• Nicola J. Newton and Abigail J. Stewart, Department of Psychology, University of Michigan.

  • We extend our thanks to the Personality and Social Contexts research group and Jacquelynne Eccles for their insightful comments concerning earlier drafts of this article, as well as Kai Cortina, Kathy Welch, and David Winter for statistical consultation. Collection of the data reported in this research has been supported by the Boston University Graduate School, National Science Foundation Visiting Professorships for Women, the Society for the Psychological Study of Social Issues, the MacArthur Foundation Network for Research on Successful Midlife Development, Radcliffe Research Support and Midlife Program Grants from the Henry A. Murray Research Center, National Institute of Mental Health subgrants under Grants 1-RO1-MH43948 and 1-RO1-MH47408, National Institute on Aging Training Grant T32-AG0017, and the University of Michigan. Computer-accessible and other data from several previous waves are archived at the Henry A. Murray Research Center, Radcliffe College, 10 Garden Street, Cambridge, MA 01238.

Address correspondence and reprint requests to: Nicola J. Newton, Department of Psychology, University of Michigan, 530 Church Street, Ann Arbor, MI 48109-1403. E-mail:


It has been argued that the predominant focus of midlife personality development is generativity; other research has found that social roles influence both its onset and its expression. In this article, we examine women's midlife personality development and its relationship to career and family commitments. Results for a sample of 90 women indicated that commitment to particular social projects in early midlife was associated with different patterns of identity, intimacy, and generativity levels at age 62. In addition, women who added social projects to existing projects during adulthood expressed similar levels of identity, intimacy, and generativity at age 62 as women whose social project commitments had stayed the same. For smaller subsamples of women in the study, longitudinal analyses assessed changes in personality development within middle age as well as the relationship between personality and the maintenance or addition of social projects. These results highlight important variation among women who followed different adult life paths.

Much has been written regarding middle age and personality, but little scholarship focuses on the nature of personality change within this life period. Research has focused on whether there is a midlife crisis (Levinson, Darrow, Klein, Levinson, & McKee, 1978; Levinson & Levinson, 1996) or on what may be predominant personality themes in middle age (generativity according to Erikson, 1982, or executive personality according to Neugarten, 1967, 1968). Little or no theory has grappled with the process of personality development or change within the period, despite the fact that various researchers have found evidence of it, perhaps especially among women (Helson, Mitchell, & Moane, 1984; Helson, Soto, & Cate, 2006; Josselson, 2003; Kroger, 1997; Raskin, 2002; Stewart, Ostrove, & Helson, 2001). Moreover, the issue of personality change in middle age must be considered along with acknowledgment that many middle-aged women face changed circumstances across the period and that they differ from each other in those circumstances (see Staudinger & Bluck, 2001). For example, a 40-year-old mother of college-age children has different concerns from a 60-year-old grandmother preparing for retirement. Similarly, a business executive in the prime of her career in her 40s has different concerns from a stay-at-home mom of the same age, and both may develop different concerns and outlooks as they age. Thus, our focus is on women's diverse paths in adulthood, and their implications for personality change in middle age.

There is growing support among personality researchers for approaches that highlight elements of both change and continuity in adult personality, especially in research with women (Helson, Pals, & Solomon, 1997; Helson & Soto, 2005; Roberts, Helson, & Klohnen, 2002; Vandewater & Stewart, 1998). Researchers not only find that women's personalities develop after the age of 30 (Helson, Jones, & Kwan, 2002) but also emphasize the impact of social roles—such as wife, mother, and career woman—on personality (Vandewater & Stewart, 1998) as well as the influence of major social events and changes in the social environment (Helson et al., 1984, 1997; Stewart & Vandewater, 1993).

One factor that seems to change less than Erikson (1950, 1959/1980, 1968) initially hypothesized is identity. Women's focus on it may remain relatively high in early middle age as women continue to develop their talents and careers; it may then remain stable or decline later in the period. In a similar way, generative concerns, which Erikson (1982) proposed would arise earlier and then dominate in middle age, may actually begin relatively low, only increasing by the end of the period (Stewart & Vandewater, 1998). In tracking the developmental course of these personality themes in middle age, we are not particularly interested in how individuals temporarily or permanently resolve their stance toward each concern (McAdams, de St. Aubin, & Logan, 1993; Ochse & Plug, 1986; Sneed, Whitbourne, & Culang, 2006). Instead, the important issue is how much they are focused on these themes (Erikson, 1982; MacDermid, Heilbrun, & DeHaan, 1997; Stewart & Vandewater, 1993; Van Manen & Whitbourne, 1997; Vandewater & Stewart, 1998). In our study, we use Erikson's (1982) concepts of identity, intimacy, and generativity to examine how early commitments adult women make concerning work and family affect their later personality development and how their personalities change over this period.

