs to understand the factors and processes that contribute to low self-esteem. Although there are many correlates of self-esteem, this paper focuses on gender, age, and social class and whether these factors intersect in their influence on self-esteem. The strong and consistent finding regarding gender and self-esteem is that compared with men, women have lower levels of self-esteem in adulthood (Josephs, Markus, & Tafarodi, 1992). Although boys and girls start with very similar levels of self-esteem in early adolescence (between ages 11 and 13), they gradually diverge throughout the teenage years and adulthood with boys gaining a sense of positive self-worth and girls losing that sense (Rosenfield, 1999). Lower levels of self-esteem help to explain the fact that girls and women are more likely to experience higher levels of distress and depression than boys and men (Avison & McAlpine, 1992; Rosenberg, 1985; Rosenfield, 1989). Although there are several explanations for this relationship in the mental health literature, Rosenfield’s (1999) is compelling. She argues that men and women have different social structural experiences that begin in early childhood and are reflected in the relative power that men and women have in society. Relative power, in turn, influences self-appraisals. As a result, Rosenfield suggests that ‘‘Given the power, the responsibility in the public domain, receipt of support, and value placed on masculine pursuits, males generally tend toward high self-esteem’’ (Rosenfield, 1999, p. 220). Although Rosenfield uses this argument to explain gender differences, it can also be extended to explain differences in self-esteem among men. Working class men, for instance, have considerably less power than do middle and upper class men, and with increasing age, the likelihood of men engaging in valued masculine pursuits declines considerably (e.g., working for pay and playing football). Other compelling explanations for gender differences in self-esteem focus on issues related to reflected appraisals and social comparisons. Girls and women, more than boys and men, are socially judged based on what they look like. If they diverge from socially constructed cultural ideals of beauty, others may think poorly of them, and through reflected appraisals, their self-esteem may suffer. Furthermore, girls and women take part in processes of comparison whereby they compare their beauty with that of others. If a woman feels less beautiful than the women with whom she interacts, her selfesteem will be negatively affected (Abell & Richards, 1996; Furman, 1997). There is little doubt that men also engage in comparative beauty exercises and value their corporeal appearances (Oberg & Tornstam, 1999). However, the fact that women rate their appearance with higher importance than men do suggests that physical appearance may be more salient to identities and self-esteem among women than among men. In the literature on the relationship between self-esteem and social class, social class is typically measured using proxies such as occupational status, income, and education. In general, adults who have lower occupational status, income, and education tend to have lower levels of social esteem than their respective counterparts. For instance, individuals who are employed in good jobs that are characterized by high autonomy, high prestige, and low routinization have higher levels of self-esteem than do unemployed individuals or individuals who work in bad jobs (Gecas & Seff, 1990; Pugliesi, 1995; Rosenberg & Perlin, 1978). Furthermore, individuals with more income and education have higher levels of self-esteem than do persons with lower incomes and less educational attainment (Mirowsky &

78

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

Ross, 1996). Linking gender and social class, some research demonstrates that women’s lower levels of self-esteem may be partly explained by the fact that relative to men, women tend to be employed in poorly paid jobs that provide less autonomy and fewer rewards (Pugliesi, 1995). Explanations for the relationship between self-esteem and social class also often focus on reflected appraisals and social comparisons (Rosenberg, 1981; Rosenberg & Perlin, 1978). The reflected appraisals argument suggests that members of the working class have poor self-esteem because they are judged negatively by those with whom they interact based on their low status jobs, incomes, and education. At work, for example, managers may perceive themselves as superior to the workers they employ and treat them accordingly. Reflected appraisals are linked to social comparisons. As the argument goes, members of the working class are more likely to view themselves as inferior to members of the middle or upper classes than vice versa. In short, the combination of reflected appraisals and social comparisons negatively influences self-esteem among the working class. In support of these arguments, Rosenberg and Perlin (1978) argued that working class children might not experience the negative perceptions and attitudes about their class as acutely as their parents. As a result, they hypothesized that the effects of social class on self-esteem would be larger in adulthood than in childhood. Their research supported this hypothesis, noting that the strongest relationship between social class and self-esteem is among adults (Rosenberg & Perlin, 1978). Although this work is now dated, the limited research that has since been conducted on the relationship between self-esteem and social class confirms these findings (see Turner & Roszell, 1994 for a review). Perhaps more importantly, the theoretical underpinnings of the argument are sound. The social class homogeneity of many school settings shelters children from social comparisons and reflected appraisals that negatively affect self-esteem. Unfortunately, the samples Rosenberg and Perlin and others used to test their hypotheses were of persons under the age of 65. Hence, it is unclear how class-based reflected appraisals and social comparison arguments play out in later life. Indeed, the exclusion of persons aged 65 and older is a characteristic of all of the work described above. Only a handful of cross-sectional studies have explored the relationship between age and selfesteem into old age, and these studies have produced mixed results. Some show that self-esteem remains stable or increases as individuals age, others suggest that it decreases, and still others demonstrate that there is a curvilinear relationship between age and self-esteem (for an overview, see Giarrusso, Mabry, & Bengtson, 2001). Two explanations for the relationship between self-esteem and aging have been put forth. The maturation perspective suggests that as individuals age, they become more accepting of who they are. This perspective draws on Erikson’s developmental stages and argues that the process of social comparison is not as salient in later life because at this stage individuals develop ‘‘ego integrity’’ and a general acceptance of their life’s accomplishments (Dietz, 1996). Hence, developmental approaches to the study of aging and self-esteem would predict stable or increasing levels of self-esteem in later life. Role perspectives have also been used to explain the relationship between aging and self-esteem. According to role perspectives, the loss of social roles that is associated with old age will result in lower levels of self-esteem. Thus, the role perspective argues that as people retire and disengage from active parenting, their self-esteem will suffer (Dietz, 1996). A key problem with both of these perspectives is that they do not consider how structured power relations change throughout the life course or how they influence on development processes. Further, although longitudinal data are required to accurately assess whether either of these perspectives adequately explains the relationship between age and self-esteem, only cross-sectional studies have examined this relationship to date.

