Evaluative Organization of the Self-Concept in Younger,Midlife,and Older Adults

Abstract

Showers’s model of evaluative organization is a fruitful way to conceptualize the self-concept and is related to self-esteem, depressive symptoms, and emotion regulation. In this study, the authors developed and piloted a card-sort measure to compare the evaluative organization of the self-concept in healthy older, midlife, and younger adults (approximately two thirds of participants were women). The results support the primary hypothesis that older adults would exhibit greater compartmentalization of the self-concept than younger and middle-aged persons. Older adults may think about their positive and negative self-aspects in a more differentiated fashion (i.e., categorize positive and negative self-aspects into separate roles) than younger and midlife persons, who are more integrative (i.e., categorize positive and negative self-aspects under the same role heading) in their self-concepts. The results are consistent with cognitive priming, memory, and emotion regulation studies, which suggest that the way persons organize information about themselves may be different in younger, midlife, and older adults.

Keywords

self-concept older adults midlife compartmentalization roles

The self-concept is a rich construct that has been approached in a variety of ways to elucidate questions regarding human development and well-being. A common conceptualization is that the self-concept is a contextualized, dynamic, changeable structure (Diehl & Hay, 2007; Kling, Ryff, & Essaz, 1997); perspectives that, importantly, accommodate change and stability during human development and adult aging (Diehl & Hay, 2007). Social-cognitive perspectives on the self-concept have been particularly influential in this regard (Woolfolk, Novalany, Gara, Allen, & Polino, 1995). Woolfolk et al. (1995) elegantly described social-cognitive views of the self as being a dynamic system of interconnected constructs and as a “configuration of elements” that can vary in complexity. The current study draws from these ideas, as well as role theory (Biddle, 1986) and self-schema models (i.e., cognitive generalizations about the self; Markus, 1977). A central hypothesis, which has yet to be addressed in the literature, is that the way in which persons organize self-concept knowledge into role categtories is relevant to understanding age-group differences in emotion. For example, understanding the organization of the self-concept in older adults may provide a new persepctive from which to learn how older adults maintain high levels of psychological well-being in later life, issues that are at the heart of what some have termed the “paradox of aging” (Kunzmann, Little, & Smith, 2000).

This study draws from Showers and colleagues’ model of evaluative organization, which is an information-processing model of the self-concept that has implications for psychological well-being (Showers, 1992a, 2002; Zeigler-Hill & Showers, 2007). The model of evaluative organization posits that it is not the content but the degree to which elements of the self-concept are compartmentalized or integrated under role headings that predicts psychological outcomes. Compartmentalization or self-concept differentiation occurs when persons organize positive and negative self-aspects in different, isolated parts of themselves. For example, a compartmentalized woman may list positive traits, behaviors, feelings, and thoughts to describe herself as a grandmother, volunteer, and gardener but, in contrast, might list negative characteristics in other roles, such as wife (Table 1). Thus, this hypothetical person has both positive and negative self-aspects, but they are compartmentalized or isolated from one another because they are manifest in distinct life roles. In contrast, a person who is integrated would mix positive and negative traits, behaviors, feelings, and thoughts under the same role heading. For example, an integrated perspective on the role of mother might include the positive traits of loving and dependable but also the negative traits of impatient and anxious.

Table 1.

Illustrative Examples of Compartmentalized Versus Integrated Self-Concept Card-Sort Results

Compart-mentalized self-concept
Role	Grandmother	Volunteer	Wife	Gardener
Traits	Loving +	Helpful +	Annoyed −	Studious +
	Caring +	Friendly +	Impatient −	Knowledgeable +
	Fun +	Responsible +	Angry −	Adventurous +
	Kind +
Integrated self-concept
Role	Mother	Writer	Wife	Friend	Athlete
Traits	Loving +	Thoughtful +	Supportive +	Supportive +	Lazy −
	Impatient −	Procrastinator −	Demanding −	Trusting +	Procrastinator −
	Dependable +	Creative +	Loyal +	Anxious −	Adventurous +
	Concerned −		Giving +	Impatient −

In research with undergraduate students and some clinical populations, Showers and colleagues discovered that compartmentalized and integrative approaches to the self-concept have implications for psychological well-being (Showers, 2002). When compartmentalization occurs, it is usually positive; persons tend to think of themselves in mostly positive terms, and this positivity is isolated from negative self-knowledge (Table 1). Most of the time, compartmentalization serves to limit accessibility to negative self-knowledge and can be associated with higher self-esteem and lower depression than integration (Showers, 1992a). Positive compartmentalization is considered the baseline style of organization for most persons and is generally adaptive, because at baseline, most persons have low levels of stress and rare activation of negative self-knowledge (Showers, 2002). In contrast, when negative self-aspects are dominant in a person’s self-descriptions, an integrated self-concept is associated with better outcomes, such as fewer depressive symptoms, than a compartmentalized organization. For these persons with poorer mental health, compartmentalization can be a liability because of the inaccessibility of positive information from the dominant negative self-concept. More recent work uncovered additional “hidden” vulnerabilities of compartmentalization. For example, persons who are compartmentalized might be more vulnerable to fluctuations in self-esteem than integrated persons, perhaps because compartmentalization predisposes individuals to “higher highs” and “lower lows” (Zeigler-Hill & Showers, 2007). Furthermore, compartmentalization was associated with slower recovery from sad mood than for integrative organization (Showers & Kling, 1996).

