Origins of Adolescents' Ideal Self: An Intergenerational Perspective

Journal of Personality and Social Psychology Copyright 2007 by the American Psychological Association
2007, Vol. 92, No. 3, 557–574 0022-3514/07/$12.00 DOI: 10.1037/0022-3514.92.3.557
Origins of Adolescents’ Ideal Self: An Intergenerational Perspective

Marcel Zentner and Olivier Renaud
University of Geneva
The present longitudinal study traced the ideal self of 120 adolescents to the ideals that parents hold for
themselves and for their children. Ideals were assessed using Q sorts for personality attributes and life
goals. After permutation analysis was used to control for random similarity, moderate parent– child
ideal-self similarity was evident. Three intermediate transmission steps accounted for this intergenera-
tional similarity: (a) transfer of parents’ ideal self to ideals for their children, (b) children’s perception,
and (c) acceptance of these parental ideals. The last 2 processes related to parenting practices, with
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
parental warmth augmenting parent– child concordance and restrictiveness reducing it. Compared with
This document is copyrighted by the American Psychological Association or one of its allied publishers.
boys’ ideal self, that of girls was more similar to the ideals that parents hold for their children and for
themselves. Analyses of patterns over time suggested a growth of ideal-self stability across adolescence.
Keywords: ideal self, internalization, intergenerational transmission, parenting practices, Q sort
Whereas the self can be defined as a person’s knowledge about research, we propose to address these questions by adopting an
him- or herself, comprising the attributes that the person thinks he intergenerational approach. More specifically, we analyzed an
or she actually possesses, the ideal self is a representation of the intergenerational arc that leads from the parents’ ideal self to the
attributes the person would like to have—an image of the person ideals that parents hold for their children to the ideals that children
to be ideally. The ideal self is recognized as important for at least believe their parents hold for them to the children’s own ideal self.
two reasons. First, the ideal self is a motivator that functions as an
incentive for future behavior, a self “to be approached or avoided”
Ideal-Self Transmission Processes
(Cross & Markus, 1991, p. 232). Second, the ideal self is an
evaluator of the actual self. This view can be traced to James The transmission of values, ideals, or standards across genera-
(1890), who proposed that self-evaluations depend on the degree to tions has often been described in terms of internalization. The
which the self’s actual successes coincide with the self’s aspira- concept of internalization is generally used to refer to a process
tions. More recent views have also stressed the importance of whereby parental and societal standards are gradually integrated
knowing specific self-values and their weightings to understand into the child’s self-system. Internalization involves the transfor-
self-esteem (e.g., J. Block & Robins, 1993; Harter, 1999). In mation of a subject’s “real or imagined characteristics of his
addition, building on the ideas of Horney (1950) and Rogers environment into inner regulations and characteristics” (Schafer,
(1961), numerous studies have provided compelling evidence for 1968, p. 188; see also Grolnick, Deci, & Ryan, 1997). Internal-
the role of discrepancies between ideal and actual views of the self ization concepts have been propounded in theories such as sym-
in accounting for emotional distress, in particular, depressive dis- bolic interactionism (Cooley, 1902; Mead, 1934), psychoanalysis
orders (e.g., Moretti & Higgins, 1999; Ogilvie & Clark, 1992; (Freud, 1923/1961; Sandler & Rosenblatt, 1962), and social learn-
Tangney, Niedenthal, Covert, & Hill-Barlow, 1998). Considering ing (Aronfreed, 1968; Bandura, 1971). However, as noted by
the attention that has been given to the ideal self as an ingredient Cashmore and Goodnow (1985), these theoretical propositions
of self-esteem, emotional distress, and the self-system, surprisingly have often lacked empirical legitimation. More recently, there has
little is known about its developmental origins. Do children differ been a reexamination of conceptualizations of internalization,
in their ideal self? If so, how do their unique ideals for the self which have led to new, empirically more accessible formulations.