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

79

2.1. Feminist gerontology and self-esteem Although studies on self-esteem tend to ‘‘control’’ for gender, class, and age or to consider the links between two of these factors, we are unaware of any research that explores the intersections among all three as they relate to self-esteem. Yet, feminist research on intersectionality (see Hill Collins, 2000) and on aging and old age tells us that class, age, gender, race, ethnicity, sexuality, and so on intersect in creating systems of advantage for some and disadvantage for others (Arber & Ginn, 1995; Calasanti, 1996; Calasanti & Slevin, 2001; Estes, 1999; McMullin, 2000). A significant advantage of this paper is that it focuses on three sets of intersecting power relations—gender, age, and class. Unfortunately, data limitations force us to lump people of various racial and ethnic backgrounds and sexual orientations together as if their experiences were the same. A starting point from which much feminist gerontology departs is the theoretical premise that power relations fundamentally structure systems of inequality (Browne, 1998; Stoller & Gibson, 1997). Hence, the intersecting structures of gender, class, and age influence various labor market, educational, reproductive labor, state, and health outcomes (see McMullin, 2004). Regarding self-esteem, the preceding review of the literature identified two explanations for differences in self-esteem. First, self-esteem is influenced by the relative control that individuals have over their lives, and gender, class, and age structure this control in complex ways. It is through these complexities that differences in selfesteem emerge. We might expect for instance that positions of authority within families might buffer the negative influence of class position on levels of self-esteem for working class men. We might also expect that the control that professional women have in paid work might buffer the negative effects on selfesteem that comes with their lack of control in determining who is responsible for the undervalued work of caring for children or older adults in families. For older adults, changes in self-esteem may be attributed to changes in power that come with role loss (i.e., retirement) rather than the loss of role itself. However, here, class and gender structure these effects. Working class men may feel a sense of empowerment with positive benefits to self-esteem that come with no longer having to work for anyone. This may stand in contrast to the experience of middle and upper class men who will no longer have anyone under their control when they retire. The effect of class-based power for women may be similar to that of men in later life; however, importantly, family caring and the relative control of men and women in the household never end. Thus, the relationship among class, age, and self-esteem cannot be separated from gender. The second set of explanations for differences in self-esteem focuses on the processes of social comparison and reflected appraisal. Although arguments in the literature on self-esteem are not generally framed this way, these processes are themselves influenced by the intersecting structures of gender, class, and age. For instance, research has shown that for young people at least, self-worth is embodied for women in a way that it is not for men. Relative to men, women are more likely to make social comparisons and reflected appraisals of themselves based on socially constructed ideals of feminine beauty. For girls and women, identity is tied to their relationships with boys and men. Hence, to the extent that boys and men believe that a girl or woman is beautiful and therefore worthy of affection, the level of self-esteem for these girls and women will be heightened. Class is an issue here because processes of social comparison and reflected appraisal may vary if a woman’s sense of identity is also linked to her paid work. Because middle class jobs and professional employment tend to be more meaningful than working class jobs, the connection between identity and paid work is more likely among middle and upper class women than it is among working class women. Alternatively,