Social Roles and Women's Personality Development

Neugarten proposed that the “social clock,” or the normative timing for undertaking social roles such as wife and mother (Neugarten, Moore, & Lowe, 1965/1996), may provoke further personality development. Neugarten and her colleagues pointed out that there are widely shared beliefs and conventions about the usual and ideal flow of life events. They noted social beliefs about the appropriate ages for certain life changes, such as getting married, having children, and retiring. Neugarten and Datan (1973/1996) also illustrated the influence of changing historical and social time on the social clock for different cohorts by comparing the different life trajectories of women in 1890 and 1966, suggesting that social norms about these behaviors are different during different historical periods. Expanding further on Neugarten and colleagues' work and applying it to a more recent cohort of women, Helson et al. (1984) studied individual differences in patterns of social clock adherence, and they highlighted the variability in commitment to “on time” social clock projects in a group of women who graduated from Mills College in 1958 and 1960. They found that these women's midlife personalities had been influenced by their earlier choice of social clock projects.

In the current study, we extend this research on the social clock to include the implications for later personality development of early commitments to social projects. As we consider the literature concerning personality development and the influence of social roles, it is important to remember that women's social roles have changed considerably in scope, especially since the 1970s, in terms of normative societal expectations for marriage, family, and career.

Erikson's Personality Development Themes

Identity According to Erikson (1968), a particular emphasis on creating a personal identity based on knowledge of one's self and how one fits into the greater societal context is associated with adolescence and early adulthood. He viewed women's identities as necessarily provisional until they made an intimate life commitment; women were destined to be predominantly nurturing and accommodating (see Franz & White, 1985, for an excellent critique).

For many women, midlife is a time of equal focus on identity and generativity (Helson, 1992; Helson, Stewart, & Ostrove, 1995; Stewart, Franz, & Layton, 1988). Helson et al. (1995) found that women's identities were affected by social roles if they had been able to make individual choices about family and careers. Equally, midlife identity for women of the baby boom was influenced by the number and quality of early adult role involvements, including career commitments (Josselson, 2003; Kroger, 2002; Vandewater, Ostrove, & Stewart, 1997). Renegotiating identity has also been found to be a continuing concern for women in midlife and beyond (Kroger, 1997). In short, contemporary research and theory suggest that identity themes may remain quite high for many women at the start of middle age, beginning to decline by late middle age and depending on the cohort studied and the influence of their social roles.

IntimacyErikson viewed intimacy as the developmental stage when young adults, having adequately resolved their identity issues, are “… eager and willing to fuse [this] identity with that of others” (1950, p. 263), in both friendship and love relationships (1959/1980, p. 101). Consistent with Erikson's theorizing, Ochse and Plug (1986) found general intimacy themes to be predominant between the ages of 20 to 25 years.

Espin, Stewart, and Gomez (1990) analyzed letters from a Latin American emigrée to her teacher; she had already resolved intimacy issues at the age of 22, by which time she was married and had her first child. In this and another study of a young woman's personal documents (diaries and letters; Stewart et al., 1988), adoption of adult social roles was associated with changes in expression of intimacy themes. In short, intimacy themes should be low in middle-aged women, according both to Erikson's theory and to the existing research.

Generativity According to Erikson (1950), midlife is the time when it is most likely that men and women concern themselves with creating a legacy that outlives them. Erikson defined this focus as generativity, or “the concern in establishing and guiding the next generation” (p. 267). However, evidence also suggests considerable variability regarding when generativity first emerges (McAdams, 2001; Peterson & Stewart, 1993) and when it peaks (Stewart et al., 2001; Peterson & Stewart, 1993; Zucker, Ostrove, & Stewart, 2002). Further difficulty has been observed in viewing Erikson's stages as discrete (Josselson, 2003). Additionally, Kroger (1997) suggested that generativity and identity are often connected in middle age. Based on the findings we reviewed here, and in keeping with Erikson's theory, we believe that generativity not only will be an important theme in middle age, but also will increase from early to late midlife.