come about? Why does John aspire to be primarily successful with One influential theory suggests that intergenerational similarity
girls, whereas Paul aspires to be successful in his studies? Why results from a two-step process of internalization:
does Mary aspire to be assertive and independent, whereas Ellen
places much more importance on being altruistic? In the present Internalization, it is suggested, involves both the child’s perception of
the parents’ position (a perception that may be accurate or inaccurate)
and the child’s acceptance or rejection of what is perceived to be the
parent’s viewpoint. Failure to internalize may then come about either
Marcel Zentner and Olivier Renaud, Department of Psychology, Uni- through inaccurate perception or through rejection. (Grusec & Good-
versity of Geneva, Geneva, Switzerland. now, 1994, p. 4)
This research was supported by a grant from the Jacobs Foundation. We
thank Sylvette Delaloye, Emmanuelle Dupont, Florence Dokic, Sabrina
A number of studies have provided support for this model, indi-
Cappuccio, and Laure Bezençon for assisting in these studies. Our thanks
extend to Tatjana Aue for insightful comments on versions of this article.
cating that perceptual accuracy and acceptance account for inter-
Correspondence concerning this article should be addressed to Marcel generational similarity in attitudes (Cashmore & Goodnow, 1985)
Zentner, Department of Psychology (FAPSE), University of Geneva, 40, and parental values (Knafo & Schwartz, in press).
Bd. du Pont d’Arve, CH-1205, Geneva, Switzerland. E-mail: Although no research to date has examined intergenerational
marcel.zentner@pse.unige.ch transmission of the ideal self, evidence suggests that it may operate
557
558 ZENTNER AND RENAUD
in ways that are consistent with the foregoing proposition. For cording to this model, the similarity between the ideals that
example, research conducted within self-discrepancy theory (Hig- parents want their child to endorse and the child’s ideal self
gins, 1987) has provided evidence for similarity between adoles- should be accounted for by (a) accurate perceptions of the
cents’ ideal self and their perceptions of parental ideals. Adoles- parental ideal and (b) acceptance of the perceived parental
cents were chosen as target population because stable ideal-self ideal. Grusec and Goodnow (1994) referred to this similarity as
representations are not believed to emerge before adolescence. The internalization. However, the mere similarity between parental
same holds for the ability to consider multiple perspectives on the and children’s viewpoints does not automatically imply inter-
self, including the ideals that significant others hold for oneself nalization in the traditional sense of a progressive intake of a
(e.g., Keating, 2004; Moretti & Higgins, 1999). Adolescents were parental position by the child. Thus, we prefer to use the more
asked to spontaneously generate sets of up to 10 attributes that neutral term assimilation to denote the extent to which a child’s
described their ideal self and the ideals the adolescents believed ideals are similar to the ideals that parents want their child to
their parents held for them. The results showed an appreciable endorse. This part of the model is strictly analogous to the
match between children’s ideal self and the ideals that they be- two-step model (see Figure 1, right part). To account for
lieved their parents held for them. The extent of match, defined as parent– child congruence in the ideal self, we added an addi-
percentage overlap, ranged from 25% in adolescents to 44% in tional link reflecting the relationship between parents’ ideal self
young adults (e.g., Moretti & Higgins, 1999). These findings are and their child-related ideals (see Figure 1, left part). We call
consistent with the notion of acceptance as defined by Grusec and this link (parental) ideal-self transposition as a way of describ-
Goodnow (1994) and provide an initial step in the analysis of ing that the ideals that parents hold for themselves are, even-
intergenerational relationships in the ideal self. tually, transposed onto the ideals they hold for their children.
The present extension of the two-step model by transposition is
based on previous research pointing to a substantial concor-
Aims of the Present Investigation dance between parents’ own values and the values they hold for
their children (Knafo & Schwartz, in press; Whitbeck & Gecas,
The overarching purpose of this prospective, longitudinal study 1988). Our proposition, then, is that parent– child congruence in
was to provide a more extensive examination of family relations in the ideal self should be accounted for by two main component
the ideal self, one that goes beyond the perceptions of the adoles- processes: (a) ideal-self transposition and (b) ideal-self assim-
cent and also includes the parents’ actual ideals for themselves and ilation, the latter reflecting both acceptance and accuracy (see
for their children. Within this general scope, particular emphasis Figure 1).