80

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

economically well-off women, either in their own right or through their relationship with men, may have more resources to invest in maintaining their beauty, in ‘‘feeling good about themselves,’’ and in living up to the middle class ideal of ‘‘taking care of yourself.’’ Of course, middle class men are also subjected to this cult of self-care in which physical fitness and nutrition are heralded in attempts to look young and fit. Yet, there remains the question of identity salience. Corporeal appearance is more important for women than it is for men (Oberg & Tornstam, 1999), and as such, perceptions of attractiveness likely figure more into evaluations of self-worth for women than they do for men. In any case, gender and class intersect in the processes through which social comparisons and reflected appraisals are made. How does age influence this relationship? In most Western nations, beauty is socially constructed with youth in mind (Abu-Laban & McDaniel, 2001; Calasanti & Slevin, 2001). Consumer beauty products are marketed specifically to ‘‘combat’’ (as if engaged in a war) the effects of aging (Calasanti & Slevin, 2001). For women, and to a lesser extent men, to be attractive and successful is to be young and beautiful. Social comparisons and reflected appraisals of beauty and self-worth are often made with youth as a referent. Indeed, self-appraisals are also made with a younger self as the ideal to which the older self is judged (e.g., I am not as young as I used to be and I feel younger than I look). Of course, the connotations of these comparisons are most often negative (see Hurde Clarke, 2001). Hence, through processes of self-comparisons and reflected appraisals, one might expect that older people would have lower levels of self-esteem than younger people. Oberg and Tornstam (1999), in one of the few studies that systematically explores the relationship among gender, age group, and body image, showed that women cared more about their appearances than did men, regardless of age. Notably, however, 60% of men in this study agreed that their looks were important to them, and compared with younger men, men aged 75 and older were more likely to agree that their appearance was important. The results of this study also showed that compared with men, women worry more about how their looks will change as they grow older. Except among 75–85-yearold women, women were less satisfied with their bodies than were men. Although the level of body satisfaction does not vary by age group among men, older women had higher levels of body satisfaction than did younger women. In short, appearance matters for both men and women of all ages but the relationship among gender, age, and body image is complex (Oberg & Tornstam, 1999). Some of the complexity may be due to the relative influence of gender, class, and age identities (e.g., professional woman vs. woman professional) in people’s lives (see above). 2.2. Feminist research and the use of quantitative data analysis Most research on self-esteem takes an individualistic approach to interpretations of the relationships among socioeconomic status (SES), age, and sex, and self-esteem. These variables are thought to be attributes that individuals possess, and to the extent to which these variables vary individually, selfesteem will as well. Although this paper explores the relationship among SES, sex, age, and self-esteem using variables, we conceptualize these variables as incomplete proxies for the structures of class, gender, and age relations. Some readers may be offended by the use of quantitative analyses in feminist research. Indeed, we understand and appreciate feminist critiques of science and empiricism. However, just as scientific principles have stood in the way of feminist understanding, rejecting quantitative analyses as antifeminist may do the same. Over the last three decades, feminists have criticized their respective disciplines for ignoring women and thereby providing distorted pictures of social life. In doing so, feminist researchers have

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

81

been particularly critical of the ‘‘scientific method’’ and all that it entails. Harding (1987) is among the best known of these critics. According to Harding, feminist critiques of science are themselves problematic because they have tended to confuse research methods, methodologies, and epistemologies and have, as a result, underestimated the distinctiveness of feminist research. In Harding’s view, research methods are the techniques used to gather evidence (e.g., quantitative vs. qualitative methods). Methodologies are theories and analyses of ‘‘how research does or should proceed’’ (e.g., what questions need to be asked to assess particular theories) whereas epistemology is a ‘‘theory of knowledge’’ (e.g., positivism) (Harding, 1987, p. 3). Feminist scholarship has challenged ‘‘ways of knowing’’ by prioritizing women’s experiences in research, by developing research programs that are of interest to women and that are critical of hegemonic order, and by placing the research subjects and researchers on the same ‘‘critical plane’’ (Harding, 1987). According to Harding (pp. 9–10), these three issues are responsible for producing the best of the new feminist research and scholarship. They can be thought of as methodological features because they show us how to apply the general structure of scientific theory to research on women and gender. They can also be thought of as epistemological ones because they imply theories of knowledge different from the traditional ones. Notably, however, these are not issues that are bound to a single research method. Indeed, it is the nature of the research question that best determines what method should be used (Jayaratne & Stewart, 1991) and it is mistaken to assume that ‘‘one method is a priori more feminist than another’’ (Risman, Sprague, & Howard 1993, p. 608). In this paper, we use quantitative methods but we do not adapt a strictly positivist stance in which hypotheses are proposed that need to be tested. Such approaches are too restrictive and ‘‘bring with them the weight of old assumptions of science that tend to view the researchers as expert, to frame the inquiry before it has begun, to specify variables rather than ideas, and to predetermine the relationships that matter’’ (Connidis & McMullin, 2002, p. 596). Exploring research questions is akin to building puzzles. Quantitative data may provide insights into parts of the puzzle; qualitative data may provide insights into other parts. Theory and interpretation are what bring the parts of the puzzle together into a coherent picture. One interesting piece of the research puzzle that is often neglected in positivist work is an explicit statement how ideas become research questions. The idea for this paper was inspired by informal conversations that McMullin has had with women in varying social contexts. When discussions centered around growing old, McMullin noticed an interesting trend. Women in lower SES groups tended to reflect more negatively on the experience of growing old. Their assessments of aging were tied, almost exclusively, to their bodies. To the extent that their bodies were deteriorating relative to class-based idealized and socially constructed norms of beauty and youth, such assessments were negative. For these women, their self-esteem seemed threatened by the fact that they perceived themselves as ‘‘looking old.’’ On the other hand, women in higher SES groups tended to cherish the experience of growing old. They described the experience as liberating because men were no longer paying attention to the way they looked. These women talked about how they loved growing old because of the self-confidence and power they had gained over the years. Like Petty, these women had achieved high levels of self-esteem and self-confidence later in life; in many ways, these women felt empowered with increasing age (see Browne, 1998; Gibson, 1996).