The relationship between social project commitment and generativity in the middle years for women from the Radcliffe class of 1964 was examined by Stewart and Vandewater (1993). Specifically—and pertinent to the current study—women who had committed to families but not careers at age 28 were more concerned with generativity at age 43, whereas those committed to careers but not families at age 28 had little focus on generativity at age 43. Although parenting is generally associated with generativity in theory, Peterson and Stewart (1996) demonstrated that career women can also be generative through occupational mastery and helping others. Peterson (2002), in a study of intergenerational roles, also found that women who provided care for a parent, particularly fathers, expressed higher levels of generativity.

The present study extends the research outlined above by examining (a) women's personality change across the period of early and late middle age and (b) patterns of personality change as a function not only of social role commitments, but also of changes in those commitments during early middle age. We will test four hypotheses concerning social project commitment and personality change in midlife. First, we propose that commitment to a family project by age 43 will be associated with more expressions of generativity for women at age 62, whereas commitment to a career project by 43 will be associated with more expressions of identity at 62. Second, we propose that women who add a new social project in middle age will have patterns of personality development similar to those who have maintained that same social project. Third, we propose that, in both cases, the different development trajectories will result in different within-age profiles of personality for the women concerned. Specifically, women with only family projects will express more generativity than identity, whereas women with only career projects will express more identity than generativity. Furthermore, women with both projects will be more balanced in their expression of these two themes. Finally, regardless of social clock projects, we propose that women's expressions of generativity will increase on average, while expressions of identity and intimacy will remain relatively stable or decrease over the same period.


Radcliffe Longitudinal Sample

Women of the Radcliffe College Class of 1964 have been studied since they were first-year students in 1960 when 244 women took part in the initial survey (see Stewart, 1978, 1980; Stewart & Vandewater, 1993, for a description of the original sample); since then, over 100 women have participated in each subsequent assessment. The women who participated in this overall study were White, generally came from privileged backgrounds, and received an excellent education. In the most recent follow-up in 2005, 105 women completed a mailed or online questionnaire that sought basic demographic information as well as open- and closed-ended questions concerning a variety of personality data as well as life events and influences. By 2005, respondents averaged 62 years old; 60% of these women were currently married (46% had been divorced at some stage), 63% were engaged in salaried employment, and 85% reported being in above-average to excellent health.

Historically, these women represent a transitional cohort, as described by Stewart and Vandewater (1993). Born during World War II, they were subject to conflicting gender norms as a generation. They spent their childhood and adolescence under pressure to conform to traditional gender roles, but during their early adulthood they experienced increased opportunities for women in graduate education and the labor force as well as second-wave feminist ideas about gender.

For our study, in addition to the 2005 data, we included data collected in 1986 and 1996, when the women were 43 (N = 103) and 53 (N = 119) years old, respectively. This procedure allowed us to conduct longitudinal analyses (from ages 43 to 62). These analyses involved a subsample of 42 women who had participated in all three data collection waves. The representativeness of the longitudinal sample of women compared to those who only participated once or twice was ascertained using a series of separate t tests; there were no significant differences in the demographic, health, and life change variables available.


Social projects Following Helson's work regarding adherence to the social clock (Helson et al., 1984), we assessed the extent to which the Radcliffe women stayed with two social projects (career and/or family) at age 28 and again at age 43. Women who had married and had children by age 28 but did not pursue careers were categorized as having made a commitment to family only. Women were categorized as committed to career only if they were focused on a career, whether married or unmarried, and if they had not had children by age 28. A third group was created to indicate commitment to both projects. Age 28 social project commitments were used to provide us with an early adult baseline for assessing changes in social projects between early adulthood (age 28) and midlife (age 43).

Personality development themes In 2005, as in previous data collections, three of Erikson's (1950, 1968, 1959/1980, 1982) themes—identity, intimacy, and generativity—were assessed from open-ended questions, following the coding system for assessing these themes originally developed by Stewart et al. (1988) and revised by Stewart, Franz, Paul, and Peterson (1991). Thus, expressions of goals for the next few years and high points in the last few years were identified as reflecting these three factors in terms of their similarity to the definitions and examples contained in Erikson's accounts. This coding system has been used extensively to assess these three themes in both examinations of personal documents, such as letters, diaries, and autobiographical writings, as well as open-ended questions in survey studies (Espin et al., 1990; Franz, 1988; Peterson, 1993; Peterson & Stewart, 1990; Stewart et al., 1988).