Organization of the Self-Concept in Older Adults

To date, no studies have examined compartmentalization and integration of the self-concept in older adults. Several lines of evidence suggest that older adults might be more positively compartmentalized than younger persons. First, compartmentalization is associated with greater vulnerability for negative moods when negative self-aspects are activated (Showers, Limke, & Zeigler-Hill, 2004). Because older adults may be more vulnerable to the effects of negative mood induction than younger persons (Knight, Maines, & Robinson, 2002; Mienaltowski & Blanchard-Fields, 2005), they may have greater compartmentalization in the organization of their self-concept. Second, Showers et al. (2004) noted that compartmentalization is likely to be less cognitively effortful than integration, because it is easier to organize information into basic categories (i.e., good and bad) than to integrate across categories. Integrative organization of the self-concept may be linked to executive function capabilities (Showers & Zeigler-Hill, 2007) which decline with age (Raz & Rodrigue, 2006). Normal age-related cognitive decline might result in more simplified organization of the self-concept in older relative to younger adults. Third, in a cogntive priming study about self-reported emotion, older adults were more likely than younger adults to store positive and negative self-knowledge in a distinct fashion, which also would support greater compartmentalization in older adults (Ready, Robinson, & Weinberger, 2006).

If older adults are more compartmentalized than younger persons, on the basis of studies with younger persons, there may be implications for how older adults respond to negative events and vulnerability for depression, as well as self-esteem and emotional well-being (Showers, 1992a, 2002; Zeigler-Hill & Showers, 2007). For example, risk for depression is not increased with age, but depressive symptoms do increase after a significant stressor, such as chronic illness or injury (Baltes & Smith, 2003). The model for evaluative organization of the self-concept suggests that reactivity to negative life events might be greater in persons who are compartmentalizated versus more integrative in their self-concepts; thus, this model might be a useful way to think about vulnerability to depression in older adults after a negative life event. If older adults are more compartmentalized than younger persons, they may be at higher risk for depressive reactions following a negative life event.

Furthermore, emotional well-being is greater in young-old persons relative to younger adults (Consedine & Magai, 2006), because the ratio of positive to negative affect generally is more favorable in older than younger persons. One way to better understand these age-group differences in emotional well-being may be to understand age-group differences in the evaluative organization of the self-concept. Compartmentalization may facilitate orientation to positive moods and events relative to negative stimuli and may contribute to or reinforce “positivity” effects that often are found in older adults (Mather & Carstensen, 2005).

Questions pertaining to compartmentalization versus integration in the self-concepts of older adults are important to pursue. Whereas research has focused on other aspects of the self-concept in older adults, such as self-concept differentiation (Diehl, Hastings, & Stanton, 2001), future and ideal selves (Ryff, 1991), and change in the face of significant life challenge (Kling et al., 1997), models of evaluative organization have yet to be addressed.

Content of the Self-Concept

To date, research on the evaluative organization of the self-concept has not considered the impact of content of the self-concept on psychological outomes or, for that matter, on the organization of the self-concept itself. Showers’s evaluative model of the self-concept does not consider the content of the self-concept to be a centrally important aspect. Showers and Zeigler-Hill (2007) speculated that the content of the self-concept or the domains of a person’s life that seem important are likely to fluctuate and to be influenced by transient events. These content areas are not likely to have an association with the organization of the self-concept, which is viewed as centrally important.

In fact, there is limited research on the content relative to the organization of the self-concept in aging (Diehl & Hay, 2007). However, it could be useful to explore links between roles and the organization of the self-concept in an aging context, because roles change as compared across the life span. Role theory posits that roles are comprehensive patterns of behavior that are linked to identity and that also may change during adult development (Biddle, 1986). Midlife adults might have more life roles than younger or older adults (Lachman, 2004). Furthermore, older adults have greater opportunity for unconventional roles because of declining demands on their time for traditional work and parenting (Herzog, Markus, Franks, & Holmberg, 1998). These changes could be linked to organization; for example, midlife persons might have more roles and traits constituting their self-concepts, leading to greater complexity. Older adults might find enjoyable their freedom to inhabit less traditional roles, as well as the ability to relinquish or minimize some traditional roles (e.g., primary caregiver to children), which might contribute to overall well-being and greater positive compartmentalization in later life. Thus, we sought to consider how role content and role organization differ by age group and to explore associations between these two aspects of the self-concept.

The Current Study

In the current study, a self-concept card-sort task was developed to include traits that were relevant, common, and spontaneously reported by younger, midlife, and older adults (Part 1). The card-sort task was modeled after Showers (1992a) influential work. Next, the card-sort task was piloted in independent samples of younger, midlife, and older adults to explore how compartmentalization differed by age group (Part 2).

On the basis of Showers (2002) cognitive model of evaluative organization of the self-concept, older adults were expected to exhibit greater positive compartmentalization than younger persons. Positive compartmentalization is indicated when few negative traits are used to describe the self-concept and they are isolated from the more numerous positive aspects. Older adults were expected to think of themselves in different roles as more exclusively positive or negative rather than listing a mix of positive and negative traits; this compartmentalization was expected to occur less often in midlife and younger persons.

Given Showers and colleagues’ suggestion that integrative organization of the self-concept might be more cognitively effortful, as reviewed above, we included measures of executive functions in Part 2 to test this hypothesis. In exploratory analyses, we hypothesized that greater executive function capabilities would be associated with more integrative organization of the self-concept.

Furthermore, current emotions were assessed, because studies often find mean age-group differences in positive and negative emotions (i.e., negative emotions are endorsed less for older than younger persons; Charles, Reynolds, & Gatz, 2001), and we sought to determine if age-group differences in emotion were significantly associated with self-concept measures. On the basis of previous research, we hypothesized that compartmentalization would be associated with greater positive mood and less negative mood than integration (Showers & Zeigler-Hill, 2007).

The primary focus of this study was on the compartmentalization versus integration of the self-concept in different age groups. However, in secondary analyses, we sought to explore associations between numbers of roles, role content, and the evaluative organization of the self-concept. These issues have not been studied before. Despite Showers and Zeigler-Hill’s (2007) suggestion that role number or content are not relevant to the evaluative organization of the self-concept, in exploratory analyses, we speculated that the freedom to inhabit new roles with aging may contribute to increased positivity and compartmentalization in older adults.