82

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

McMullin’s discussions with these women and our reading of the literature on self-esteem led us to explore the following research questions in this paper: 1. Is there a relationship among age, SES, and self-esteem? If so, what is it? 2. Does gender influence the relationship among age, SES, and self-esteem? If so, how? We consider these research questions using quantitative data analyses that examine three-way interactions among age, gender, and income. Quantitative analyses are appropriate here because we are not examining process or meaning but rather the extent to which self-esteem varies based on key social factors. In the concluding section of this paper, we interpret our results within the context of the structured sets of power relations that are signified by age, gender, and income variables.

3. Methods 3.1. Sample Our sample is drawn from the National Population Health Survey (NPHS) conducted by Statistics Canada. The NPHS is a 1994 telephone survey of a national probability sample of Canadian residents across its 10 provinces. Using a multistage, stratified random sampling procedure, Statistics Canada interviewers surveyed 19,600 households in which one person in each household was selected to provide detailed personal information for the longitudinal component of the survey. Persons living on Native reserves, military bases, institutions, and some remote areas in Ontario and Quebec were excluded. Of the 18,342 possible respondents aged 12 and older, 17,626 participated, resulting in a response rate of 96.1%. After a list-wise deletion of cases with missing values, the total sample was further reduced to 16,051 (n = 15,694 weighted). This reduced sample was virtually identical to the original subsample in terms of age, gender, and marital status. 3.2. Self-esteem Self-esteem is a subset of six items derived from Rosenberg’s (1965, 1979) original 10-item scale. The items were scored using a five-point Likert scale to indicate how strongly one agrees or disagrees with a series of statements measuring self-worth. These statements are as follows: (1) You feel that you have a number of good qualities, (2) You feel that you are a person of worth at least equal to other people, (3) You are able to do things as well as most other people, (4) You take a positive attitude toward yourself, (5) On the whole, you are satisfied with yourself, and (6) All in all, you are inclined to feel you are a failure (reverse coded). The derived six-item scale in the NPHS had an internal consistency of a=.85. 3.3. Age, gender, and SES Age is an ordinal variable coded in 5-year intervals (12–14, 15–19, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70–74, and 75–79), except for the last interval, which includes those aged 80 and over. A quasi-continuous measure is created by setting each interval to the

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

83

midpoint of the range. In the last interval (80 and over), all respondents are set to 90 years of age. Gender is coded 1 for females and 0 for males (reference category in multivariate analysis). Because gender, age, and class influence marital status and because marital status has been found to have independent effects on various measures of self-worth, we included it in our analysis. Marital status is measured using three dummy variables for married (including common-law), previously married (including widowed, divorced, and separated), and single (reference category in multivariate analysis). Education and household income are the two measures of SES that are included in this analysis. Occupation was not used because questions regarding occupation were only asked of persons currently in the workforce, thus excluding individuals who had retired. Theoretically, there are advantages and disadvantages to using both household and personal income. We report the findings on household income here but note that it did not change when the analyses were run using personal income instead. Education is based on an item involving a combination of qualitative and ordinal categories. The original item is a 12-category variable, reflecting different levels of education attainment. While some categories appeared to be ordered, others do not. Thus, the original variable is recoded into eight categories: (1) no formal schooling, (2) elementary level education, (3) some secondary level education, (4) high school diploma, (5) some education beyond high school, (6) college diploma or trade certificate, (7) undergraduate university degree, and (8) graduate degree (MA or PhD) or a degree in Medicine. For multivariate analyses, this measure is treated as a continuous variable. Unfortunately, due to limitations in the data, we were not able to separate widowed individuals from those who were separated or divorced. Household income is coded into the following 11 intervals in the 1994 NPHS: (0) no income, (1) less than $5000, (2) $5000–9999, (3) $10,000–14,999, (4) $15,000–19,999, (5) $20,000–29,999, (6) $30,000–39,999, (7) $40,000–49,999, (8) $50,000–59,999, (9) $60,000–79,999, and (10) $80,000 and more. An 11-item scale is created by setting each scale value to the midpoint of the interval (e.g., 3=$12,499; mean = 44,900, S.D. = 27,700). For multivariate analysis, this variable is treated as continuous. Table 1 shows the basic descriptive statistics for the study. There are more women (55%) than men (45%) in our study, which reflects the age composition of the sample (e.g., the inclusion of individuals Table 1 Sample characteristics (n = 16,051 unweighted) Variables

Mean

S.D.

