Parental contexts of adolescent self-esteem: A developmental perspective

Journal of Youth and Adolescence, VoL 18, No. 1, 1989

Parental Contexts of Adolescent Self-Esteem: A Developmental Perspective Roberta S. Isberg, t Stuart T. Hauser, z A l a n M. J a c o b s o n , 3 Sally I. P o w e r s , 4 Gil N o a m , 5 B e d o n n a Weiss-Perry, 6 and D o n n a Follansbee 7 Received November 6, 1986; accepted August 3, 1988

Relationships between parental behaviors and adolescent self-esteem were analyzed in a group o f 95 early adolescents from multiple settings. The study was designed to investigate hypotheses regarding associations between obThis study was supported through a Research Training Grant No. MH 16259 (Dr. Isberg) from the NIMH, a grant from the National Institute of Child and Human Development (NICHD Grant No. 5 R01 HD18684-02), and a Research Scientist Development Award No. 5 K-02-MH-70178 (Dr. Hauser) from the NIMH. qnstructor in Psychiatry, Harvard Medical School. Received M.D. from Harvard University. Currently studying adolescent development and working with the school consultation program of the Massachusetts Mental Health Center. 2Associate Professor in Psychiatry, Harvard Medical School. Received M.D. from Yale University and Ph.D. from Harvard University (Psychology). Currently studying family contexts of adolescent development. 3Chief of Psychiatry, Joslin Diabetes Center, and Associate Professor in Psychiatry, Harvard Medical School. Received M.D. from The University of Chicago. Currently studying psychological consequences of diabetes mellitus. 4Research Associate, Henry A. Murray Research Center of Radcliffe College. Received Ed.D. from Harvard University (School of Education). Currently studying family coping processes in response to stressful events. SDirector of Evaluation Research, the Children's Unit of McLean Hospital. Received Dipl. Psych. from Freie Universitat, Berlin (Clinical Psychology). Currently studying relationships between psychopathology and development among adolescent psychiatric patients. 6Former Research Associate, Adolescent and Family Development Project, Harvard Medical School. Received Ph.D. from Ohio State University (Psychology). Current interests in assessing ego development and family systems. ~Research Associate, Adolescent and Family Development Project, Harvard Medical School, and Coordinator, Parent-Place, Judge Baker Guidance Center. Received Ph.D. from the University of Miami (Clinical Psychology). Research interests in family studies and adolescent development.

0047-2891/89/0200-0001506.00/0 @ 1989 Plenum Publishing Corporation

2

Isberg e t

aL

served parental interactions (e.g., accepting and devaluing) and adolescent self-esteem. Parents" verbal interactions with their adolescents were assessed through application o f the constraining and enabling coding system to transcribed family discussions, generated through a revealed differences procedure. Adolescent self-esteem was measured with the Coopersmith Self-Esteem Inventory. Parent interaction-self-esteem associations were examined in the pooled sample, as well as in specific sub-groups based on gender, health, and ego development (measured by the Washington University Sentence Completion Test). Boys had more numerous associations between their self-esteem and parental interactions than girls, and psychiatrically ill boys had particularly high associations. Parental interactions were f o u n d to be most strongly related to adolescent self-esteem f o r adolescents at the lowest levels o f ego development. Our findings are consistent with the view that increasing individuation in self-esteem regulation occurs during adolescent development, such that adolescents at higher levels o f ego development evaluate themselves more independently o f parental feedback than do their less mature peers.

Are parental behaviors related to self-esteem development in adolescents? Theories from psychology, sociology, and psychoanalysis emphasize the importance of the opinions of others, and particularly appraisal by significant others, in the development of self-esteem (Adler, 1927; Cooley, 1912; Coopersmith, 1967; see Cotton, 1983, for review; A. Freud, 1965; Holmbeck and Hill, 1986; James, 1980; Kohut, 1971; Mead, 1934; Rogers, 1951; Rosenberg, 1965, 1979; Sullivan, 1953; White, 1965). An extensive research literature on the influences of child-rearing practices also suggests that parental acceptance, interest, closeness, warmth, and respect are positive influences on children's and adolescents' self-esteem (Bachman, 1970; Block, 1985; Coopersmith, 1967; Gecas, 1972; Loeb et al., 1980; Maccoby and Martin, 1983; Rosenberg, 1965). Yet few studies have examined the relationship between observed parental behaviors and self-esteem during adolescent development (Litovsky and Dusek, 1985; Wylie, 1979). Researchers have generally related adolescents' self-esteem to their perceptions of their parents' attitudes and behaviors (Bachman, 1970; Litovsky and Dusek, 1985; Rosenberg, 1979). A vast interdisciplinary literature examines parental practices and children's self-esteem. Coopersmith (1967) explored the relationship between children's reports of self-esteem and their mothers' reports of parenting, in a group of 10- and 11-year-old white males. He concluded that "acceptance of the children by their parents" and "respect . . . for individual action" (p. 236) are two important antecedents of self-esteem. Rosenberg (1963) found that teenagers' recollections of extreme parental indifference during childhood were associated with low self-esteem in adolescence. In his questionnaire survey of 2200 tenth-grade boys, Bachman (1970), found that the boys' reports of parental closeness, communication, respect, and low punitiveness