was placed on identifying processes and factors that would account Third, we took a closer look at issues of directionality in the
for parent– child similarity in the ideal self. To this end, elabora- transmission of ideals between parents and children. There is a
tions along five lines were required. broad consensus today that processes of socialization are best
First of all, it is necessary to examine whether parent– child conceptualized in transactional terms (e.g., Bugental & Grusec,
similarity in ideal self actually exists. Even though the answer to 2006; Kuczynski, 2003). Although the above-mentioned lines of
this question may seem trivial, it is not. Many parents like their theorizing often conceptualized transmission of parental beliefs
children to be intelligent, conscientious, sociable, self-confident, and values as going from parent to child, the possibility that
and so forth, and many parents and children hold these ideals for parents may be receptive to their children’s ideals should also be
themselves. Thus, any profile-based similarities between parent considered. For example, a mother may wish her child to endorse
and child ideals may simply reflect shared societal ideals. Indeed, artistic ideals (such as becoming a performing artist), and this wish
previous work found the base-rate similarity in self-ideals to be may be taken over by the child. However, some mothers may
high, in the range of .7 (e.g., Harter, 1999; Klohnen & Mendel- develop artistic ideals for their child as a reaction to their child’s
sohn, 1998). Hence, two ideal-self components need to be distin- artistic motivations. One important parameter to consider in this
guished: a prototypical component, which reflects consensually context is the consistency of a child’s ideal self. As long as a
shared ideal-self representations in a culture, and a unique com- child’s ideal self is inconsistent, lasting child-to-parent effects are
ponent, which refers to the distinctive or unique ideals of an unlikely. However, as the ideal self matures and stabilizes, influ-
individual. If, as expected, ideals were heavily shared within a ences between parent and child ideals may go both ways. Little is
given culture, a first question to be clarified is the extent to which known about ideal-self stability during adolescence, but some
transmission of unique ideals within families occurs. Although indirect evidence comes from research on values. Although values
internalization is occasionally understood as a process relating to and ideal selves are distinct constructs, in a broad sense, they can
parent– child transmission of unique parental values (e.g., Rohan both be characterized as representations of desired hypothetical
& Zanna, 1996), in other contexts, it is seen as a “developmental end states. Values begin to exhibit moderate levels of stability only
process in which children (as well as adolescents and adults) from age 13 (Whitbeck & Gecas, 1988), and they appear to
progressively integrate societal values and proscriptions into a become increasingly consistent thereafter (Rokeach, 1973). These
coherent sense of self” (Grolnick et al., 1997, p. 136). In our view, results are consistent with Piaget’s notion that the capacity to
internalization encompasses both perspectives. However, these reason in terms of possibilities and hypotheticals does not emerge
perspectives need to be clearly distinguished both conceptually before adolescence (Inhelder & Piaget, 1958; see also Keating,
and empirically. 2004). In accordance with these considerations, we addressed
Second, if unique parent– child similarity exists, how does issues of directionality with an eye on growth in ideal-self consis-
ideal-self transmission within the family occur? We propose tency.
that this question can be successfully studied from an extension Fourth, to better understand the processes involved in trans-
of the two-step model of internalization mentioned above. Ac- mission, we sought to identify factors that affect the extent of
ORIGINS OF THE IDEAL SELF 559
Figure 1. Schematic illustration of the intergenerational links supposed to characterize ideal-self transmission
from parent to child. Accuracy and acceptance account for assimilation (similar to the two-step model of
internalization), and assimilation and transposition account for parent– child similarity in the ideal self (extension
of the two-step model). Following are capsule definitions of all five kinds of concordances in parent– child
ideals. Congruence ⫽ parent– child similarity in the ideal self. Transposition ⫽ similarity between parents’ ideal
self and their ideals for their children. Assimilation ⫽ similarity between parental ideals for their children and
their children’s ideal self. Accuracy ⫽ similarity of parental ideals for their children and the perception of these
ideals by the children. Acceptance ⫽ similarity between perceived parent ideals and the children’s ideal self.