Part 1 Method

Participants

Participants in the measurement development arm of the study were younger (aged 20 to 39 years; n = 20 [n = 16 or 80% women]) (Table 2), midlife (aged 40 to 64 years; n = 26 [n = 22 or 85% women]), and older (aged 65 years and older; n = 21 [n = 15 or 71% women]) adults; data from this sample were previously published (Ready, Carvalho, & Weinberger, 2008, Study 3). The older and midlife adults were recruited from the western Massachusetts area through posters, fliers, advertisements in newspapers, and talks at various community organizations. Individuals from an institutional review board (IRB)–approved pool of older adults who were interested in participating in research also were contacted for participation. The younger adults were University of Massachusetts undergraduate students recruited through an online signup for psychology experiments.

Table 2.

Descriptive Statistics for Demographic and Cognitive Factors by Age Group (Part 1)

Variable	Younger	Midlife	Older
Age (years)	19.9 (1.0)	54.8 (6.8)	74.0 (6.6)
Income^a	3.8 (1.9)	4.3 (1.0)	4.5 (0.9)
Education (years)^b	14.3 (0.8)	16.6 (3.3)	17.9 (2.9)
HVLT–R delayed recall score	10.5 (1.4)	10.0 (1.7)	10.1 (1.6)

Note: HVLT–R = Hopkins Verbal Learning Test–Revised. Data are expressed as M (SD).

Intervals for annual household income: 0 = less than $10,000, 1 = $10,001 to $20,000, 2 = $20,001 to $30,000, 3 = $30,001 to $40,000, 4 = $40,001 to $50,000, 5 = $50,001 or greater.

The value for younger adults was significantly (p < .05) lower than for middle-aged and older adults.

Inclusion criteria were good to excellent self-reported health, the ability to travel to campus for participation in the study, English speaking, and at least 9 years of formal education. Exclusion criteria included self-reports of a current or past neurologic illness, current or past major psychiatric diagnoses, current or past diagnoses of substance abuse or dependence, or cognitive impairment. The Hopkins Verbal Learning Test–Revised (Benedict, Schretlen, Groninger, & Brandt, 1998) delayed recall task was used to screen for cognitive impairment because memory tasks are particularly sensitive to age-related changes in cognition (De Jager & Budge, 2005). Data from persons scoring in the impaired range for their age on delayed recall (i.e., >2 standard deviations lower than the age-group mean) were excluded from analyses. The older and middle-aged adults were reimbursed $25, and the undergraduate students received extra credit points for participation.

A multivariate analysis of variance (MANOVA) was conducted to test for group differences in demographic and cognitive variables (income, education, and Hopkins Verbal Learning Test–Revised delayed recall) to characterize the sample and reduce the Type 1 error rate associated with multiple analyses of variance. There was a significant effect for age group, Wilks’ lambda = 0.70, F(6, 106) = 3.43, p < .01, partial η² = 0.16 (Table 2). Tests of between-subjects effects indicated a significant effect of age group on education, F(2, 55) = 8.50, p < .01, partial η² = 0.24; younger adults had significantly less education than midlife persons (p < .05, Cohen’s d = 0.96) and older adults (p < .01, Cohen’s d = 1.69). This is because the younger adult sample was composed of undergraduates pursuing their academic degrees.

Procedure

All participants were screened for inclusion and exclusion criteria over the telephone by a trained research assistant; qualifying individuals were invited to participate. In individual assessment sessions, all participants signed an IRB-approved consent form. The study was approved by the internal review board at the University of Massachusetts Amherst.

Next, participants were given several blank note cards and asked to write down different roles that characterized their self-concept. Participants were told,

Everybody thinks about themselves in different ways. I want you to write down different roles that characterize yourself on the top of these note cards. The roles should be ones that are important in how you think about yourself. Some examples might be yourself at work or at home; or yourself as a significant other or son/daughter. As I mentioned, everyone thinks of themselves in different ways, so I’d like you to take a few minutes to think and then list the roles in which you think about yourself. Please come up with at least 3 different roles of yourself and keep going until all important roles are represented. Write one aspect on the top of each note card.

All participant questions were answered. Next, participants were instructed,

Now I would like you to list traits or characteristics of yourself under each role. For example, for (pick an aspect), list what you are like in this role. You can include both positive and negative traits. Please provide several traits per aspect but there is no upper limit. Write the aspects on the note cards.

As a final task, all participants were asked to indicate whether each trait was mostly positive or mostly negative by writing a plus sign (+) or minus sign (–) beside each trait, respectively.

Results

Developing the Standardized Card-Sort Task

All traits that were spontaneously reported by younger, midlife, and older adults were tallied by frequency and valence with the goal of selecting the most common traits across the age groups for a standardized card-sort task, to be piloted in Part 2 of this study. Valence sorting meant that positive and negative traits were tallied separately; the few traits that appeared as both positive and negative were recorded separately. Overall, participants used 67 negative traits and 171 positive traits (a 1:2.56 ratio of negative to positive traits), supporting other research demonstrating that individuals without psychological disorders think of themselves in more positive than negative terms (Kendall, Howard, & Hays, 1989; Schwartz, 1986). There were significant gender differences neither in numbers of positive and negative traits (p values > .15) nor in the ratio of positive to negative traits (p = .15). Six traits were mixed because some participants rated them as positive and some as negative (i.e., concerned, organized).

The most frequent traits used by at least one person from each age group were selected for inclusion in the standardized card-sort task. For example, the most frequent positive trait was loving, which was listed 90 times by 12 older, 2 midlife, and 17 younger participants¹; next was caring, which was used 67 times by 6 older, 13 midlife, and 14 younger persons. The most common negative traits were annoyed and anxious, used 8 and 5 times, respectively, by 1 older adult each, 2 midlife persons each, and 2 younger participants. The goal was to select traits that were relevant for persons from all three age groups. A few traits were used by only two of the three age groups (i.e., listener, knowledgeable, procrastinator, lazy) but were included because they were frequently reported. In the end, 36 traits were selected: 25 were positive, 7 were negative, and 4 were mixed (see the Appendix for a comprehensive list; the potential ratio of negative to positive traits ranged from 1:4.14 to 1:2.27, depending on whether the mixed traits were used positively or negatively). More positive traits than negative traits were selected for the card-sort task, because most persons spontaneously use more positive than negative traits to describe themselves, as was evident in our study and in results from other labs (Kendall et al., 1989; Linville, 1985; Schwartz, 1986; Showers, 1992b). Furthermore, Showers (1992a) used a greater proportion of positive to negative descriptors (approximately 63% positive and 37% negative).