Self-esteem Age Education Income

19.90 4.49 40,000

3.00 1.59 26,700

n

%

Gender Female Male

8890 7161

55.4 44.6

Marital status Married Single Previously married

8621 4351 3078

53.7 27.1 19.2

84

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

over the age of 65). The majority of the sample is married (54%) followed by single individuals (27%) and those who were previously married (19%). The average score on the self-esteem measure is 19.0 (S.D. = 3.00). Average household income and education levels are $40,000/year (S.D.=$26,700) and 4.49 (S.D. = 1.59), respectively. The value for education corresponds roughly to completed high school and partial postsecondary level education, suggesting that on average, individuals in this survey have at least a high school level education. Table 2 OLS regression of self-esteem on gender, age, marital status, income, and education: 1994 NPHS Variables

Model 1

Model 2

Model 3

Model 4

Age

6.99E 02*** (0.045) 1.26E 04* ( 0.021) 0.20*** ( 0.033) 5.91E 02* ( 0.008) 0.26*** ( 0.038) 0.21*** (0.110) 9.22E 06*** (0.045)

6.21E 03** (0.045) 1.05E 04 ( 0.021) 0.20*** ( 0.033) 6.75E 02 ( 0.009) 0.26*** ( 0.038) 0.23*** (0.122) 8.25E 06*** (0.073) 1.25E 06 ( 0.014) 3.28E 05 ( 0.012) 5.19E 08 ( 0.009) 3.24E 09 (0.019)

8.65E 04 ( 0.006) 7.38E 06 ( 0.070) 0.15* ( 0.025) 0.10 ( 0.013) 0.27*** ( 0.041) 0.23*** (0.123) 8.14E 06*** (0.072) 1.47E 03 ( 0.016) 3.04E 05 ( 0.011) 3.38E 08 ( 0.006) 3.38E 09 (0.019) 1.32E 02*** (0.07) 2.05E 04 ( 0.031)

18.82 .031

18.76 .031

18.73 .033

1.09E 04 ( 0.001) 1.23E 05 (0.002) 8.26E 02 (0.014) 0.12 ( 0.016) 0.28*** ( 0.041) 0.23*** (0.12) 1.09E 05*** (0.097) 1.22E 03 ( 0.014) 5.37E 05 ( 0.020) 3.40E 08 (0.006) 1.73E 09 (0.010) 2.00E 02* (0.100) 5.33E 04 ( 0.080) 1.30E 07** ( 0.051) 3.26E 09 (0.015) 5.33E 04 ( 0.011) 3.93E 05 (0.023) 18.61 .033

Age-squared Female Previously married Single Education Income Education  Age Education  Age-squared Income  Age Income  Age-squared Female  Age Female  Age-squared Female  Income  Age Female  Income  Age-squared Female  Education  Age Female  Education  Age-squared Intercept r2 * P < .05. ** P < .01. *** P < .001.

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

85

4. Results In this study, we test for two-way and three-way interactions among age, SES, gender, and self-esteem (see Table 2). In Model 1, we regress self-esteem on age, gender, marital status, income, and education (main effects model). We include a quadratic term for age to capture any potential non-normality in the relationship between age and self-esteem. Both age and age-squared variables are significant and indicate that older persons report lower levels of self-esteem than do younger persons. In addition to age, women report lower levels of self-esteem than do men. Those who were previously married and those who are single report lower levels of self-esteem than those who are married. Both measures of SES are significantly related to self-esteem, net of these other predictors. As education and income increase, selfesteem also increases. In Models 2 and 3, we introduce interaction terms for age by SES and age by gender, respectively. The nonlinearity in the age by self-esteem relationship requires two sets of interaction terms for each variable (education, income, and gender). The only interaction term to reach statistical significance is age by gender (see Model 3). The dummy variable for single, education, and income continue to be significantly related to self-esteem in Model 3. Finally, in Model 4, we introduce three-way interaction terms for age, gender, SES, and selfesteem. Only one term reaches statistical significance. There is a significant three-way interaction among age, gender, and income in predicting self-esteem. The coefficient representing the interaction was significant at the P < .01 level. To interpret the three-way interaction, we graphed the relationship separately for women (see Fig. 1) and men (see Fig. 2). The pattern of association is similar with for both men and women. Regardless of sex, income differences in self-esteem are virtually nonexistent in early adolescence through to middle age. By about age 62, however, there is a divergence in self-esteem by income groups; individuals in the highest income groups (1 S.D. above the mean for household income) report higher levels of self-esteem than those in the lowest income group (1 S.D. below the mean). This gap widens until age 90 where income differences are the most pronounced. For men, average self-esteem scores are somewhat higher across age groups when compared with women. Moreover, the income gap in self-esteem at age 90 is somewhat wider