Two questions asking about future goals and recent high points were coded: “If you could do anything you wished in the next 10 years, what would you do?” and “Looking back over the last nine years, what do you consider major high points, or the most satisfying activities?” The length of responses to these questions was a few lines (ranging from 29 to 34 words for goals and 52 to 76 for high points). A typical example for goals is: “Stay healthy, keep traveling. Phase myself out of my job gracefully, find other ways to spend my time productively after retirement. Handle my parents' deterioration effectively.” and for high points is:

There are two major high points both of which involved risk. One was leaving a good, upper level management position after 15 years working for the same organization and the second was leaving a marriage of 25 yrs. Both took much courage and I am not a “risk taker” by heart; however, I have been happier, more complete & more fulfilled by doing so.

These representative responses not only demonstrate the presence of more than one overall theme (identity, intimacy, or generativity) in many answers, but also highlight the presence of different aspects of each theme in a single response. Thus, subcategories were used as guides to codable content; however, only overall themes (of identity, intimacy, and generativity) were scored. We summed responses to the two questions (future goals and high points) for analysis purposes.

Identity References to identity in response to the two questions were coded if they focused on career identity or on personal aspects of identity. Statements scored for career identity included responses such as “Widened scope of my job; took hard courses to improve background knowledge”; coding for this particular category often required taking into account the woman's actual occupation. Expressed themes regarding personal identity included, for example, statements dealing with traits or stable characteristics: “I lack confidence in myself,” or “What I'm doing now, but with better organizing skills.”

Intimacy Statements scored for intimacy focused on issues of relationships, such as mutuality, sexuality, and more general themes. They included “… resuming dating; new relationships,”“… the discovery of so much joy in this partnership,” and “Keep our sex life going.” These references were summed across the two questions to give an overall intimacy score.

Generativity Statements concerning generativity focused on references to generative legacy, generative productivity, generative caring, and the need to be needed. Responses to the two questions were again summed. Expressions included references to generative legacy, including children, grandchildren, or a personal contribution of some kind to society. For example, responses such as “The nurturing, even in their young adulthood, of my four children,” or “Seeing students (especially graduate students) blossom with their own independent research programs” were coded for generative legacy. Statements coded for generative productivity mentioned some type of work on a personal project that had lasting impact, for example, “… creating gardens that are truly comforting to be in” or “Finishing my science textbook.” Expressions of generative caring included statements concerned with the capacity to care for others, such as “Doing as much as possible for my parents in their declining years …” Finally, statements scored for the need to be needed included expressions of being needed by others, such as “Feel of some use” or “I would like to be of service to others.”

Responses from the 1986 wave had been previously scored as described in Stewart and Vandewater (1993). The first current author coded responses to the same high points and future goals questions from the 1986 data collection, achieving 85% inter-rater agreement with the coding undertaken at that time by graduate students in psychology. Both current authors then coded the responses to the open-ended questions from 1996 and 2005 for identity, intimacy, and generativity themes without knowledge of any other information about participants, except when occupation was necessary in coding for identity. Their inter-rater agreement (calculated as twice the number of the raters' agreements on the presence of a category divided by the sum of both raters' scoring for that category, per the account in Smith, Feld, & Franz, 1992) was 91% for the combined 1996 and 2005 data. Following standard practice to avoid assessing sheer verbal fluency, scores were corrected by regression for their correlation with the number of words written in responses (Winter, 1973; Smith et al., 1992).


Two sets of results are presented here: (a) those addressing the relationship between social projects at age 43 and levels of identity, intimacy, and generativity at age 62 (N = 90) and (b) those addressing changes in levels of identity, intimacy, and generativity over time (between ages 43 and 62; N = 42). For both sets of results, we used general linear modeling in order to assess change in identity, intimacy, and generativity at ages 43, 52, and 62 for each social project pattern: career, family, or their combination.

Social Projects and Personality Development

Table 1(a) shows the results of a repeated-measures analysis of variance (ANOVA). Overall mean levels of expressed identity, intimacy, and generativity themes for the 90 women for whom we have data at age 62 were significantly different among the three social project groups to which they were committed by age 43, F(2, 86) = 36.45, p < .001, inline image. Within social project, each group exhibited a different profile of personality themes. Consistent with our prediction, career-only women at the age of 62 expressed significantly more identity themes than both intimacy and generativity themes, F(1,13) = 11.04, p = .006, inline image, and F(1,13) = 6.90, p = .02, inline image, respectively. Similarly, family-only women expressed more generativity themes than both intimacy and identity themes, F(1,39) = 30.86, p < .001, inline image, and F(1,39) = 11.84, p = .001, inline image, respectively. Women committed to both family and career expressed both more identity and more generativity themes than intimacy themes, F(1,35) = 14.83, p < .001, inline image, and F(1.35) = 43.39, p < .001, inline image, respectively.