Part 2 Method

Participants and Procedure

Participants were 24 younger (n = 16 women [67%]), 25 midlife (n = 22 women [88%]), and 25 older (n = 13 women [52%]) adults and were drawn from the same populations as for Part 1 (Table 3); no persons participated in both studies. Inclusion and exclusion criteria and compensation were the same, except that cognition was screened over the telephone by a trained research assistant, using the Telephone Interview for Cognitive Status–Modified (TICS-m) (Welsh, Breitner, & Magruder-Habib, 1993), rather than screening for cognitive impairment during the session. Persons scoring 30 or below on the TICS-m were excluded from the study because of suspected cognitive impairment (Brandt, Spencer, & Folstein, 1988; Welsh et al., 1993). Qualifying individuals were invited to participate. In individual assessment sessions, all participants signed an IRB-approved consent form before taking part in the study. As in Part 1, a MANOVA was conducted to determine if the age groups differed for income; education; TICS-m total, positive, and negative emotions; and executive functions. The results indicated no significant age-group effect, Wilks’ lambda = 0.70, F(12, 96) = 1.54, p > .10, partial η2 = 0.16). The lack of significant age-group differences in education was surprising, because there had been significant differences in Part 1. It appears that the older adults in Part 2 were slightly less well educated than those in Part 1.

Table 3.

Descriptive Statistics for Demographic, Affect, Cognitive, and Self-Concept Factors by Age Group (Part 2)

Variable	Younger	Midlife	Older
Age (years)	19.5 (1.3)	57.8 (5.1)	73.6 (6.5)
Income^a	3.7 (2.0)	3.6 (1.9)	3.7 (1.5)
Education (years)	13.8 (2.7)	15.7 (3.4)	15.9 (3.6)
TICS-m score	39.0 (3.3)	40.7 (4.2)	38.7 (3.8)
Positive emotion	3.2 (0.8)	3.3 (0.8)	3.7 (0.7)
Negative emotion	1.7 (0.6)	1.6 (0.6)	1.5 (0.5)
Executive function composite	0.1 (0.6)	0.2 (0.7)	-0.3 (0.7)
Self-concept measures
Number of roles	5.5 (2.2)	6.4 (2.9)	6.3 (2.3)
Traits per role^b	9.6 (5.6)	16.0 (7.5)	11.3 (7.2)
Positive traits per role^c	8.4 (5.0)	14.5 (6.5)	10.1 (6.2)
Negative traits per role	1.2 (1.1)	1.5 (1.7)	1.2 (1.8)
Compartmentalization (Φ)^d	0.4 (0.3)	0.3 (0.2)	0.6 (0.3)

Note: TICS-m = Telephone Interview for Cognitive Status–Modified. Data are expressed as M (SD).

Intervals for annual household income: 0 = less than $10,000, 1 = $10,001 to $20,000, 2 = $20,001 to $30,000, 3 = $30,001 to $40,000, 4 = $40,001 to $50,000, 5 = $50,001 or greater.

The value for midlife persons was significantly (p < .05) greater than for younger adults.

The value for midlife persons was significantly (p < .05) greater than for younger and older adults.

The value for older adults was significantly (p < .05) greater than for younger and midlife adults.

Measures

The self-concept rating task

Participants were given several blank note cards and asked to write down different roles that characterized their self-concepts; directions here were identical to those in Part 1. Participants listed as many roles as necessary to fully capture their self-concepts and wrote one role at the top of each card. Next, participants were provided with the 36 traits generated in Part 1, each written on an index card. Participants were asked to place a trait under a role if it characterized them in that role. Participants could use as few or as many traits as needed to describe themselves; that is, participants did not have to use all the traits. Blank note cards were provided so that participants could use the same trait under multiple roles, if they so desired. Participants were restricted to using the adjectives provided.

Next, participants indicated whether each trait was mostly positive or mostly negative by writing a plus sign (+) or minus sign (–) beside each trait, respectively. To score the self-concept task, total positive and negative traits were tallied and divided by the number of roles that were generated, producing summary scores that were used in analyses: traits per role, negative traits per role, and positive traits per role.

The Φ coefficient was used to measure compartmentalization, following Showers et al. (1992a). The Φ coefficient is based on the χ² statistic and determines the extent to which an individual’s organization of positive and negative traits, listed for each role, deviates from an expected organization on the basis of the overall ratio of positives to negatives in that individual’s sort. Phi ranges from 0 (a perfectly integrated sort) to 1 (a perfectly compartmentalized sort) and was calculated separately for each participant. Phi is dependent neither on the number of roles nor on the ratio of positive to negative traits in the self-concept description.

For example, Person A generates four roles. Under each of three roles, four positive traits are listed, and under the fourth role, four negative traits are listed. In contrast, Person B generates four roles but, under each role, lists one negative and three positive traits. Person B is more integrated than Person A, and thus Person B would have a lower Φ score than Person A. To further illustrate, the perfectly compartmentalized example in Table 1 would have a Φ coefficient of 1; the integrated example would have a Φ coefficient equal to 0.28.

Potential correlates of the evaluative organization of the self-concept

Subtests from the Delis-Kaplan Executive Function System assessed executive functions, including the Trail Making Test (Condition 4), Verbal Fluency (Letter), and Design Fluency (Total). These tasks measure set-shifting, divided attention, retrieval, working memory, and cognitive flexibility. Raw scores were converted to z scores on the basis of results for the entire sample. The z scores for the three cognitive tasks were averaged to create a composite executive function score; higher scores indicated better performance.