Parent Contexts of Adolescent Self-Esteem

3

predicted self-esteem more strongly than any other family or individual background variable (p. 125). These studies used a retrospective approach to determine antecedents of current self-esteem, risking contamination of memories by present self-perceptions and intervening family history. Many empirical studies have been based on theoretical models of the parent's role in self-esteem development (Rosenberg, 1979; see Shrauger and Schoeneman, 1979, for review). Rogers' (1951) interpersonal theory stressed that self-image develops out of interaction with the environment: children come to respect and gain assurance in themselves when parents accept the views and values of the child. Sullivan (1947, p. 10) proposed that "the self may be said to be made up of reflected appraisals. If these were chiefly derogatory, as in the case of an unwanted child who was never loved . . . then the self-dynamism will itself be chiefly derogatory." Mead (1934) advanced a "symbolic interaction" model to explain how inferences about the way others see us constitute the primary source of information for self-image formation (Leahy and Shirk, 1985, p. 130). Rosenberg (1979) explored developmental aspects of these interactional theories by investigating differences in self-concept formation among different age groups of children and adolescents. In his study of 1988 Baltimore school children, ages 8-18, he asked children whom they trusted most to "be right" about whether they were smart, good, or good-looking, and whom they trusted to know their innermost thoughts, feelings, and wishes. For all types of knowledge about the self, younger children trusted their parents, more than themselves, to know what they were like. However, by age 15 and older, children were far more likely to attribute prime knowledge to themselves than to their parents. McGuire (1984) tested the developmental postulate that children become less dependent on parents and on others for self-image information as they mature. He found that children's spontaneously produced self-descriptions contained fewer references to parents and to significant others with increasing age, from age 7 to 17. The convergence of these findings suggests that self-definition becomes less directly tied to relationships with parents as children progress through adolescence. Psychoanalytic theory (Blos, 1962, 1979) also views adolescent individuation as the relinquishment of childhood dependence on parents as idealized omnipotent figures. The younger child's self-esteem has been regulated by a rigid internalization of parental standards, referred to, in psychoanalytic terms as, the "oedipal superego." During adolescence, these internalized parental standards become less rigid and powerful as the adolescent's "ego ideal" acquires more influence. In contrast to the more primitive superego, the ego ideal is comprised of a more flexible set of personal expectations and aspirations for the self (Blos, 1962, p. 184). These more autonomous standards for self-evaluation "render the constancy of self-esteem and of mood increasingly independent from external sources" (BIos, 1979, p. 143).

4

isberg

e t aL

The psychoanalytic view of adolescent development is, as Hill and Holmbeck (1986) point out, in larger measure than has been thought, a social-cognitive view: "Detachment is a process of modifying childhood representations" (p. 48). The parallels between social-cognitive and intrapsychic development have been made explicit in the work of Jane Loevinger. The Washington University Sentence Completion Test of ego development (Loevinger and Wessler, 1970; Loevinger et al., 1970) assays an individual's way of understanding experience, interpersonal relationships, and the self in order to determine his or her level of psychological maturation. Individuals at the earliest stages of ego development, "preconformists," are impulsive, demanding, and have exploitative or dependent interpersonal styles. At the more advanced, "conformist" stages, increasing self-awareness enables the individual to express inner states verbally, but usually in terms of clich6s, stereotypes, and moral judgments. At the most advanced, "postconformist" stages, individuals cope with inner conflict through a high degree of self-perception, and have interpersonal styles that emphasize mutuality, autonomy, empathy, and acceptance of individual differences. Loevinger's description of ego development suggests that at the lower stages, an individual's self-esteem would be regulated primarily by external events: getting needs met and getting praise. Individuals at higher levels would be able to evaluate themselves more independently of the feedback from others, and would thus have a more stable, internally regulated sense of self, similar to the "constancy of self-esteem" in older adolescents described by Blos. Because Loevinger's Sentence Completion Test tracks a process of development that influences the regulation of self-esteem, we use this instrument in our study to test the hypothesis that more mature adolescents become increasingly independent of parental feedback for the regulation of self-esteem. Most of the research on children's regulations of self-esteem is based on children's reports of their perceptions and feelings about their parents. Gecas and Schwalbe (1986) reviewed studies of perceived parental behaviors and children's self-esteem. They found moderate associations (correlations ranging from the .30s to the .40s; p. 39) between these two variables, but cautioned against generalizing this association to the relationship between actual parental behavior and children's self-esteem. Children's perceptions of parents' behaviors may be distorted by a variety of personal needs for self-image consistency and maintenance of self-esteem (Rosenberg, 1979). Shrauger and Schoeneman's (1979) review of research on "the looking glass self" points out that people's self-perceptions agree substantially with the way they perceive themselves as being viewed by others. But there is no clear indication that self-evaluations are influenced by actual feedback received from others. We have yet to discover whether parental influences on self-esteem

Parent Contexts of Adolescent Self-Esteem

5

can be o b s e r v e d in family interaction, or whether they can only be inferred from children's perceptions and representations of parent behaviors. The few studies based on observations of parent-child interaction (Bell and Bell, 1983; Block, 1985; Loeb et al., 1980) have yielded differing findings: parental warmth and acceptance have positive associations with self-esteem in certain parent-child dyads, and in some age groups, but not in others. Recent development of techniques for microanalytic coding of transcripts of family discussions (Bell and Bell, 1983; Grotevant and Cooper, 1985; Hauser et al., 1984) have made it possible to test the relationship between children's individual characteristics and specific types of family behaviors observed in the parent-child interaction. The usefulness of this method for assessing family interaction has been demonstrated, as these studies have revealed complex and meaningful relationships between observed family behaviors and individual characteristics. In addition, gender differences have been found in the reactions of sons and daughters to maternal and paternal influences (Bell and Bell, 1983; Block, 1985; Cooper and Grotevant, 1987; Grotevant and Cooper, 1985; Loeb et al., 1980). Previous analyses (Jacobson et al., 1984) have also shown that medical and psychiatric illness are linked with variations in adolescent self-esteem, and may therefore have an impact on the interaction between parental behaviors and adolescent self-esteem. In light of these studies citing the importance of individual child differences, we track psychosocial development, gender, and health status as factors that may mediate the interaction between parental behaviors and adolescent self-esteem. In the present study, parents' accepting and devaluing comments to their adolescents are observed during semistructured family discussions. Our first set of inquiries focuses on parental behaviors and adolescent self-esteem. We examine these relationships in pooled and separate samples of normal, medically ill, and psychiatrically ill adolescents. Our general hypothesis is that parental accepting would be positively correlated and parental devaluing negatively correlated, with adolescent self-esteem. As suggested by previous contributions to this literature, reviewed above, there exists the strong possibility that so general an hypothesis will not hold across diverse adolescent and family groups. Through our varied subsamples (differing in gender and health status) we explore the possibility that such associations may be specific to certain gender pairings (e.g., mother-daughter) or illness groups (e.g., psychiatrically ill adolescents). Our second set of inquiries explores how individual differences in psychosocial maturation affect individuation in self-esteem regulation. From this model we expect to find that, with increasing levels of ego development (Loevinger, 1976), adolescents would display weaker associations between their self-esteem and their parents' accepting and devaluing feedback.

6

lsberg et al.