ideal-self transmission within the family. There are reasons, parent– child value profiles (Balswick & Macrides, 1975; Ro-
both theoretical and empirical, to assume that congruence in han & Zanna). Thus, we anticipated that supportive parenting
parent– child ideals may be related to parenting practices and to would be positively related to assimilation and its components
gender. For example, according to Knafo and Schwartz (2003), accuracy and acceptance, whereas restrictive parenting would
parenting affects the accuracy with which parental values are be negatively related to them.
perceived by children in three ways: It affects the availability of A role for gender is suggested by a substantial literature
the parental message, the understandability of the message, and indicating that boys and girls tend to be socialized in different
the child’s motivation to attend to the parental message. More ways. A common theme running through this literature is that
specifically, these investigators predicted that the parenting girls are encouraged to attend to intimate others’ needs, con-
dimension of warmth should contribute to accurate perception, form to their expectations, and evaluate themselves in terms of
whereas “autocratic parenting undermines accuracy” (Knafo & acceptance by their intimates. In contrast, boys are encouraged
Schwartz, 2003, p. 598). Overall, their prediction was sup- to develop an identity that is more distinct from the desires and
ported. Parenting is also likely to play an important role in wishes of their intimates (see, e.g., J. H. Block, 1983; Cross &
acceptance for at least two reasons. First, children tend to have Madson, 1997; Pomerantz, Saxon, & Kenney, 2001, for re-
more positive views of their parents if the parents use respon- views). Consistent with this gender pattern in socialization,
sive, supportive, and affectionate parenting strategies (e.g., adolescent girls score higher on measures of perspective taking,
Rohan & Zanna, 1996). A positive parental image, in turn, that is, the tendency to adopt the point of view of others
should increase the child’s receptivity to accepting parental (Eisenberg, Carlo, Murphy, & van Court, 1995; Eisenberg,
ideals. Second, warm and authoritarian parents likely have Miller, Shell, McNalley, & Shea, 1991). From this observation,
different ways of expressing their ideals. The former may be one would expect girls to perceive parental viewpoints such as
more inclined to use induction when expressing ideals as if they values more accurately than do boys, and indeed, they do (e.g.,
were recommendations that need to be openly discussed or Acock & Bengtson, 1980; Knafo & Schwartz, 2003). In addi-
negotiated with the child. A long-standing view in socialization tion, because conformity with parental expectancies is more
theory holds that parents’ use of inductive reasoning promotes encouraged in girls than in boys, one should also find more
internalization of their norms for behavior, especially prosocial endorsement of parental ideals by girls compared with boys. In
behavior (e.g., Hoffman, 1970). Authoritarian parents, in con- addition to a main effect for child gender, effects of parent
trast, may be more likely to express their child-related ideals as gender had to be considered. For example, mothers spend about
if these were obligations. In doing so, these parents risk induc- twice as much time with their adolescent children as fathers do
ing psychological reactance in their children (Brehm & Brehm, (Montemayor & Brownlee, 1987). This should give maternal
1981). Consistent with this proposition, it has been found that ideals a greater chance than paternal ideals to be perceived
authoritarian parenting is associated with dissimilarity in accurately. A final point regarding gender differences can be
traced to early theories of internalization and identification, anchor points of standard rating scales can have nonequivalent
according to which behavior and value transmission is primarily meanings across participants. As noted by Cronbach (1955)
from parents to their same-sex rather than to opposite-sex long ago, this may lead to distorted similarity estimates. In
offspring (e.g., Kohlberg, 1966; Winch, 1962). However, be- contrast, the Q sort’s forced-rating format ensures commensu-
cause findings regarding effects of the gender composition of rability in the response profiles provided by different individ-
the parent– child dyad have been mixed and inconsistent (e.g., uals (e.g., J. Block, 1971; Ozer, 1993). In the present research,
Troll & Bengtson, 1979), we did not formulate hypotheses we used a Q sort based on a comprehensive list of adjectives to
regarding interaction effects of parent and child gender. assess ideals about personality traits and life aspirations. Ado-
Fifth, the present study introduces some methodologically lescents were selected as target population for the reasons
distinctive features in comparison with earlier work. How to mentioned earlier, that is, their capacity to form ideal-self
measure the ideal self has been a matter of controversy in prior representations and their skills in considering multiple perspec-
research. Proponents of self-discrepancy theory, for example, tives on the self. We chose to include life aspirations in addition
have strongly advocated using an idiographic procedure to to personality ideals because of the particular relevance that
they acquire during late adolescence, when consequential life
assess the ideal self. More specifically, the Selves Question-

naire was introduced as a means of capturing the ideal self and decisions are being prepared.