Participants also rated positive and negative emotion items for the extent to which they experienced each emotion in general on the Emotion Judgment Task (1 = none, 5 = an extreme amount). In the emotion measure, there were 30 positive (e.g., confident, delighted) and 30 negative (e.g., disappointed, angry) items. The emotion measure has strong internal consistency reliability for each valence (positive and negative) among older and younger adults (Cronbach’s α ranged from .97 to .98 for younger and older adults; Ready et al., 2006).

Analyses

A MANOVA was conducted to test for age-group differences on card-sort outcomes, to control for Type 1 error rate; significant effects were followed with post hoc Tukey multiple comparisons. For role content analyses, roles were sorted by two authors (J.O.C. and A.M.A.) into one of five categories: family relationship, nonfamily relationship, occupation, hobby, and other. Roles that did not accurately match the available categories (e.g., Irish) were sorted in the “other” category. Initial agreement was strong (intraclass correlation coefficient = .88). Discrepancies were resolved via discussion with the lead author (R.E.R.). Next, role categories were tallied within age groups, and group differences were tested using χ² tests. Correlation analyses explored associations between self-concept factors, roles, mood, and executive functions.

Results

Age-Group Differences in Compartmentalization and Trait Variables

A MANOVA was conducted to determine age-group differences for compartmentalization and trait variables (Table 3). There were significant age-group effects, Wilks’ lambda = 0.62, F(8, 128) = 4.39, p < .001, partial η² = 0.22. Tests of between-subject effects indicated that age group had significant effects (p values < .01) on traits per role, F(2, 67) = 6.01, partial η² = 0.15; positive traits per role, F(2, 67) = 7.38, partial η² = 0.18); and compartmentalization, F(2, 67) = 6.83, partial η² = 0.17. As expected, post hoc Tukey multiple comparisons indicated that older adults were significantly more compartmentalized than younger (p < .05, Cohen’s d = 0.67) and middle-aged (p < .01, Cohen’s d = 1.18) persons, with medium to large effect sizes. Midlife persons reported significantly more traits per role than younger adults (p < .05, Cohen’s d = 0.97). Midlife persons also reported significantly more positive traits per role than younger (p < .01, Cohen’s d = 1.05) and older (p < .05, Cohen’s d = 0.69) adults; again, these differences were of medium to large effect sizes.

The compartmentalization findings deserved further scrutiny because there were some persons with perfectly positively compartmentalized sorts (i.e., Φ = 1); that is, they indicated no negative traits. Older adults were perfectly positive in their self-concept descriptions (n = 7 [4 men, 3 women], 28%) significantly more often than younger (n = 2 [8%]) and midlife persons (n = 2 [8%]) (χ² = 6.29, df = 2, N = 70, p < .05). When age-group averages for Φ (i.e., compartmentalization) were computed and compared for a second time, excluding persons with perfectly positive sorts, an overall analysis of variance for age-group differences remained significant (younger M = 0.4, SD = 0.2; midlife M = 0.3, SD = 0.1; older M = 0.5, SD = 0.2); F(2,56) = 4.0, p < .05, partial η² = 0.13), with older adults being significantly (p < .05, Cohen’s d = 1.26) more compartmentalized than midlife persons, and the effect size was large. Exploratory analyses determined if there were mean differences in positive emotion, negative emotion, or executive functions for persons who were perfectly compartmentalized compared with all others; t tests revealed no significant differences (p > .20).

Roles and Organization of the Self-Concept

The total number of times that a role was mentioned within each age group was tallied to determine how prevalent were roles in each age group. There was a significant difference in the number of family roles mentioned (χ² = 6.85, df = 2, N = 71, p < .05; Table 4). Middle-aged participants reported more of these roles, with younger adults reporting the fewest. Finally, there was a significant age-group difference for the other category (χ² = 7.33, df = 2, N = 71, p < .05), with older adults having the most and younger adults having the fewest other roles. Overall, women reported more family roles than men (χ² = 11.56, df = 5, N = 71, p < .05). Within age groups, there were no significant gender differences in roles.

Table 4.

Role Frequencies by Age Group (Study 2)

	Younger	Midlife	Older
Family^a	37	63	48
Nonfamily interpersonal	35	39	18
Occupational	32	29	20
Hobby/recreation	15	13	22
Othera	13	26	31

Significant (p < .05) age-group difference.

Correlations Between Self-Concept, Role, Mood, and Execution Function Variables

Correlations were calculated to determine if compartmentalization was associated with other self-concept factors, mood, and executive functions (Table 5). Compartmentalization was significantly and negatively associated with positive and negative traits per role and positively with number of other roles. It is possible that older adults exhibit more compartmentalization because of increased role flexibility in later life. Positive traits per role were positively and significantly associated with negative traits per role; persons who listed more of one type of role were more likely to list the other as well. Greater executive functions and greater negative mood were positively associated with more positive traits per role. Persons who listed more positive traits per role were significantly more likely to list more family roles and fewer other roles. Persons who reported more “other” roles were significantly less likely to report interpersonal roles (familial or nonfamilial).

Table 5.

Correlations between Self-Concept, Execution Function, and Mood (Study 2)

	2	3	4	5	6	7	8	9	10	11
1. Compartmentalization	−0.55**	−0.52**	−0.07	−0.20	−0.05	0.01	−0.06	0.13	0.11	0.25*
2. Positive Traits per Role		0.47**	0.26*	0.25*	0.06	0.31**	−0.11	−0.11	−0.19	−0.27*
3. Negative Traits per Role			0.23	0.24	−0.10	−0.10	−0.13	−0.08	−0.19	−0.04
4. Executive Function Composite				0.01	0.17	0.09	0.05	0.13	−0.12	0.10
5. Negative Emotion					−0.34**	0.03	0.06	−0.11	−0.21	−0.10
6. Positive Emotion						0.06	−0.09	0.06	0.18	−0.03
7. Family Roles							0.18	0.07	−0.16	−0.30*
8. Non-familial Interpersonal Roles								0.11	−0.11	0.34**
9. Occupational Roles									.02	−0.10
10. Hobby-Recreational Roles										0.03

Note. 11 = Other Roles.

p < .05, ** p < .01

Discussion

Showers and colleagues’ model of evaluative organization (Showers, 1992a, 1992b, 2002; Showers & Kling, 1996; Showers et al., 2004) was extended to midlife and older adults in the current study. In Part 1, using Showers work as a model, traits were generated from spontaneous self-concept descriptions provided by younger, midlife, and older adults. From these, traits were selected for a standardized card-sort task on the basis of their prevalence and cross-age applicability. Selected traits included more positive than negative attributes because people spontaneously describe themselves using more positive than negative terms (Kendall et al., 1989; Schwartz, 1986).