Fig. 1. Age, income, and self-esteem: women only.

86

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

Fig. 2. Age, income, and self-esteem: men only.

among men. Finally, the rate of decline in self-esteem and the rate of divergence in income with age appear steeper for women. In the presence of these interactions, education remains significantly and positively related to self-esteem, and single persons have lower levels of self-esteem than do married persons.

5. Discussion and conclusions This paper takes a modest step in piecing together the complexities of the relationships among SES, gender, age, and self-esteem. Earlier in this study, we note that the relative salience of body image to identity formation acts as a mechanism through which self-esteem is made manifest in individuals. Intersecting structures of power, combined with the interpersonal processes of social comparison and reflected appraisals, likely operate simultaneously in influencing the processes through which body images influence identities. Hence, the identities of those in structurally powerful positions in society may not be as reliant on corporeal appearance as are the identities of those in structurally less powerful positions. This, combined with the relative control that various groups exercise in society, influences one’s sense of self-worth. In general, our findings support these premises. We show that, contrary to some past research that suggests that age has little influence on self-esteem or that self-esteem increases with age, levels of self-esteem are lower in older age groups for both men and women. Furthermore, in all age groups, women have lower levels of self-esteem than do men. Corresponding with past research, income tends not to influence levels of self-esteem for young men or women but does so for those in middle age. By including persons aged 65 and over in our study, we see that the most pronounced income differences are for those in later life. Hence, our research adds support to the cumulative advantage/disadvantage hypothesis that suggests that class advantage/disadvantage is heightened in old age (Dannefer & Sell, 1988; O’Rand, 1996a, 1996b). With respect to self-esteem, the cumulative class advantage/disadvantage (as measured by income) seems somewhat greater for men than for women. However, there also seems to be a cumulative advantage related to gender

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

87

that benefits men more than women (hence the steeper decline in self-esteem for women than for men). Adherents of positivism may dismiss these findings because the coefficients and explained variance in our analyses are very small. Some feminist scholars, on the other hand, may dismiss our results because they are grounded in quantitative analysis. Yet, because these findings make substantive and theoretical sense, they add valuable, albeit incomplete, information to the research puzzle. Hence, it is premature to dismiss these findings as antifeminist simply because they rely on quantitative analyses or to do so based on their limited mathematical magnitude. Our feminist and interpretive approach implies that sex, income, and chronological age are proxies for gender, class, and age structures; they are not merely variable attributes of individuals. Feminist gerontology informs the interpretations of our results by beginning with the assumption that class, age, and gender intersect in structuring all aspects of social life. The theoretical arguments presented in this paper are that power, social comparisons, and reflected appraisals influence self-esteem. Hence, it is not individual sex, income, and age differences that explain the variation in self-esteem found in this paper. Rather, we argue that gender, class, and age relations inform the relative power that individuals hold in society as well as the interpersonal processes of social comparison and reflected appraisals. Those with less power have fewer life chances available to them and are poorly positioned to make favorable social comparisons and reflected appraisals. Hence, women, members of the working class, and older adults suffer from lower levels of self-esteem because of their structurally disadvantaged position in society. We would be remiss to not speculate about the relationship between our findings and physical health. It is well established that changes in physical health status with age are associated with negative changes in self-perceptions (see Avison & Cairney, 2003; Mirowsky, 1995). The loss of independence and altered or disputed interpersonal relationships that result from physical illness all contribute to an eroded sense of self-worth (e.g., Penninx et al., 1996). However, the interrelationships among age, class, gender, physical health, and self-conceptions are apt to be quite complex. For example, individuals who are in poor health may suffer from lower levels of self-esteem, and income, age, and gender all influence physical health. Indeed, the findings presented in this paper mirror the findings that one might expect if physical health was the dependent variable (see Cairney & Arnold, 1996). The possibility that gender by income differences in self-esteem among older adults may be attributable to gender by income differences in physical health has not been explored in the literature and is worthy of further investigation. In particular, the extent to which self-esteem is connected to gendered differences in coping responses to different kinds of physical illness is a particularly interesting question for further consideration. This paper only provides a glimpse at the relationship between social structure and self-esteem, and as such, we can only speculate about how and why gender, class, and age influence selfesteem levels. More research is needed on this topic and also on the relationship between selfesteem and education throughout the life course. The theoretical speculation presented in this paper about class, gender, and age would have been better supported had education, gender, and age produced a significant interaction. The fact that our analyses did not produce such an interaction may also speak to the limitations of quantitative data analysis. Thus, further study would benefit from a qualitative research design and continued use of feminist epistemology and methodologies.