METHODS The analyses we report below are based on data collected in the first year of a longitudinal study of high school-age adolescents and their families. The aim of the study was to trace the course of adolescent development within the context of family interaction. The teenage subjects and their families responded annually to several paper-and-pencil tests, and to semistructured individual and family interviews.

Sample Subjects in these analyses are 95 adolescents from two-parent families, drawn from three populations: psychiatrically hospitalized teenagers (N -38), diabetic teenagers (N = 22), and nonpatient high school volunteers (N = 35). The psychiatric subjects consisted of successive 12-16-year-old admissions to a private psychiatric hospital's children's unit who agreed to participate in the study. There were only rare refusals to participate, probably because the testing and interviews took place at the patient's and family's convenience, within the hospital, along with the rest of the routine psychiatric evaluation procedures. In addition, patients and families received $30 for participation in the family interviews. All patients diagnosed as having thought disorder or organic mental disorder were excluded from the sample. While traditional parent guidance is a part of the hospital's program, family therapy was not a required treatment for the psychiatric patients. The insulin-dependent diabetic adolescents included all 12-16-year-old successive admissions to an inpatient program for diabetics who agreed to participate in the study. Again, there were rare refusals, as the testing and interviews took place within the hospital, at the subjects' convenience, and subjects were paid. The program is for intensive education and reevaluation of nonacutely ill diabetics. The high school students were drawn from 230 freshman volunteers attending a suburban high school. Freshman volunteers were included to provide comparable ages to the other samples, and to allow for following of the control group through high school years, without adding yet a new factor, the variable of being in or out of high school, to subgroups of the sample. Two of the authors (S.T.H. and S.I.P.) visited approximately one half of the freshman homeroom classrooms at the start of the school year to invite the participation of interested students in a study of "how teenagers and their families change during the high school years." The students were told that they and their families would be paid for participating in the study. Two hundred thirty students returned signed parental permission slips. Out of this total pool of volunteers, 76 students from one- and two-parent families were

Parent Contexts of Adolescent Self-Esteem

7

then included in the longitudinal project. Their inclusion in the study was based on family structure (one or two parents living at home) and socioeconomic class in order to match this samples as closely as possible with the psychiatric and diabetic samples. The 95 subjects in the analyses we report in this paper represent a subset of the entire sample o f adolescents who participated in the larger longitudinal project. For the family interaction/self-esteem analyses, we selected all subjects who had two parents participating in the family interview, a n d complete data on self-esteem and ego development. To be sure that our nonhospitalized sample was reasonably representative of the distribution of ego development levels in this age group, we compared our sample to three other samples o f high school students (Block, 1984, p. 30; Browning, 1987; Sullivan et al., 1970). When we compared the distribution of preconformists, conformists, and postconformists in our sample and in each of the three other samples, chi-square analyses revealed no significant differences. When we compared the demographic characteristics of the three populations in our sample, matching was accomplished for gender and age, but not for social class (Table I). Using the Hollingshead Two-Factor Index (Hollingshead, 1957), we found a higher representation of lower and workingclass families in the psychiatric group than in the other two groups (Table I). A one-way analysis of variance (ANOVA) was performed to assess whether adolescents in different social classes displayed significant differences in selfesteem. A second one-way A N O V A assessed social class effects on the frequencies of parents' accepting and devaluing comments to their adolescents.

Table I. Demographic Characteristics of Sample

Subject group~ Gender Male Female

High school N % 15 43 20 57

SESb Upper middle and middle Lower middle Working and lower

32 3 0

Average age Group mean Standard deviation

14.6 .56

9t 9 0

Diabetic N % 12 55 10 45

12 3 7 14.0 1.2

54 14 32

Psychiatric N % 18 47 20 53

16 12 9

43 33 24

14.4 1.0

*All comparisonsamong the three health-status groups were nonsignificant except for SES (X2 = 22.7, p < .002). bSES based on Hollingshead (1957) Two-Factor Index. SES data on one psychiatric subject was missing.

8

Isberg et aL

There were no significant differences in either the mean self-esteem or parent accepting and devaluing scores for the three social class groups. It also should be noted that despite their similarity in average age, the high school sample has a narrower age range than the other two samples. Consequently, all of the subsequent analyses control for the possible confounding influences of age through partial correlations and hierarchical multiple regression techniques. Data Collection

Individual adolescent subjects and families were seen at their hospital or school. In the first session, the adolescent subjects completed the Ego Development Test (Loevinger & Wessler, 1970; Loevinger et al., 1970) and the Self-Esteem Inventory (Coopersmith, 1967). The psyciatric patients completed these self-report measures several weeks following their admission. Although we cannot be certain, it is unlikely that their ego development and self-esteem were markedly different at the time of the family session two months later. In the family session, the adolescent and two parents each met separately with the interviewers to complete a Kohlberg Moral Judgment Interview (Colby and Kohlberg, 1986). The family members were then brought together for a "revealed differences procedure" (Strodtbeck, 1958), in which differences in moral judgment responses among the three family members were revealed. The family was instructed to defend their differences and attempt to come to a family decision. The experimenter then left the room, and the family discussion was audiotaped. Transcriptions of these audiotaped discussions form the data base for the family analyses. Measures

Parents" Interactions with Adolescents Parent responses during the family interview were assessed using the Constraining and Enabling Coding System (CECS; Hauser et al., 1984, 1987). Following this system, speeches of each family member can be scored according to whether they are "enabling" (accepting, empathic, explaining, focusing, problem-solving, curious) or "constraining" (distracting, withholding, judgmental, indifferent, gratifying, devaluing) of the autonomous functioning of other family members. In the present study, we examine only those speeches that directly express parental valuing or devaluing of the adolescent's contributions to the discussion: parents' accepting, empathic, devaluing, and judgmental speeches. These particular speech categories were chosen

Parent Contexts of Adolescent Self-Esteem

9

on the basis of the theoretical and empirical work reviewed above, which led to the study's hypothesis about the effects of parental valuing and devaluing of children. Interrater agreement for source and object of speeches were 99 and 84%, respectively. Interrater reliabilities using Cohen's kappa (1960) were applied to a speech-by-speech comparison of ratings by pairs of coders. Kappa values were .79 for accepting, .74 for empathic responses, and .65 for devaluing. Since judgmental speehes occurred too infrequently to calculate the kappa statistic, percent agreement was used to assess their reliability, which was 96% (Hauser et al., 1984). Raters were blind to the adolescents' ego development level, self-esteem score, and health status. Preliminary signs for validity of the constructs coded by CECS have been favorable: both adolescents' and parents' CECS scores have shown significant correlations in the expected directions with independent measures of adolescent and parent ego development (Hauser et al., 1984). Two scores were derived for each parent on the basis of the entire transcribed interview: (I) the number of accepting + empathic speeches made by the parent to the adolescent, and (2) the number of devaluing + judgmental speeches made by the parent to the adolescent. Correlations between the two types of speeches by the same parent were nonsignificant (r = .06 for mother's accepting + empathic scores with her devaluing + judgmental scores; r = - .03 for father's). This suggests that different aspects of parental behaviors are being measured by the different scales. To permit interfamily comparisons, the total number of mother-toadolescent, or father-to-adolescent, speeches was controlled for in all the analyses (through partial correlations or through hierarchical multiple regression equations).