Table 1. 
Identity, Intimacy, and Generativity Themes at Age 62 by Social Project Group (Cross-sectional and Longitudinal)
Age 62 themesSocial Project Commitment by Age 43
  1. Note. Means (and standard deviations) are reported for both (a) age 62 themes and (b) age 62 themes for those women who participated at all three data collections. Means sharing a common alphabetical subscript within social project group (reading down columns) are not significantly different from each other at p < .05.

(a) Cross-sectional(N = 14)(N = 40)(N = 36)
Identity 1.98 (1.70)a0.48 (1.17)d1.25 (1.66)f
Intimacy 0.16 (0.76)b0.18 (0.75)d0.14 (0.55)e
Generativity 0.27 (1.37)b1.81 (1.67)c1.97 (1.49)f
(b) Longitudinal(N = 6)(N = 18)(N = 18)
Identity 1.26 (1.36)a0.21 (1.16)c1.05 (2.00)e
Intimacy 0.07 (0.55)a0.18 (0.80)c0.04 (0.46)d
Generativity−0.12 (0.98)a2.17 (1.96)b2.33 (1.43)e

We also analyzed the relationship between addition or maintenance of a social project between the ages of 28 and 43 with levels of identity, intimacy, and generativity at age 62 for those women for whom we also had age 28 social project data (N = 80; see Table 2(a)). For this analysis, the sample was divided into four groups: (a) career-always women between ages 28 and 43, (b) those who added a family by age 43 to their career commitment at age 28, (c) family-always women, and (d) those who added a career by age 43 to their family commitment at age 28.

Table 2. 
Identity, Intimacy, and Generativity Themes at Age 62 by Social Project Group Status Between Ages 28 and 43 (Cross-sectional and Longitudinal)
Age 62 themesSocial project groups
Career always 28 → 43Family only at 28 → both family and career at 43Family always 28 → 43Career only at 28 → both career and family at 43
  1. Note. Means (and standard deviations) are reported for both (a) age 62 themes and (b) age 62 themes for those women who participated at all three data collections: 1986, 1996, and 2005. Means sharing a common alphabetical subscript within social project group (reading down columns) are not significantly different from each other at p < .05.

(a) Cross-sectional(N = 14)(N = 9)(N = 23)(N = 34)
Identity 1.98 (1.70)a2.10 (1.82)d0.35 (1.24)f 1.00 (1.38)g,
Intimacy 0.16 (0.76)b0.26 (0.54)c0.01 (0.52)f0.20 (0.77)h
Generativity 0.27 (1.37)b1.77 (1.54)d2.11 (1.94)e1.86 (1.47)i
(b) Longitudinal(N = 6)(N = 5)(N = 11)(N = 16)
Identity 1.26 (1.36)a 2.75 (2.17)b,c−0.16 (1.18)e0.67 (1.33)g
Intimacy 0.07 (0.55)a0.21 (0.59)c0.06 (0.71)e0.07 (0.56)g
Generativity−0.12 (0.98)a1.81 (0.92)b2.65 (2.16)d2.28 (1.62)f

The overall pattern of results for this repeated-measures ANOVA was similar to the pattern found for age 43 commitment to social projects in that mean levels of identity, intimacy, and generativity themes were significantly different for the four groups, F(2, 75) = 37.92, p < .001, inline image. Group profiles within each social project were also similar to the previous findings. Career-always women expressed more identity themes than either intimacy or generativity, F(1,13) = 11.04, p = .006, inline image, and F(1,13) = 6.90, p = .02, inline image, respectively, whereas family-always women exhibited higher levels of generativity compared to both intimacy and identity, F(1,22) = 23.34, p < .001, inline image, and F(1,22) = 9.00, p < .007, inline image, respectively. Those women who added a social project to an already-existing commitment by the age of 43 exhibited the same pattern as women who had been committed to that project from early adulthood. For family-only women who added a career commitment, identity and generativity were both higher than intimacy: F(1,8) = 6.16, p = .04, inline image for identity; F(1,8) = 6.50, p = .03, inline image for generativity. For career-only women who added a family commitment, both expressed identity and generativity were significantly higher than intimacy themes: F(1,33) = 8.35, p = .007, inline image for identity; F(1,33) = 35.78, p < .001, inline image for generativity; and generativity was also significantly higher than identity, F(1,33) = 4.80., p = .04, inline image. As predicted, career-only and career-addition women resembled each other in their levels of identity, whereas family-only and family-addition women were closer on levels of generativity.