In Part 2, the card-sort traits were piloted to explore how compartmentalization versus integration of the self-concept manifests in older versus younger and midlife persons. Older adults were significantly more compartmentalized than younger and midlife persons. A particularly surprising finding was that 28% of older adults described their self-concepts in purely positive terms, completely excluding negative traits from their sorts. However, even when these extreme persons were excluded, older adults were more compartmentalized than midlife persons, indicating greater divisions or differentiation between their positive and negative selves.

Greater compartmentalization in the self-concept of older adults was expected on the basis of a number of lines of evidence. First, greater compartmentalization in older relative to younger persons was found in a cognitive priming study of emotion self-report, in which older adults stored positive and negative self-knowledge more distinctly than younger persons, who were more integrated (Ready et al., 2006). Second, older adults are move vulnerable to negative mood induction, which is consistent with ideas that compartmentalized persons are more vulnerable to lower lows because positive self-knowledge is isolated from negative self-knowledge and is thus less accessible when negativity predominates (Zeigler-Hill & Showers, 2007).

Reasons why older adults are more compartmentalized than younger persons are difficult to discern. Greater compartmentalization could reflect cognitive simplification of how one thinks about oneself (Labouvie-Vief, 2005), and indeed, compartmentalization was associated with reporting of fewer traits overall. It is unclear, however, if compartmentalization was associated with reduced cognitive resources (Raz & Rodrigue, 2006). Significant associations between compartmentalization scores and executive function scores were not evident, which suggests that the complexity with which participants organized self-knowledge did not map onto the capacity for complex thinking (e.g., set-shifting, divided attention, working memory, and cognitive flexibility). Alternatively, our measures of executive functions may have been insufficient to measure the cognitive flexibility required for self-concept integration.

There may be other factors driving compartmentalization in older relative to younger adults. In a study of older adult women, Kling et al. (1997) found that some life domains were lowered in importance if persons felt that they were not doing well in that domain; in this way, negative self-knowledge might be isolated to particular roles that are less important for the self-concept. Future studies could assess the importance of roles to pursue this hypothesis. Along similar lines, Atchley (1982) suggested that over time, persons take credit for good results in life and blame external events or persons for bad results; over time, continually taking credit for good outcomes could result in positive self-concepts that are resistant to change. That is, negative information might not easily be incorporated into self-views.

Furthermore, older adults might selectively attend to or remember positive information better than negative (Ready, Weinberger, & Jones, 2007). Thus, there may be self-selective biases at work on the self-concept and over a lifetime; these processes of relegating or isolating or diminishing negative attributes could lead to positively compartmentalized self-views. Finally, whereas developmental and aging perspectives on age-group differences on the self-concept are appealing and supported by some longitudinal data (Charles et al., 2001), in this cross-sectional study, cohort effects cannot be ruled out. Perhaps there are generational and historical influences that support dissociation between one’s positive and negative self-views that differentially affected our age groups.

In undergraduates, compartmentalization is associated with greater self-esteem and fewer depressive symptoms (Showers, 1992a). If the same is true for older adults, perhaps compartmentalization is one mechanism by which older adults organize their self-knowledge to promote greater well-being, or perhaps it is a consequence of greater emotional health in older age. Contrary to expectations, there were no associations between self-concept measures and positive and negative mood. Given previous research that found links between compartmentalization and well-being (Showers, 1992a), we would have expected linkages between compartmentalization and mood, which is a core component of well-being (Diener, 1994). The lack of significant effects could be due to the small sample size or to the fact that our assessment of mood was limited; a broader assessment of well-being might have been more fruitful. Another possibility is that compartmentalization can have hidden vulnerabilities for well-being and self-esteem (Zeigler-Hill & Showers, 2007), and thus the links between compartmentalization and mood might be mixed.

Future work should determine how the evaluative organization of the self-concept relates to other conceptualizations of the self-concept. For example, Diehl and colleagues have studied self-concept differentiation (Diehl et al., 2001; Diehl & Hay, 2007; Diehl, Jacobs, & Hastings, 2006). Self-differentiation refers to the extent to which self-representations are the same or different in different roles and contexts (Diehl & Hay, 2007). It would be fascinating to know if and how self-concept differentiation relates to compartmentalization. Compartmentalized persons might be vulnerable to self-concept differentiation because the positivity and negativity with which they view themselves differs by role; that is, roles are highly dissociable on the basis of valence of trait self-descriptions. In contrast, integrated persons might experience fewer differences in self-knowledge across roles and may experience lesser differentiation. Studies along these lines could be illuminating about the self-concept and mental health in later life and might reveal another “hidden vulnerability” of compartmentalization (e.g., self-concept differentiation).

As noted above, some participants produced purely positive descriptions of their self-concepts; they were perfectly compartmentalized because they indicated no negative traits. These positive descriptions were most prevalent in older adults, consistent with ideas that some older adults might adjust to the aging process, and to the associated decline in resources, with self-protective optimism (Labouvie-Vief, 2005). Nonetheless, these purely positive self-views were extreme and seemingly unrealistic because, as Showers (2002) noted, negative self-aspects are inevitable. Exploratory analyses determined if there were differences in emotion or cognition between persons who had these extreme views of their concepts and persons who reported at least some negative traits, but results were not significant.