Self-Esteem Adolescent self-esteem is assessed with the Coopersmith Self-Esteem Inventory (SEI; Coopersmith, 1967). In addition to using the total score as a measure of global self-esteem, Coopersmith divides the items into four conceptual domains: home, school, social, and general self-esteem. Subsequent factor analyses (Kokenes, 1974, 1978) have identified multiple dimensions of self-esteem in the SEI, in part reflecting the domains defined by Coopersmith. These scales enable us to explore whether family interactions are related to adolescent self-esteem in numerous contexts, or only in those contexts, such as home and school, in which parents, or adult authority figures, are central (Gecas, 1972). In our analyses we use the global self-esteem score (total self-esteem), and the scales that tap self-esteem in the conceptually relevant home and school domains.

I0

lsberg et aL

Split-half and retest reliabilities for the SEI are favorable (Coopersmith, 1967; Robinson and Shaver, 1973; Wylie, 1979). In addition, the inventory scores have been shown, in a variety of studies (Demo, 1985; Jacobson et al., 1984; Litovsky and Dusek, 1985), to have meaningful relationships with other individual and family characteristics in this age group. The SEI presents the subject with 58 self-descriptive statements that can be checked off as "like me" or "not like me." Eight items comprise a "lie" scale, intended to identify subjects who give socially desirable responses in their self-descriptive responses. In our sample, the lie scale scores did not significantly correlate with the SEI total, or other subscale, scores, and did not discriminate between subjects at different levels of ego development. Ego Development

We evaluate psychosocial development using The Washington University Sentence Completion Test of ego development (Loevinger and Wessler, 1970; Loevinger et al., 1970). This measure samples the way in which an individual's level of ego development is expressed in written responses to a variety of incomplete sentences, which call for personally meaningful self-expression. Ego development level is then estimated through application of a conceptually based and empirically refined scoring system. Scores are derived using the recommended methods (Loevinger et al., 1970), plus the mimeographed manual provided for those male stems not included in the published manuals. Interrater reliabilities on the separate items (using intraclass correlations) range from .70 to .80. Two types of ego development scores as obtained for each subject: (1) by summing the 36 items scores to produce item sum scores, and (2) by using the ogive distribution rules (Loevinger and Wessler, 1970) to calculate stage scores. On the basis of stage scores, we divide our subjects into two groups, which were evaluated separately in later analyses: (1) preconformists (I2, delta, delta/3), and (2) conformists and postconformists (I3, I3/4, I4, I4/5).

RESULTS Health-Related and Gender Differences in Self-Esteem, Ego Development, and Parent-Adolescent Interactions Health-Related Differences

Our first analyses examined differences in self-esteem, ego development, and parent-adolescent interactions among the three groups of healthy, med-

Parent Contexts of Adolescent Self-Esteem

II

ically ill, and psychiatrically ill subjects. Group differences (Table II) were evaluated with one-way analyses of covariance (covarying for age). Post- hoc analyses, using Scheff6's method, determined which groups differ significantly from each other. Self-esteem scores differed significantly among the three health-status groups (F[2, 91] = 14.5, p < .0001, for total self-esteem; F[2, 91] = 14.8, p < .0001, on the home scale; and F[2, 91] = 9.9, p < .0001, on the school scale). Post hoe analyses revealed that psychiatric adolescents had significantly lower total self-esteem scores than the high school students, and significantly lower home and school self-esteem scores than both the high school and diabetic groups. Ego development item sum scores also differed significantly (F[2, 91] = 44.7, p < .0001), with the psychiatric group scoring significantly lower than both the high school and diabetic groups. Parent interaction scores differed significantly among the groups for three of the four speech categories. Mothers of high school students expressed significantly more accepting + empathic speeches (F[2, 89] = 4.8, p < .01) than mothers of diabetics. Fathers of high school students and psychiatric patients expressed significantly more accepting + empathic speeches (F[2, 89] = 9.0, p < .0001) than fathers of diabetics. Fathers' devaluing + judmental scores also differed significantly among the three groups (F[2, 89] = 3.1, p < .05), with the fathers of psychiatric patients expressing the highest number of devaluing + judgmental speeches.

Gender Differences Gender was not associated with any significant differences in the SEI scores. In contrast, girls scored significantly higher than boys in ego development (F[I, 91] = 5.3, p < .05, for item sum scores). There were also differences in the way parents spoke to sons and daughters. Both mothers and fathers expressed significantly more accepting + empathic speeches to daughters than to sons (F[1, 90] = 10.9, p < .001, for mothers, and F[1, 90] = 9.2, p < .005, for fathers).

Self-Esteem and Parent-Adolescent Interactions: Multiple Regression Analyses of Pooled Sample The next set of analyses examined the extent to which specific parent interactions contributed to adolescent self-esteem, after first accounting for possible influences of demographic and health-status variables. Hierarchical multiple regression analyses were carried out. Health status was entered as the first variable. Age, gender, and social class were entered together as the second variable. (Health status, gender, and social class were entered as

12

lsberg et al.

0

r~

oo.~

~.

~

,~

o

~.

~

i~

~

0o

e~

"n.

o

~.