perceived parental ideals (Moretti & Higgins, 1999). This in-
strument elicits up to 10 participant-generated self-descriptions Method
to tap dimensions that may be uniquely important and accessi-
ble to the individual. However, some authors have criticized Participants
this approach because it remains unclear whether the partici-
pant’s choices of attributes are due to their salience and acces- Families were recruited via letters sent to families with one child
sibility or simply to limitations in the participant’s working in the age range between 15 and 19 years. The addresses were
vocabulary (Tangney et al., 1998). Open-response formats also obtained through the Department of Public Instruction of the city
restrict the range of statistical techniques that can be used for of Geneva, Switzerland. Families who were interested were asked
to return a self-addressed, prestamped postcard. Each family who
fine-grained quantification. At the opposite end of purely idio-
returned a card was then contacted by project staff to determine
graphic assessments is the conventional variable-centered ap-
eligibility for enrollment in the study. The sample at Time 1
proach, which has been used in the great majority of studies on
consisted of 120 adolescents (M ⫽ 17.5 years, range 15.0 –19.8
parent– child resemblance (see Loehlin, 2005, for a review of
years) and their parents, 115 mothers and 68 fathers (M ⫽ 50.0
this literature). For example, in recent studies on the transmis-
years, range 38 –71 years). Of the adolescents, 69% were girls
sion of values, the child’s importance rating for a certain value
and 31% boys. Over 90% of the original sample, 109 adoles-
is the dependent variable, and the parents’ score on the same
cents and their parents (105 mothers, 62 fathers), participated in
value is the predictor (e.g., Albert & Trommsdorff, 2003;
the assessments at Time 2, which took place approximately 7
Schönpflug, 2001; see also Larson & Almeida, 1999, for a
months later.
similar approach). An important limitation of this approach is
Of the adolescents in the sample, 69% were high school stu-
that it is confined to an analysis of parent– child congruence on dents, 11% were in a professional training school, and 8% were in
single dimensions computed across persons one at a time (e.g., college. Thirty percent of the adolescents came from single-parent
collectivism, extrinsic motivation, sociability). In contrast, con- homes in which the adolescent lived with his or her mother in the
gruence assessed via the person-centered approach is based on vast majority of cases, and 70% of the adolescents were living with
profile similarities that capture the congruence of the child– both parents. This distribution is relatively representative of the
parent dyad across responses on an overarching individual- entire population of Geneva (21% and 79%; population reference
differences domain (such as life goals, personality, attitudes). In values provided by OCSTAT, 2002). Although, in the biparental
other words, by using a person-centered approach, it is possible category, both parents were encouraged to participate, in 20% of
to compare a child and his or her parent in terms of a system of the cases, the father could not participate. The vast majority of
interrelated attributes rather than in terms of single variables, participants were of European descent (95%) and French mother
each considered separately. A second limitation is that a tongue (80%), which approximates the reference values of the city
variable-centered correlation or regression is computed between population (92% and 70%, respectively). To report family income,
the parents’ and the child’s scores on a specific characteristic participants were presented with a series of sequential income
across all participants and thus characterizes the sample rather ranges from which to choose, resulting in the following distribu-
than the dyad (see Luo & Klohnen, 2005, for an extensive tion: 18%, 40,000 –70,000; 33%, 70,000 –100,000; 33%, 100,000 –
analysis of the limitations of the variable-centered approach). 150,000; 9%, 150,000 –200,000; and 7%, above 200,000 Swiss
To overcome these limitations, we used the Q sort as a francs. (In the current period, 1 dollar is equal to 1.3 Swiss francs,
time-honored method of person-centered assessment. The Q and 1 euro corresponds to 1.5 Swiss francs). In comparison, the
sort captures the unique ordering of multiple attributes within median family income of the city population is between 90,000
the individual being described rather than scores on single and 100,000 Swiss francs. With respect to educational background
traits. Thus, this method is suited to capturing the intergenera- of the parents, 42% held an advanced high school degree, 14% had
tional concordances in systems of ideals. In addition, the Q sort obtained a degree from an advanced professional training school,
limits responses that result from a social desirability bias, which and 38% held a university degree. In the city population, the