88

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

Another limitation of this study is that it only explores three of the dominant structures of inequality in North America. There is little doubt that the findings presented in this paper would be different if structures of race, ethnicity, sexuality, and so on were considered. However, because of data limitations in the NPHS, we must leave this task to other researchers. Finally, our study relies on cross-sectional data that confound cohort and aging effects. At the same time, our research contributes to the literature on self-esteem in moving away from developmental, physiological, or role-loss approaches to aging. Our theoretical stance is that an age-based power structure exists in society and that this power structure acts as a detriment to self-esteem in later life. Indeed, cross-sectional data allow us to explore this possibility. The facts that power decreases in later life, that old age is unvalued in society, and that beauty is socially constructed with youth in mind place older people at risk of experiencing low levels of self-esteem relative to younger people. Nonetheless, further research would benefit from both longitudinal research designs and retrospective life history accounts of self-esteem.

Acknowledgements We thank Toni Calasanti who provided us with many helpful comments and suggestions as we were writing this paper.

References Abell, S. C., & Richards, M. H. (1996). The relationship between body shape satisfaction and self-esteem: An investigation of gender and class differences. Journal of Youth and Adolescence, 25, 691 – 703. Abu-Laban, S. M., & McDaniel, S. (2001). Beauty, status, and aging. In N. Mandell (Ed.), Feminist issues. Race, class, and sexuality (3rd ed.) (pp. 108 – 133). Toronto: Prentice-Hall, Chapter 5. Arber, S., & Ginn, J. (Eds.) (1995). Connecting gender and ageing: A sociological approach. Buckingham: Open University Press. Avison, W. R., & Cairney, J. (2003). Social structure, stress and personal control. In H. Steven, L. Zarit, I. Pearlin, & K. W. Schaie (Eds.), Personal control in social and life course contexts ( pp. 127 – 164). New York: Springer. Avison, W. R., & McAlpine, D. (1992). Gender differences in symptoms and depression among adolescents. Journal of Health and Social Behavior, 33, 77 – 96. Browne, C. V. (1998). Women, feminism, and aging. New York: Springer. Cairney, J., & Arnold, R. (1996). Social class, health and aging: Socioeconomic determinants of self-reported morbidity among the non-institutionalized elderly in Canada. Canadian Journal of Public Health, 87(3), 199 – 203. Calasanti, T. M. (1996). Incorporating diversity: Meaning, levels of research, and implications for theory. Gerontologist, 36, 147 – 156. Connidis, I. A., & McMullin, J. A. (2002). Sociological ambivalence and family ties: A critical perspective. Journal of Marriage and Family, 64(3), 558 – 567. Calasanti, T. M., & Slevin, K. F. (2001). Gender, social inequalities, and aging. New York: Altamira. Dannefer, D., & Sell, R. R. (1988). Age structure, the life course and ‘‘aged heterogeneity’’: Prospects for research and theory. Comprehensive Gerontology B, 2, 1 – 10. Dietz, B. E. (1996). The relationship of aging to self-esteem: The relative effects of maturation and role accumulation. International Journal of Aging and Human Development, 43, 249 – 266. Estes, C. L. (1999). The new political economy of aging: Introduction and critique. In M. Minkler, & C. Estes (Eds.), Critical gerontology ( pp. 17 – 35). Amityville, NY: Baywood, Chapter 1.