Roles and Traits and the Self-Concept

Contrary to expectations, there were no age-group differences in the total number of roles used to characterize the self-concept across the age groups. However, midlife persons used more traits per role when describing themselves. Thus, somewhat consistent with other studies, it appears that midlife persons may be unique in the complexity of the roles they inhabit (Evandrou, Glaser, & Henz, 2002; Marks, 1998; Spitze, Logan, Joseph, & Lee, 1994). The roles of midlife can be sources of conflict and stress but also of well-being (Spitze et al., 1994) consistent with our finding that midlife persons were the most integrated in their self-views. Furthermore, overall, midlife persons were most positive in their self-concepts. Whereas midlife often is described as a stressful time with multiple and often conflicting demands (e.g., occupational, familial; Helson & Soto, 2005), at least in the current healthy sample, midlife persons viewed their self-concepts in a generally positive light. For example, midlife persons reported the most family roles, reflecting that midlife persons often are at the center of their families (e.g., “sandwiched” between the young and old; Lachman, 2004), and family roles were associated with the most positive traits per role. Family roles also were frequent, although less numerous, in older adults. Older adults may have fewer family roles than midlife persons, but they figure more prominently in their self-concepts, perhaps because of roles lost due to retirement or smaller social networks (Carsensen, 1992; Hertzog, McGuire, & Lineweaver, 1998).

As expected, older adults reported more “other” roles than younger and midlife persons, suggesting that more traditional or easily classified roles (i.e., employee, student, parent) are replaced in later life with other pursuits. We speculated that perhaps it is the greater freedom to embody self-selected roles in later life, in part, that contributes to compartmentalization, and indeed, there was a significant association between compartmentlaization and the number of other roles.

Limitations

There were several limitations to this study. First, the samples were small, were ones of convenience, and were composed disproportionately of women. Few gender differences were found, but the samples were too small to adequately test for gender effects on the self-concept. Second, as noted above, the data were cross-sectional and do not speak to development or change in the self-concept over time. Third, the card-sort traits included more positive than negative descriptors. Whereas this was a deliberate choice and was made to reflect how persons actually describe and think about themselves and was modeled on the greater ratio of positive to negative descriptors in Showers (1992a), it is possible that a card sort with a different balance of positive to negative traits would have produced different results (Woolfolk et al., 1995). Fourth, directions to the card-sort task provided only two narrow examples of a self-concept role and thus might have unintentionally biased participants in their responding. In the future, a wider range of role examples should be provided to participants. Fifth, the samples were psychologically and physically healthy, and different organizational schema would be expected for persons with poorer mental health.

Conclusions

Despite these limitations, there were intriguing findings consistent with role theory, the model of evaluative integration, and previous research. Furthermore, this study expands the scope of age-related research on the self-concept by piloting a new tool and by introducing a new construct (i.e., compartmentalization) into the discourse, one that is likely to differ in persons of different ages. Furthermore, a novel aspect of our study was a secondary focus on the content of the self-concept, because recently, there has been more research on the structure than the content of the self-concept in aging contexts (Diehl & Hay, 2007).

All indications are that the self-concept is a dynamic structure that evolves in response to context and the demands of adult development at different stages (Diehl & Hay, 2007; Helson & Soto, 2005). Many of these changes have implications for mental health, mood regulation, and well-being. The current study opens some new perspectives on the self-concept from an age-based perspective. Midlife persons, although certainly struggling with complex demands, may view themselves in positive yet integrated ways. Older adults may think more categorically and positively about their self-concept than do midlife and younger persons and may embody the least number of “traditional” life roles.

Footnotes

Appendix

The positive traits for the standardized card sort were loving, caring, helpful, supportive, fun, friendly, dependable, hard-working, happy, creative, giving, trusting, interested, kind, understanding, adventurous, thoughtful, loyal, respectful, studious, determined, listener, sharing, knowledgeable, and responsible. The negative traits were annoyed, anxious, worried, impatient, procrastinator, lazy, and bored. The mixed traits were concerned, organized, demanding, and angry.

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Atchley

R.C.

(1982). The aging self. Psychotherapy: Theory, Reserach, and Practice, 19, 388-396.

Baltes

P. B.

Smith

(2003). New frontiers in the future of aging: From successful aging of the young old to the dilemmas of the fourth age. Gerontology, 49, 123-135.

Benedict

R.H.B.

Schretlen

Groninger

Brandt

(1998). The Hopkins Verbal Learning Test–Revised: Normative data and analysis of inter-form and test-retest reliability. Clinical Neuropsychologist, 12, 43-55.

Biddle

B. J.

(1986). Recent developments in role theory. Annual Review of Sociology, 12, 67-92.

Brandt

Spencer

Folstein

(1988). The telephone interview for cognitive status. Neuropsychiatry, Neuropsychology and Behavioral Neurology, 1, 111-117.

Carsensen

L. L.

(1992). Motivation for social contact across the life span: A theory of socioemotional selectivity. In Jacobs

J. E.

(Ed.), Nebraska Symposium on Motivation: 1992, developmental perspectives on motivation (Vol. 40, pp. 209-254). Lincoln: University of Nebraska Press.

Charles

S. T.

Reynolds

C. A.

Gatz

(2001). Age-related differences and change in positive and negative affect over 23 years. Journal of Personality and Social Psychology, 80, 136-151.

Consedine

N. S.

Magai

(2006). Emotional development in adulthood: A developmental functionalist review and critique. In Hoare

(Ed.), Handbook of adult development and aging (pp. 123-148). New York: Oxford University Press.

De Jager

C. A.

Budge

M. M.

(2005). Stability and predictability of the classification of mild cognitive impairment as assessed by episodic memory test performance over time. Neurocase, 11, 72-79.

10.

Diehl

Hastings

C. T.

Stanton

J. M.

(2001). Self-concept differentiation across the adult life span. Psychology and Aging, 16, 643-654.

11.

Diehl

Hay

E. L.

(2007). Contextualized self-representations in adulthood. Journal of Personality, 75, 1255-1284.