is particularly important when many options are desirable, as respective percentages are 33%, 13%, and 23%. Hence, our sample
with ideals. Most important, because of response sets, the was biased toward higher education levels, but it was relatively
representative for ethnicity, language, family composition, and to sort the cards, as described, into piles ranging from “those
income. attributes you wish your child would develop the least (1) to
those that you wish your child would develop the most in the
future (9).” Instructions for desired life aspirations were for-
Procedure and Measures mulated in analogous terms ranging from “life goals you wish
Time 1. At Time 1, families were invited to two sessions, both your child would least (1) to those you wish your child would
held at the University of Geneva. Separated by a minimum of 4 days, most (7) pursue in his/her future life.” Finally, for perceptions
each session lasted about 2 hr, although the time needed to complete of parent ideals, adolescents were asked to sort the cards
the procedures varied across participants. The participants were tested ranging from traits/life goals “my mother/father would like me
in groups of 4 –20. Two research assistants were always present to to have the least to those my mother/father would like me to
give an introductory explanation of the tasks, monitor the sessions, have the most.”
and answer questions while the participants completed the tasks. After Parenting dimensions were assessed with the revised Block
the initial introduction, parents and adolescents were placed in differ- Child Rearing Practices Report (CRPR; Rickel & Biasatti, 1982),
ent rooms to prevent communication or exchange of information administered to both parents and adolescents after completion of
between family members. Across the two sessions, adolescents were the Q sorts. The revised CRPR includes two dimensions, nurtur-
instructed to complete six different Q sorts: ideal self, perception of ance and restrictiveness. Nurturance relates to flexible and warm
maternal ideals, and perception of paternal ideals, each being done for child-rearing practices, such as encouraging independence and
two content domains, personality traits and life goals. Parents, in turn, showing support and affection (e.g., “I express affection for my
were asked to complete four different Q sorts: ideal self and the ideals child, for example, by holding and hugging my child”; “I encour-
they held for their child, each Q sort being done for personality traits age my child to be independent from me”; “I encourage my child
and life goals. To avoid carryover, parents provided their self-related to be curious, to explore and to question things”). Restrictiveness
Q sorts in one session and their child-related Q sorts in the other. relates to parental attitudes such as expecting respect, protective-
Adolescents provided their ideal-self Q sort and one of the two ness, restriction of showing feelings, and belief in physical pun-
perceived parental ideals Q sorts in one session and the second Q sort ishment (e.g., “I do not allow my child to get angry with me,” “I
relating to perceived parental ideals in the second session. teach my child to keep control of his/her feelings at all times,” “I
The personality Q sort was based on Block’s Q set for use do not allow my child to question my decisions”). Dekovic,
with lay people (see J. Block, 1961/1978, Appendix H). It Janssens, and Gerris (1991) found self-reports based on this in-
contains 70 trait-descriptive items, which are unambiguous and strument to be significantly related to observations of actual pa-
easy to understand. For the present version, items that are of rental behaviors. On the basis of the original report, we used the 19
particular significance in an adolescent population were added. most discriminating items, 8 for nurturance and 11 for restrictive-
In addition, some of the original items were replaced for the set ness (see Rickel & Biasatti, 1982). For mother reports, coefficient
to approximate the item pool of the most current model of alpha was .76 for nurturing and .77 for restrictiveness. In the child
personality, the five-factor model. These changes produced 78 reports of maternal parenting, the respective alphas were .79 and
items, which are listed in Table A1 of Appendix A. The life .77. For father reports, coefficient alpha was .68 for nurturing and
aspiration Q sort was based on the comprehensive set of 38 life .79 for restrictiveness. The respective alphas for child reports of
goals recently proposed by Roberts and Robins (2000). The father parenting were .79 and .77. To increase the reliability of the
authors demonstrated that this set of goals is not redundant with parenting measure, we computed a composite score for each parent
personality, as defined by the five-factor model. The items are for each parenting style from the two adolescent ratings and the
listed in Table A2 of Appendix A. parent’s self-rating.