Changes in Identity, Intimacy, and Generativity Over Time

We conducted a second set of analyses that capitalized more directly on the available longitudinal data at three time points for the smaller sample of women who had responded to all three waves of data collection (N = 42). We conducted chi-square analyses to compare the distribution of social roles for these women with the larger group of women who participated at age 62. We found no differences between the larger sample and the smaller longitudinal sample in the distribution of age 43 social projects, χ2(1,2) = .268, p = .61, or in the addition or maintenance of social project between ages 28 and 43, χ2(1,3) = .098, p = .75; these results can therefore be safely generalized to the full sample.

The mean levels of expressed personality themes over time (see Table 3) showed significant differences overall, F(2, 40) = 34.10, p < .001, inline image, due to the predicted pattern of a linear increase in generativity between ages 43 and 62, F(1,41) = 4.62, p = .04, inline image. However, identity and intimacy themes showed no significant decrease over time, but rather maintained their relatively low levels in comparison to generativity. Within-age contrasts between personality themes showed significant differences within each age. At age 43, the themes differed overall, F(2, 40) = 3.88, p = .03, inline image, with generativity themes significantly higher than intimacy, F(1, 41) = 7.03, p = .01, inline image, but not identity. At age 53, personality themes were also significantly different overall, F(2, 40) = 28.95, p < .001, inline image, with intimacy themes lower than both generativity, F(1, 41) = 25.16, p < .001, inline image, and identity, F(1, 41) = 22.20, p < .001, inline image. We found the same pattern at age 62, F(2, 40) = 30.66, p < .001, inline image, that is, intimacy was lower than both generativity, F(1, 41) = 36.85, p < .001, inline image, and identity, F(1, 41) = 5.71, p = .02, inline image. However, for the first time at age 62, identity themes were also significantly lower than generativity themes, F(1, 41) = 6.56, p = .01, inline image.

Table 3. 
Identity, Intimacy, and Generativity Themes at Ages 43, 53, and 62
  1. Note. N = 42. Means (and standard deviations) are reported. Means sharing a common alphabetical subscript across time (reading across rows) are not significantly different from each other at p < .05; means sharing a common alphabetical subscript within time (reading down columns) are not significantly different from each other at p < .05.

Identity0.79 (2.01)a,b,w1.15 (1.20)d,w0.72 (1.63)e,w
Intimacy0.30 (0.88)a,x0.19 (0.60)c,x0.11 (0.62)f,x
Generativity1.15 (1.82)b,y1.42 (1.42)d,y1.91 (1.80)g,z

Changes as Related to Social Project

We also conducted longitudinal analyses for women for whom we had social project data (see Tables 1(b) and 2(b)), although social project group sizes were very small compared to those in the cross-sectional analyses. Nevertheless, relationships between earlier social project commitment and pattern of personality themes at age 62 were the same as previously found: Personality profiles were significantly different overall, F(2, 38) = 20.18, p < .001, inline image, as well as within each of the three social project groups. Similarly, the longitudinal relationship between personality and maintenance or addition of social projects produced similar results both overall, F(2, 33) = 33.62, p < .001, inline image, as well as within each social project group.


Overall, as expected, we found that women's social project commitment by age 43 related to levels of identity and generativity at age 62. Identity concerns were most frequent at age 62 for women with careers but no families. These women had very low levels of concern with intimacy, as did all social project groups, and very low levels of generativity. This lack of expressed generativity among the career-only women provides some support for Erikson's (1982) suggestion that parenthood, as a key type of caring for the next generation, is associated with generativity. A mirror image pattern characterized the women who had made family commitments with no career throughout adulthood; they were strongly focused on generativity at age 62 and less focused on identity or intimacy. In addition, identity themes were as high as generativity themes at age 62 for women who added a career to their family commitments. Moreover, identity themes were significantly higher for these women than for those who were consistently committed to family. Finally, expressed generativity at age 62 was significantly higher than intimacy for all women committed to family projects by age 43, regardless of their commitment to careers.

Looking across ages, generativity as a personality factor increased for our sample of midlife college-educated women between the ages of 43 and 62, whereas identity and intimacy factors were unchanged over the same time period. These findings support our prediction that the women's personality profiles would change, given that generativity became successively more dominant from ages 43 through 62. In fact, only at age 62 did expressions of generativity occur significantly more often than both identity and intimacy.