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

89

Furman, F. K. (1997). Facing the mirror: Older women and beauty shop culture. New York: Routledge. Gecas, V., & Seff, M. A. (1990). Social class and self-esteem. Psychological centrality, compensation and the relative effects of work and home. Social Psychological Quarterly, 53, 165 – 173. Giarrusso, R., Mabry, J. B., & Bengtson, V. L. (2001). The aging self in social contexts. In R. H. Binstock, & L. K. George (Eds.), Handbook of aging and the social sciences (5th ed.) (pp. 295 – 312). San Diego: Academic Press. Gibson, D. (1996). Broken down by age and gender: ‘‘The problem of old women’’ redefined. Gender and Society, 10, 433 – 448. Harding, S. (1987). Introduction: Is there a feminist method? In S. Harding (Ed.), Feminism and methodology ( pp. 1 – 14). Bloomington: Indiana University Press, Chapter 1. Hill Collins, P. (2000). Moving beyond gender: Intersectionality and scientific knowledge. In M. M. Ferree, J. Lorber, & B. B. Hess (Eds.), Revisioning gender ( pp. 261 – 284). New York: Altamira. Hurde Clarke, L. (2001). Older women’s bodies and the self: The construction of identity in later life. Canadian Review of Sociology and Anthropology, 38(4), 441 – 464. Jayaratne, T. E., & Stewart, A. J. (1991). Quantitative and qualitative methods in the social sciences: Current feminist issues and practical strategies. In M. Fonow, & J. Cook (Eds.), Beyond methodology: Feminist scholarship as lived research ( pp. 85 – 118). Bloomington: Indiana University Press, Chapter 5. Josephs, R. A., Markus, H. R., & Tafarodi, R. W. (1992). Gender and self-esteem. Journal of Personality and Social Psychology, 63, 391 – 402. McMullin, J. A. (2000). Diversity and the state of sociological aging theory. Gerontologist, 40, 517 – 530. McMullin, J.A. (2004). Understanding social inequality: Intersections of class, age, gender, ethnicity, and race in Canada. Toronto: Oxford University Press. Mirowsky, J. (1995). Age and the sense of control. Social Psychology Quarterly, 58, 31 – 43. Mirowsky, J., & Ross, C. E. (1996). Economic and interpersonal rewards: Subjective utilities of men’s and women’s compensation. Social Forces, 75, 223 – 245. Oberg, P., & Tornstam, L. (1999). Body images among men and women of different ages. Ageing and Society, 19, 629 – 644. O’Rand, A. (1996a). The precious and the precocious: Understanding cumulative disadvantage and cumulative advantage over the life course. Gerontologist, 36, 230 – 238. O’Rand, A. (1996b). The cumulative stratification of the life course. In R. H. Binstock, & L. K. George (Eds.), Handbook of aging and the social sciences (4th ed.) (pp. 188 – 207). New York: Academic Press. Penninx, B. W. J. H., Beekman, A. T. F., Ormel, J., Kriegsman, D. M. W., Boeke, A. J. P., Van Eijk, J. T. M., & Deeg, D. J. H. (1996). Psychological status among elderly people with chronic diseases: Does type of disease play a part? Journal of Psychosomatic Research, 40(5), 521 – 534. Perlin, L. I., Lieberman, M. A., Menaghan, E. B., & Mullan, J. T. (1981). The stress process. Journal of Health and Social Behavior, 22, 337 – 356. Pugliesi, K. (1995). Work and well-being. Gender differences in the psychological consequences of employment. Journal of Health and Social Behavior, 36, 57 – 71. Risman, B., Sprague, J., & Howard, J. (1993). Comment on Francesca M. Cancian’s ‘‘Feminist Science’’. Gender and Society, 7, 608 – 609. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Rosenberg, M. (1979). Conceiving of self. New York: Basic Books. Rosenberg, M. (1981). The self-concept: Social product and social force. In M. Rosenberg, & R. H. Turner (Eds.), Social psychology: Sociological perspectives ( pp. 593 – 624). New York: Basic Books. Rosenberg, M. (1985). Self-concept and psychological well-being in adolescence. In R. L. Leahy (Ed.), The development of the self ( pp. 205 – 246). Orlando, FL: Academic Press. Rosenberg, M., & Pearlin, L. I. (1978). Social class and self-esteem among children and adults. American Journal of Sociology, 84, 53 – 77. Rosenberg, M., Schooler, C., & Schoenbach, C. (1989). Self-esteem and adolescent problems: Modeling reciprocal effects. American Sociological Review, 54, 1004 – 1018. Rosenfield, S. (1989). The effects of women’s employment: Personal control and sex differences in mental health. Journal of Health and Social Behavior, 30, 77 – 91. Rosenfield, S. (1999). Splitting the difference: Gender, the self, and mental health. In C. S. Aneshensel, & J. C. Phelan (Eds.), Handbook of the sociology of mental health ( pp. 209 – 224). New York: Kluwer Academic, Chapter 11.

90

J.A. McMullin, J. Cairney / Journal of Aging Studies 18 (2004) 75–90

Stoller, E. P., & Gibson, R. C. (1997). Worlds of difference: Inequality in the aging experience. Thousand Oaks, CA: Pine Forge. Thoits, P. (1999). Self, identity, stress, and mental health. In C. S. Aneshensel, & J. C. Phelan (Eds.), Handbook of the sociology of mental health ( pp. 345 – 368). New York: Kluwer Academic, Chapter 17. Turner, R. J., & Roszell, P. (1994). Psychosocial resources and the stress process. In W. R. Avison, & I. H. Gotlib (Eds.), Stress and mental health: Contemporary issues and prospects for the future ( pp. 179 – 212). New York: Plenum, Chapter 7.