--

--

t"

0

r'

e~-~ ~ l

~

0

.o ..o

r~

o. o.o.

e~

"e

U

Parent Contexts of Adolescent Self-Esteem

13

dummy variables.) The total number of mother-to-adolescent or father-toadolescent speeches in the transcript as entered as the third variable, to control for the frequency of speeches by the parent to the adolescent. The specific type of parent interaction was entered as the final variable. As Table III shows, of all the parent interaction categories, only fathers' devaluing + judgmental speeches contributed significantly (F = 4.0, p < .05) to lowered adolescent school self-esteem. It is interesting to note that those factors that we controlled for did contribute significantly to variation in self-esteem: health status accounted for the greatest variation in selfesteem, followed by the demographic factors of age, gender, and socioeconomic status (SES). Self-Esteem and Parent-Adolescent Interactions: Influences of Adolescent Gender and Health Status

The general multiple regression analyses above treat the group as a whole, and do not permit investigation of relationships between self-esteem and famiIy interactions that could be gender specific (e.g., present only for sons) or health status specific (e.g., present only in the psychiatric sample). By removing the variance for gender and health status, we may overlook relationships that characterize certain important subgroups. The following analyses focus on relationships between adolescent self-esteem and parent interactions in the separate gender and illness samples. Gender Partial correlations (controlling for age and total number of parentto-adolescent speeches) reveal numerous significant associations between parental interactions and boys" self-esteem, in theoretically expected directions. In contrast, there is only one significant correlation between parental interactions and girls' self-esteem (cf. Table IV). In light of the significant sex difference in ego development, a second set of partial correlation analyses also controlled for ego development (with item sum scores). Although somewhat attenuated, significant correlations were again found between boys' self-esteem and both types of maternal interactions, as well as paternal devaluing. In contrast, there were no longer any significant correlations for girls. Health Status What is the influence of medical or psychiatric illness on the relationship between adolescent self-esteem and family interactions? When partial

IsbergetaL

14

Table I11. Multiple Regression Analyses: Adolescent Self-Esteem and Parent Interactions

Health status

Age, gender, SES

R 2 AR2

B

R2

AR2

Total number mother- or father-toadolescent speeches

Type of parent interaction

B

B

R 2 ARa

R 2 z~tR2

Total self-esteem

-.40

-.40

-.40

-.40

.16

.16

.16

.16

.16/

.161

.16 /

.16 j

-.28"

-.28 ~

-.28*

-.28 ~

.25

.09 a

.25

.25

.25

.09 a

.09 d

.09 a

Mother-toadolescent speeches .08 .26 .01

Mother accepting + empathic .18 .27 .01

.01

Mother devaluing + judgmental -.13 .27 .01

.00

Father accepting + empathic .08 .26 .00

.00

Father devaluing + judgmental -.09 .26 .01

Mother-toadolescent speeches .01 .25 .00

Mother accepting + empathic .22 .27 .02

.08

.26

Father-toadolescent speeches .01 .25

.01

.25

H o m e self-esteem

-.40

-.40

-.40

-.40

.16

.16

.16

.16

.16 /

.16 j

.161

.16 /

.23 b

-.23 b

-.23 b

-.23 b

.25

.25

.25

.25

.09 a

.09 a

.09 a

.09 a

.00

Mother devaluing + judgmental -.14 .26 .02

.00

Father accepting + empathic .07 .25 .00

.00

Father devaluing + judgmental -.04 .25 .00

Mother-toadolescent speeches .02 .22 .00

Mother accepting + empathic .13 .23 .01

.01

.25

Father-toadolescent speeches - . 0 3 .25

-.03

.25

School self-esteem

-.35

.12

.12 /

-.30 c

.22

.10 e

Parent Contexts o f Adolescent Self-Esteem

15

Table 111. Continued

Health status R 2

Age, gender, SES AR z

B

R 2

Total n u m b e r mother- or father-toadolescent speeches B

AR 2

R 2

AR~

Type of parent interaction B

R 2

6J~ 2

Mother devaluing + judgmental -.35

.12

.12!

-.30 r

.22

.10"

.02

.22

.00

Father-toadolescent speeches -.35

.12

.12t

-.30 r

.22

.10"

-.03

.22

-.14

.24

.02

Father accepting + empathic .00

.25

.25

.02

Father devaluing + judgmental -.35

.12

.12t

-.30 =

.22

.10"

-.03

.22

.00

-.21

.26

.03*

*For readability, we record in the table only the beta for SES, since in all cases it was the highest beta of the three betas for age, gender, and SES. The betas for age and gender were - . 1 4 and - .05. bBetas for age and gender were - . 0 7 and - . 1 8 . CBetas for age and gender were - . 11 and - . 0 9 . "p < .o5.

"p < .Ol. /p < .OOl.

Table IV. Partial Correlations Between Parent Behaviors and Adolescent Self-Esteem Scores in Boys and Girls, Controlling for Age Mother-toadolescent accepting + empathic

Mother-toadolescent devaluing + judgmental

Father-toadolescent accepting + empathic

Father-toadolescent devaluing + judgmental

Boys Total self-esteem H o m e self-esteem School self-esteem Girls Total self-esteem H o m e self-esteem School self-esteem *D < .05. ~D < .01.

~p <

.005.

.26 ~ .12 .32*

- . 19 -.40 c -.27 ~

.34 b .22 .22

- .24 -.30* -.52 r

.15 .19 .19

-. l 1 - .04 -.10

.01 .06 .28 ~

- .07 .00 -.14

16

lsberg et aL

Table V. Partial Correlations Between Parent Behaviors and Adolescent Self-Esteem Scores in High Scho61, Diabetic, and Psychiatric Boys and Girls, Controlling for Age Mother-toadolescent accepting + empathic

Mother-toadolescent devaluing + judgmental

Father-toadolescent accepting + empathic

Father-toadolescent devaluing + j udgme nt a l

High school boys Total self-esteem Home self-esteem School self-esteem

-.03 .17 .09

-.38 -.62 b - . 14

.07 .38 .06

-.12 -.09 - .29

High school girls Total self-esteem Home self-esteem School self-esteem

-.14 - . 17 - . 19

-.24 - . 19 - . 19

.20 .28 .31

-.27 .06 - .35

Diabetic boys Total self-esteem Home self-esteem School self-esteem

.61 = .19 .21

.29 -.51 - . 14

.49 -.02 .35

Diabetic girls Total self-esteem Home self-esteem School self-esteem

.30 .39 .15

-.45 - . 17 - .54

.45 .43 .47

-.22 .22 - .26

Psychiatric boys Total self-esteem Hom e self-esteem School self-esteem

.58 b .03 .66 c

-.30 -.31 -.61 b

.68 ~ .21 .39

-.22 -.48= -.47*

- .40 = -.34 .26

.10 -.14 .10

Psychiatric girls Total self-esteem H o m e self-esteem School self-esteem

- ,25 -.07 .00

- .04 .00 - . 16

-.39 -.35 - .62*

"p < ,05. bp < .01. =p < .005.