This picture of rising generativity across the period of women's middle age is somewhat different from that suggested in Erikson's original theory, and thus it deserves further attention. It is possible that Erikson overestimated the generativity of previous middle-aged women. Furthermore, this pattern may extend to men as well because Miner-Rubino, Winter, and Stewart (2004) found no difference between the rising trajectories of generativity for women and men between their 20s and 60s. Possibly, parenthood merely marks the beginning of a long rise in generativity that continues on for both women and men into their 60s. It is also possible that just as observers have argued that contemporary young people have much longer periods of adolescent identity experimentation than their forebears, contemporary middle-aged people may have a longer period in which to develop concern for the next generation (or to remain focused on themselves). Only further research, capitalizing on longitudinal and archival data from older cohorts, could help settle this question.

Social Project Commitment and Personality Development

Our findings are consistent with the suggestion that commitment to particular social roles influences personality development in adulthood (O'Connell, 1976; Vandewater & Stewart, 1998) as well as the notion that there is at least some connection between activities aimed at families and generativity. The results also extend Vandewater and Stewart's (1998) research beyond early middle age, while highlighting the ongoing relationship between social roles and personality development across the lifespan (Roberts et al., 2002; Roberts & Mroczek, 2008; Wood & Roberts, 2006). Generativity themes expressed in 1996 and 2005 were often similar: Women mentioned their children, as well as grandchildren, at both times. Many of the women describe an intergenerational continuum of caring that includes children, grandchildren, and parents. This point is illustrated by one woman whose expressed high points were “taking care of Mom, birth of grandchild, daughter entering significant relationship, both sons getting married.”

We note that these different life paths varied in how normative they were for this generation of Radcliffe women. For example, it is clear, given the small numbers of women in our sample who added a career after starting a family or who focused solely on a career, that following these particular paths was not normative. It is equally clear, given the larger number of women who focused solely on family or added a family after establishing a career, that these family paths were more socially acceptable options for women of this generation. Indeed, Stone (2007) concluded that this emphasis on family over career is a phenomenon particular to generations of educated women who came of age prior to the 1970s, especially elite women. Relatedly, Elliot, Dale, and Egerton (2001) found that having an occupation-specific qualification predicted the speed with which women could return to the workforce after starting a family. Women with general liberal arts educations, like those who graduated from Radcliffe, may have had fewer options for adding a career or returning to either full- or even part-time employment after starting a family.

Identity, Intimacy and Generativity Levels Over Time

Support for our fourth hypothesis, which predicted that women's generativity themes would increase between the ages of 43 and 62 with a concurrent maintenance or decline in identity and intimacy themes, indicates that concern for the next generation was, as Erikson argued, an increasing priority throughout midlife and possibly beyond. However, expressions of identity were relatively high and remained relatively stable in this period, while also remaining relatively high in comparison to intimacy, which again remained stable. This pattern over time could be the result not of aging per se, but rather of particular experiences in the life course. We were limited in what we could consider as possible experiences of this sort, but we did have data on events such as divorce, catastrophic events occurring in early middle age, and stressful life events occurring in late middle age. However, introducing controls for having experienced divorce or other difficulties during preceding periods did not produce either a significant effect or any changes in the results reported here.

Although these findings accord with previous findings regarding continued focus on identity for women in midlife, the aggregate longitudinal picture of high generativity at age 43 and increasing to age 62 masks the different ways in which Radcliffe women's personality profiles varied as a result of their different predominant social roles, especially for women who added a social role. For example, women who pursued both careers and families at age 43 achieved this outcome by different paths—either by adding a family to a career or by adding a career to a family. Further, although their levels of generativity at age 62 were similar, the two groups were very different concerning identity: The women who added a career had higher levels of identity than those who added a family. At the same time, the relative pattern among the personality themes was also different. These results suggest that adding a career to an existing family social project is associated with identity themes, and that adding a family to an existing career is associated with generativity themes. This pattern may reflect a type of recency effect: The importance of the most recently added role may prompt greater expression of particular themes at age 62. Alternatively, perhaps women with remaining unmet desires in each domain sought out these new roles. Only much more detailed analyses of individual lives will help us understand the processes through which roles and personality themes influence one another over time.