12.

Diehl

Jacobs

Hastings

C. T.

(2006). Temporal stability and authenticity of self-representations in adulthood. Journal of Adult Development, 13, 10-22.

13.

Diener

(1994). Assessing subjective well-being: Progress and opportunities. Social Indicators Research, 31, 103-157.

14.

Evandrou

Glaser

Henz

(2002). Multiple role occupancy in midlife: Balancing work and family life in Britain. The Gerontologist, 42, 781-789.

15.

Helson

Soto

C. J.

(2005). Up and down in middle age: Monotonic and nonmonotonic changes in roles, status, and personality. Journal of Personality and Social Psychology, 89, 194-204.

16.

Hertzog

McGuire

C. L.

Lineweaver

T. T.

(1998). Aging, attributions, perceived control, and strategy use in a free recall task. Aging, Neuropsychology, and Cognition, 5, 85-106.

17.

Herzog

A. R.

Markus

H. R.

Franks

M. M.

Holmberg

(1998). Activities and well-being in older age: Effects of self-concept and educational attainment. Psychology and Aging, 13, 179-185.

18.

Kendall

Howard

Hays

(1989). Self-referent speech and psychopathology: The balance of positive and negative thinking. Cognitive Therapy and Research, 13, 583-598.

19.

Kling

K. C.

Ryff

C. D.

Essaz

M. J.

(1997). Adaptive changes in the self-concept during a life transition. Personality and Social Psychology Bulletin, 23, 981-990.

20.

Knight

B. G.

Maines

M. L.

Robinson

G. S.

(2002). The effects of sad mood on memory in older adults: A test of the mood congruence effect. Psychology and Aging, 17, 653-661.

21.

Kunzmann

Little

T. D.

Smith

(2000). Is age-related stability of subjective well-being a paradox? Cross-sectional and longitudinal evidence from the Berlin Aging Study. Psychology and Aging, 15, 511-526.

22.

Labouvie-Vief

(2005). Self-with-other representations and the organization of the self. Journal of Research in Personality, 39, 185-205.

23.

Lachman

M. E.

(2004). Development in midlife. Annual Review of Psychology, 55, 305-331.

24.

Linville

P. W.

(1985). Self-complexity and affective extremity: Don’t put all of your eggs in one cognitive basket. Social Cognition, 3, 94-120.

25.

Marks

(1998). Does it hurt to care? Caregiving, work-family conflict, and midlife well-being. Journal of Marriage and the Family, 60, 951-966.

26.

Markus

(1977). Self-schemata and processing information about the self. Journal of Personality and Social Psychology, 35, 63-78.

27.

Mather

Carstensen

L. L.

(2005). Aging and motivated cognition: The positivity effect in attention and memory. Trends in Cognitive Neuroscience, 9, 496-502.

28.

Mienaltowski

Blanchard-Fields

(2005). The differential effects of mood on age differences in the correspondence bias. Psychology and Aging, 20, 589-600.

29.

Raz

Rodrigue

K. M.

(2006). Differential aging of the brain: Patterns, cognitive correlates and modifiers. Neuroscience and Biobehavioral Reviews, 30, 730-748.

30.

Ready

R. E.

Carvalho

Weinberger

M. I.

(2008). Emotional complexity in younger, midlife, and older adults. Psychology and Aging, 23, 928-933.

31.

Ready

R. E.

Robinson

M. D.

Weinberger

(2006). Age differences in the organization of emotion knowledge: Effects involving valence and time frame. Psychology and Aging, 21, 726-736.

32.

Ready

R. E.

Weinberger

Jones

(2007). How happy have you felt lately? A diary study of emotion recall in older and younger adults. Cognition & Emotion, 21, 728-757.

33.

Ryff

C. D.

(1991). Possible selves in adulthood and old age: A tale of shifting horizons. Psychology and Aging, 6, 286-295.

34.

Schwartz

(1986). The internal dialogue: On the asymmetry between positive and negative coping thoughts. Cognitive Therapy and Research, 10, 591-605.

35.

Showers

(1992a). Compartmentalization of positive and negative self-knowledge: Keeping bad apples out of the bunch. Journal of Personality and Social Psychology, 62, 1036-1049.

36.

Showers

(1992b). Evaluatively integrative thinking about characteristics of the self. Personality and Social Psychology Bulletin, 18, 719-729.

37.

Showers

(2002). Integration and compartmentalization: A model of self-structure and self-change. In Cervone

Mischel

(Eds.), Advances in personality science (pp. 271-291). New York: Guilford.

38.

Showers

Zeigler-Hill

(2007). Compartmentalization and integration: The evaluative organization or contextualized selves. Journal of Personality, 75, 1181-1204.

39.

Showers

C. J.

Kling

K. C.

(1996). Organization of self-knowledge: Implications for recovery from sad mood. Journal of Personality and Social Psychology, 70, 578-590.

40.

Showers

C. J.

Limke

Zeigler-Hill

(2004). Self-structure and self-change: Applications to psychological treatment. Behavior Therapy, 35, 167-184.

41.

Spitze

Logan

J. R.

Joseph

Lee

(1994). Middle generation roles and the well-being of men and women. Journal of Gerontology: Social Sciences, 49, S107-S116.

42.

Welsh

K. A.

Breitner

J.C.S.

Magruder-Habib

K. M.

(1993). Detection of dementia in the elderly using telephone screening of cognitive status. Neuropsychiatry, Neuropsychology, and Behavioral Neurology, 6, 103-110.

43.

Woolfolk

R. L.

Novalany

Gara

M. A.

Allen

L. A.

Polino

(1995). Self-complexity, self-evaluation, and depression: An examination of form and content within the self-schema. Journal of Personality and Social Psychology, 68, 1108-1120.

44.

Zeigler-Hill

Showers

C. J.

(2007). Self-structure and self-esteem stability: The hidden vulnerability of compartmentalization. Personality and Social Psychology Bulletin, 33, 143-159.