correlations (controlling for age and total number of parent-to-adolescent speeches) were calculated separately for the three health-related groups, there were no consistent differences in the patterns of associations. However, when each group was further subdivided by gender, psychiatrically ill boys had strongly positive correlations between parents' accepting speeches and total and school self-esteem (cf. Table V). The psychiatrically ill boys also had significant negative correlations between parents' devaluing speeches and school and home self-esteem. All but one of these correlations (between mothers' accepting + empathic speeches and total self-esteem) remained significant after controlling for the effects of ego development. Such high correlations were not found in the high school or diabetic groups.

Parent Contexts of Adolescent Self-Esteem

17

Self-Esteem and Parent-Adolescent Interactions: Influences of Ego Development How does adolescent ego development influence the association between parent behaviors and self-esteem? Multiple regression analyses were performed on the pooled sample (N = 95) to assess the interaction o f ego development and parent behaviors in predicting self-esteem. In these hierarchical regression analyses, health status, SES, age, and gender were entered first, so that the effects of ego development and parent behaviors could be evaluated independently of these factors. Ego development (coded as preconformist or conformist + postconformist), total parent-to-adolescent speeches, and parent CECS score were then entered. The final term measured the interaction between ego development and parent behaviors as predictors of self-esteem. A significant increment in school self-esteem variance was accounted for by the interaction term between ego development and motherto-adolescent devaluing + judgmental scores (increment in R 2 = .05, F[8, 84] = 6.57, p < .05). This finding replicated a pattern noted in our preliminary correlation analyses (available from first author): parental behavior was significantly correlated with adolescent self-esteem in the preconformist, but not the conformist + postconformist, group. A more stringent test of this association between parent behaviors and adolescent self-esteem in the preconformist group was then carried out. Hierarchical multiple regression analyses of the preconformist group examined the proportion of variance in self-esteem scores explained by parent responses, after controlling for the contributions of health status, social class, age, and gender, and total parent-to-adolescent speeches. Mothers' devaluing + judgmental behaviors explained a significant amount of variance in the total self-esteem scores [increment in R 2 = .07, F(6, 44) = 3.96, p -.05], as well as in school self-esteem [increment in R 2 = . 10, F(6, 44) = 6.47, p = .01] o f the preconformist group. Fathers' devaluing + judgmental speeches explained a significant amount of variance in the school subscale [increment in R 2 = .07, F(6, 44) = 4.50, p < .05]. Similar multiple regression analyses of the conformist + postconformist group revealed no significant contributions by parent speeches to variance in adolescent self-esteem.

DISCUSSION The present study brings a new developmental perspective to the interaction between parent behaviors and children's self-esteem: level ofpsychosocial development, rather than chronologic age, is our measure of maturational differences in self-esteem regulation. Previous studies have shown that ego

18

lsberg etal.

development distinguishes among different family interaction styles (Hauser et al., 1984), different levels of self-esteem (Jacobson et al., 1984) and different adaptative and defensive styles (Beardslee et al., 1985) in early adolescents. We again chose ego development, rather than age, as our yardstick, because the adolescent's level of intrapsychic and interpersonal maturity, measured by the ego development test, has been shown to have an important influence on the subtle aspects of intrafamilial process that we evaluate. Our analyses reveal that subjects at lower stages of ego development have significant correlations between self-esteem and parents' devaluing interactions with them while subjects at higher stages of ego development do not (when age, health status, and other possible confounds are controlled). The more mature adolescent appears to evaluate him- or herself more independently of parents' evaluative comments than does the adolescent at a lower stage of ego development. Our findings are consistent with the contemporary view of adolescent development as a "process of individuation" (Hill and Holmbeck, 1986). This concept, supported by other empirical studies of the individuation process (Steinberg and Silverberg, 1986; Youniss and Smollar, 1985), recalls the theoretical and clinical observations of Peter Blos (1962, 1979), who described adolescent individuation as the relinquishment of childhood conceptualizations of parents as idealized omnipotent figures. Blos described how self-esteem regulation gradually gains autonomy from childhood conceptualizations of parental standards. In the present study, we find evidence for increasing autonomy from actual, rather than internalized, parental influences. Our findings are also consistent with other clinical observations. Blos notes (1979, pp. 146-147) that those adolescents who make the most violent rupture with parental standards actually "sidestep" the difficult process of disengagement from idealized parental figures. Their "wholesale rejection" of their family and past is a sign of how difficult it is for them to negotiate their separation from parents. In our sample, the psychiatric boys, who were, in most cases, hospitalized for serious conduct disorder and extreme family conflict, were having the most difficult struggles with their families and schools. Interestingly, the subjects in this group had the highest correlations between self-esteem and parental responses, even when the effect of their lower ego development is accounted for. Although they were the most oppositional to parents in their overt actions, these subjects appear to have the strongest associations between parental feedback and self-image regulation. Our findings recall Loevinger's (1976) conceptualization of the stages of ego development as a process of gaining increasing self-direction. Her description of the postconformist- as having autonomous and inner-directed ways of perceiving and integrating experience- suggests that adolescents at this stage of development could evaluate themselves more independently of