It is possible that the relative focus on generativity for the entire sample at age 62 will actually continue to increase among these women, at least for a while. The continuing increase found between ages 43 and 62 suggests that generativity includes a broader concern for successive generations (e.g., the offspring of one's offspring). It raises questions about how the social roles of parenting and grandparenting may be similar and different in terms of their influence on levels of generativity or indeed whether being without children influences generativity. Overt concerns about the next generation of students or workers (e.g., via mentoring) and about the future of the culture and the planet are sometimes expressed, but perhaps the focus in our study on one's own personal goals and high points limited the range of outlets for generativity that were more salient to the participants.

Our finding that generativity continues to increase across the period of middle age also raises further questions about when generativity is at its peak, especially for women. Stewart and Vandewater (1998) suggested that different aspects of generativity might peak at different times. For example, it is difficult to know whether the women in our study are developing new kinds of generativity themes or expressing persistent ones. Equally, part of the issue may be the relative focus on other personality themes. Perhaps the critical issue is that expressions of identity and intimacy stabilize or decline in order for generativity to dominate during middle age, and if this decline continues, whether or not generativity consequently—perhaps also ego integrity—continues to rise.

The evidence presented here both supports and extends Erikson's theory (1950, 1959/1980, 1982). We find, consistent with his theory, that by middle age expressions of intimacy themes are low, perhaps because it has been resolved. In contrast, consistent with research after Erikson, identity themes, though eventually less important than generativity themes, remain substantial at least through the early 50s for this cohort of women.

Conclusions and Limitations

Taken together, our results point to a number of tentative conclusions about lifespan development in women. Perhaps most importantly, our findings provide support for the arguments (a) that personality change takes place during the long middle ages between 43 and 62 (Helson et al., 1997, 2002; Helson & Soto, 2005; Stewart et al., 1993, 1998, 2001; Zucker et al., 2002), (b) that generativity increases during this period, and (c) that social role experiences are related to personality during it. The current findings also underscore the differing trajectories, as well as the connections among Erikson's adult developmental themes, for women in this period. In addition, they provide at least some support for Zucker et al. (2002) and Josselson (2003), who posited the existence of unbounded and inseparable developmental stages, respectively.

Participants in the current study often expressed multiple personality themes as well as multiple subthemes within personality themes. For example, this participant's response not only demonstrates the presence of more than one overall theme (i.e., identity and generativity), but that these themes represent both her occupational identity and her expressed productive generativity and generative legacy:

1. Working on my movies, being totally absorbed by the creative process. 2. Seeing my [children] launched in satisfying careers. 3. Genealogy project & trips to try to stand where our ancestors stood and see the view. 4. Creative mini-triumphs at work—designing new … databases, writing research proposals, creating a new training for supervisors, designing the first annual report … producing a concert …

Like most research, the current study has its limitations. The benefits of studying such a sample of women as the Radcliffe women, who were less constrained in their choices than women with fewer available opportunities, is offset by a diminished potential for generalizability. Additionally, the fact that these women lived through the social changes of the 1960s may have influenced not only their focus on identity development, but also their confidence in articulating that focus. It is also important to note that the relationship between social roles and personality development may not only exist in the direction in which we have emphasized here: It is quite possible that early adult personality could influence the undertaking of social roles. Unfortunately, our lack of earlier personality data, and of detailed evidence about life and personality over time, prevented us from testing this competing possibility.

Another limitation of our study is that these women are all White, middle-class college graduates, although they are representative of similar upper/middle-class samples studied in previous research (Helson et al., 1984; Stewart & Vandewater, 1993). Future research would benefit from the study of midlife personality development with a more diverse sample of women and, of course, of men. Although our study provided an unusual opportunity to examine change in women's personality development over 20 years, it is nonetheless a relatively small and homogeneous sample.

Clearly, too, we cannot generalize to all generations of women; this cohort is in some ways different from those that came before or will come after. Previous cohorts of women experienced more overt gender discrimination in such areas as education and employment; women of the baby boom were relatively well educated and were the first generation to have access to contraception as well as increased opportunities for education and occupations in their adult years. Although this cohort can be viewed as the leading edge of social changes for women (see Stewart & Torges, 2006), it is difficult to know how later cohorts of women will respond to changing gender norms.

As current and future generations age, further research will need to address how generativity develops in the “third age” (Winter, Torges, Stewart, Henderson-King, & Henderson-King, 2006), a time when retirement and bereavement issues emerge, and issues of identity and intimacy may change in relevance and form. The overall issue of how personality develops during the process of aging includes the changing relationships these personality themes have not only to one another, but also to ego integrity—Erikson's eighth and final stage. Meanwhile it is evident that for college-educated women of this generation, middle age is a time of continued growth and development.