Parent Contexts of Adolescent Self-Esteem

19

parental feedback than adolescents at lower stages of ego development. Adolescents at the conformist level, although not wholly autonomous in their ways of thinking, may be able to use their conformity to group standards as an alternative to the values of their parents. The preconformist subjects, who are most concerned with having their needs met and who interpret experience in the most concrete way, appear more dependent on parental feedback for self-image formation. In addition to this central motif of important mediating effects of psychosocial development on relationships between family processes and adolescent self-esteem, there were indications of a gender difference. Boys, especially those in the psychiatric group, had significant associations between their selfesteem and parent behaviors, in contrast to minimal associations for the girls. This was an unexpected finding. The most plausible interpretation is that the boys' self-esteem, and perhaps overall self-perceptions, remain more tied to family influences in the early adolescent years. This sex differences is not simply due to the boys' lower levels of ego development, since the contrasting correlations patterns did not disappear after we controlled for ego development. A competing directional interpretation is that the parents are responding differentially to their sons and daughters, being more accepting of sons with higher esteem, and more devaluing or judgmental toward sons with impaired self-esteem. Such consistent reactions to daughters are not occurring. We consider this explanation less likely, but it cannot be ruled out through cross-sectional data. Unlike Crain et al. (1966), we did not find higher correlations between maternal behaviors and self-esteem in the diabetic group. Rather, we found the greatest number of significant correlations among the psychiatric groups of boys and girls. With the small number of subjects in each group, it is difficult to draw conclusions from these findings. However, the results are consistent with our central hypothesis that those adolescents who are having the greatest difficulty with the normal individuation process are the most dependent on parental feedback for self-esteem. As noted above, it is important to recognize that this cross-sectional study cannot rule out the possibility that such findings reflect the influence of adolescents on their parents. Developmentally arrested and psychiatrically ill teenagers may elicit more direct parental responses to their expressions of self-esteem than normally progressing children. Analyses o f our longitudinal data will help identify which factors from early adolescence predict outcome in later years, shedding some light on the question of causal direction. There are several other method considerations to bear in mind. Children's perceptions of parent behaviors are more likely to coincide with selfevaluation than are a neutral observer's perceptions of these same bahaviors. Our observations o f parents do not have as strong a relationship with children's self-esteem as do children's own perceptions of parental behavior (Ge-

20

isberg et aL

cas and Schwalbe, 1986; Holmbeck and Hill, 1986). Additional factors, enumerated below, may attenuate the strength of the correlations between observed parent behaviors and self-esteem. The subjects' experimental t a s k - the resolution of disagreements over moral dilemmas-may not call for behaviors representative of family life. In addition, the adolescent's experience of being accepted or devalued in such an interaction may not generalize to other aspects of adolescent life. Another factor that may attenuate the findings is our scoring system. The microanalytic scores used to assess parent accepting and devaluing may not reflect family characteristics that are the most influential determinants of children's self-esteem. Accepting and devaluing may be better assessed by global ratings of parental attitudes that permeate the p a r e n t - adolescent interaction, rather than by frequencies of particular types of comments (Bakeman and Brown, 1980). In addition, other factors in the adolescents' lives may overshadow parental contributions to self-esteem. Teenagers' peer relationships, their physical and intellectual endowment, their school achievement, and their expanding contacts with the world outside the family may become increasingly more important than parental feedbback in self-esteem regulation. Parent behaviors probably have the greatest impact on self-esteem development during early childhood. By evaluating families whose children have reached adolescence, we may have missed the opportunity to observe those parent behaviors that have the greatest impact on self-esteem. When the entire heterogeneous sample is evaluated as a group, it appears that parental behaviors have little impact on self-esteem. Yet when the sample is divided into more homogeneous groups on the basis of gender, health status, and/or ego development level, parental behaviors appear to have significant influences on self-esteem within certain specific groups. The analyses from the first year of our longitudinal study suggest a developmental trend of decreasing association between parental behaviors and self-image regulation with adolescent psychosocial maturation. Using our longitudinal data from several years of adolescence, we have now distinguished several different trajectories of ego development (e.g., progression, arrest, consistent conformist [Hauser et al., 1988]). Our ongoing analyses of these groups will further clarify how individual differences in development mediate the interplay between parental behaviors and self-esteem regulation.

ACKNOWLEDGMENTS Dr. John Houlihan provided critical statistical consultation, and Richard Resulis contributed valuably to data management and computer program-

Parent Contexts of Adolescent Self-Esteem

21

ming. Barbara Turner, Eydie Kassendorf, Dena Grossier, and Stephanie B u k e m a assisted in d a t a c o l l e c t i o n . W e t h a n k , as well, t h e f a m i l i e s w h o s e g e n e r o u s p a r t i c i p a t i o n m a d e this s t u d y p o s s i b l e .

REFERENCES

Adler, A. (1927). The Practice and Theory o f Individual Psychology. Harcourt, Brace, Jovanovich, New York. Bachman, J. B. (1970). Youth in Transition (Vol. 2) The Impact o f Family Backgroun and Intelligence on Tenth Grade Boys. Survey Research Center, Institute for Social Research, Ann Arbor, MI. Bakeman, R., and Brown, J. V. (1980). Early interaction: Consequences for Social and mental development at three years. Child Develop. 51: 437-447. Beardslee, W., Jacobson, A., Hauser, S., Noam, G., and Powers, S. (1985). An approach to evaluating adolescent adaptive processes. J. Am. Acad. Child Psychiatr. 24: 637-642. Bell, D. C., and Bell, L. G. (1983). Parental validation and support in the development of adolescent daughters. In Grotevant, H. D., and Cooper, C. R. (eds.), Adolescent Development in the Family. Jossey-Bass, San Francisco. Block, J. (1985, October). Some relationships regarding the self emanating from the Block and Block longitudinal study. Paper presented at the Social Science Research Councel conference on Selfhood, Stanford, CA. Block, J. H. (1984). Sex Role Identity and Ego Development. Jossey-Bass, San Francisco. Blos, P. (1962). On Adolescence. The Free Press, New York. BIos, P. (1979). The Adolescent Passage. International Universities Press, New York. Browning, D. L. (1987). Ego development, authoritarianism and social status: An investigation of the incremental validity of Loevinger's Sentence Completion Test (Short Form). J. Personal. Social Psychol. 53: i 13- I 18. Cohen, J. (1960). A coefficient of agreement for nominal scales. Educ. Psychol. Measure. 20: 37-46. Colby, A., and Kohlberg, L. (1986). The Measurement o f Moral Judgment. (Vol. 1). Cambridge University Press, New York. Cooley, C. H. (1912). Human Nature and the Social Order. Scribner's, New York. Cooper, C. R., and Grotevant, H. D. (1987). Gender issues in the interface of family experience and adolescents' friendship and dating identity. J. Youth Adoles. 16: 257-264. Coopersmith, S. (1967). The Antecedents o f Self-Esteem. Freeman, San Francisco. Cotton, N. S. (1983). The development of self-esteem and self-esteem regulation. In Mack, J. E., and Ablon, S. L. (eds.), The Development and Sustenance o f Self-Esteem in Childhood, International Universities Press, New York. Crain, A. J., Susman, M. B., and Weft, W. B., Jr. (1966). Family interaction, diabetes and sibling relationships. Int. J. Social Psychiatr. 12: 35-43. Demo, D. H. (1985). The measurement of self-esteem: Refining our methods. J. Personal. Social Psychol. 48: 1490-1502. Freud, A. (1965). Normality and Pathology in Childhood. International Universities Press, New York. Gecas, V. (1972). Parental behavior and contextual variations in adolescent self-esteem. Sociometry 35: 332-345. Gecas, V., and Schwalbe, M. L. (1986). Parental behavior and adolescent self-esteem. J. Marriage Family 48: 37-46. Grotevant, H. D., and Cooper, C. R. (1985). Patterns of interaction in family reltionships and the development of identity exploration in adolescence. Child Develop. 56: 415-428. Hauser, S. T., Powers, S. I., Bowlds, M., and Noam, G. G. (1988). Family interiors of adolescent ego development. Family Perspect. 21: 263-282.

22

Isberg et al.

Hauser, S. T., Powers, S. I., Noam, G. G., and Jacobson, A. M. (1984). Familial contexts of adolescent ego development. Child Develop. 55: 195-213. Hauser, S. T., Powers, S. I., Weiss-Perry, B., Follansbee, D., Rajapark, D. C., and Greene, W. (1987). Constraining and enabling family coding system. Unpublished manuscript. Harvard Medical School, Boston. Hill, J. P., and Holmbeck, G. N. (1986). Attachment and autonomy during adolescence. In Whitehurst, G.A., (ed.), Annals of Child Development (Vol. 3, pp. 145-189). JAI Press, Inc., Greenwich, Connecticut. Hollingshead, A. B. (1957). Two-factor index of social position. Mimeograph, Yale University, New Haven, CN. Holmbeck, G. N., and Hill, J. P. (1986). A path-analytic approach to the relations between parental traits and acceptance and adolescent adjustment. Sex Roles 14: 315-334. Jacobson, A. M., Hauser, S. T., Powers, S., and Noam, G. (1984). The influences of chronic illness and ego development on self-esteem in diabetic and psychiatric adolescent patients. J. Youth Adoles. 13: 489-507. James, W. (1980). The Principles of Psychology. Holt, New York. Kohut, H. 0971). The Analysis of the Self. International Universities Press, New York. Kokenes, G. (1974). Grade level differences in factors of self-esteem. Develop. Psychol. 10: 954-958. Kokenes, B. (1978). A factor analytic study fo the Coopesmith Self-Esteem Inventory. Adolescence. 13: 149-155. Leahy, R. L., and Shirk, S. R. (1985). Social cognition and development of the self. In Leahy, R. L. The Development of the Self. Academic Press, Orlando, FL. Litovsky, V. G., and Dusek, J. B. (1985). Perceptions of child rearing and self-concept development during the early adolescent years. J. Youth Adoles. 14: 373-387. Loeb, R. L., Horst, L., and Horton, P. J. 0980). Family interaction patterns associated with self-esteem in preadolescent girls and boys. Merrill-Palmer Quart. 26: 205-217. Loevinger, J. 0976). Ego Development: Conceptions and Theories. Jossey-Bass, San Francisco. Loevinger, J., and Wessler, R. (1970). Measuring Ego Development (Vol. l). Jossey-Bass, San Francisco. Loevinger, J., Wessler, R., and Redmore, C. (1970). Measuring Ego Development (Vol. 2). Jossey-Bass, San Francisco. Maccoby, E. E., and Martin, J. A. (1983). Socialization in the context of the family: Parentchild interaction. In Mussen, P. H. (ed.), Handbook of Child Psychology (Vol. 4). Wiley & Sons, New York. McGuire, W. J. (1984). Search for the self: Going beyond self-esteem and the reactive self. In Zucker, R. A., Aronoff, J., and Rabin, A. I. (eds.), Personality and the Prediction of Behavior. Academic Press, Orlando, FL. Mead, G. H. (1934). Mind, Self, and Society. University of Chicago Press, Chicago. Robinson, J. P., and Shaver, P. R. (1973). Measures of PsychologicalAttitudes. Institute for Social Research, Ann Arbor, MI. Rogers, C. R. (1951). Client-Centered Therapy: Its Current Practice, Implications, and Theory. Houghton Mifflin, Boston. Rosenberg, M. (1963). Parental interest and children's self-conceptions. Sociometry, 26: 35-49. Rosenberg, M. (1965). Society and the Adolescent Self-Image. Princeton University Press, Princeton, NJ. Rosenberg, M. (1979). Conceiving the Self. Basic Books, New York. Shrauger, J. S., and Schoeneman, T. J. (1979). Symbolic interactionist view of self-concept: Through the looking glass darkly. Psychol. Bull. 86: 549-573. Steinberg, L., and Silverberg, S. (1986). The vicissitudes of autonomy in early adolescence. Child Develop. 57: 841-851. Stierlin, H. 0974). Separating Parents and Adolescents. Quadrangle, New York. Strodtbeck, F. L. (I958). Husband-wife interaction and revealed differences. Am. Sociol. Rev. 16: 468-473. Sullivan, E. V., McCullough, G., and Stager, M. (1970). A developmental study of the relationship between conceptual, ego, and moral development. Child Develop. 41 : 399-41 I.

Parent Contexts of Adolescent Self-Esteem

23

Sullivan, H. S. (1947). Conceptions of Modern Psychiatry: The First William Alanson White Memorial Lectures. The William Alanson White Psychiatric Foundation, Washington, DC. Sullivan, H. S. (1953). The Interpersonal Theory of Psychiatry. Norton, New York. White, R. W. (1965). Ego and Reality in Psychoanalytic Theory. Psychological Issues (Monograph 11). International Universities Press, New York. Wylie, R. C. (1979). The Self-Concept. (Vol. 2). Theory and Research on Selected Topics. University of Nebraska Press, Lincoln. Youniss, J., and Smollar, J. (1985). Adolescent Relations with Mothers, Fathers and Friends. University of Chicago Press, Chicago.