In the ideal-self Q sort, parents and children were instructed Time 2. Approximately 7 months after the initial assessments,
to sort the cards according to the person they would like to be families were invited to the university to participate in the follow-
or become ideally. With this instruction in mind, they were up. The purpose of the assessments at Time 2 was fourfold: first,
asked to sort the 78 personality attributes into nine categories to assess the ideal-self stability between Time 1 and Time 2;
ranging from 1 (least desired) to 9 (most desired). The partic- second, to examine whether the concurrent findings would also be
ipants were required to sort the cards according to a predeter- evident longitudinally; third, to run cross-lagged analyses between
mined discrimination-maximizing distribution (i.e., 14 items parents’ and children’s ideals; and finally, to gather additional data
into the middle category and 8 items in the remaining eight that were unrelated to the purposes of this study. So as not to
categories). The same procedure was then repeated for the 38 overtax the attentional capacities of our participants, only a selec-
life goals, which, because of their lesser number, had to be tion of the tasks administered at Time 1 was repeated at Time 2.
placed into seven categories (i.e., eight items in the middle Specifically, adolescents were asked to complete two Q sorts, one
category and five in the remaining six categories; see J. Block, for their ideal self and one for perceived maternal ideals— both
1961/1978, for details on Q-sort formatting). To describe their relating to life goals only. Both parents were asked to complete
child-directed ideals for personality, parents were instructed to two life-goal Q sorts, one regarding the ideals they held for their
let themselves dream and to imagine the ideal future character- child and one relating to their ideal self. Thus, in comparison to
istics of their child. We decided to ask parents about their ideal Time 1, analyses at Time 2 were restricted to life aspirations. In
future child because pilot work suggested that rating an ideal addition, because perception of the parental ideals was collected
present child would induce excessive comparisons with the only for mothers at Time 2, perceptual accuracy and acceptance
current child. From this general instruction, parents were asked could be assessed only for mother– child dyads.
Results particular assumptions concerning statistical distributions (ex-

cept for the assumption of independent observations), they are
Data Analytic Strategy increasingly applied even in the context of traditional statistical
tests such as correlations, t tests, and ANOVAs (Good, 1994;
Units of analysis were Q correlations, that is, profile-based
May & Hunter, 1993). In the current context, we used the
similarity scores computed within dyads. For example, the con-
permutation test to examine whether, for each of the thousands
gruence index between the parents’ ideal-self template and their
of possible random combinations (permutations), the test sta-
ideal-child template was derived by computing the correlation
tistic was equal to the statistic based on the true sample. A
between the numerical value assigned to each item in the parental
primer on permutation analysis, including a simple example
ideal-self Q sort and the value assigned to each item in the parent’s
that illustrates how the test was applied in the current context,
ideal-child Q sort. This index quantitatively expresses the degree
is provided in Appendix B. To determine whether the difference
to which the parent’s ideal self is congruent with the ideals that the
between each of the distributions was statistically significant,
parent holds for his or her child (J. Block, 1961/1978; Klohnen &
we used 100,000 permutations to maximize precision (1,000
Mendelsohn, 1998; Zentner, 2005). Using this similarity index, the
permutations are usually considered sufficient). As reported in

subsequent analyses exemplify how permutation analysis can be
Table 1 (see under Permutation test), the differences in mean

used for distinguishing true from random similarity between re-
between familial (diagonal) similarities and random (off-
lated individuals.
diagonal) similarities were consistently significant for mothers
Aside from the conventional analyses based on Gaussianity
and least squares (regressions, t tests, analyses of variance and fathers in both content domains, personality and life goals.
[ANOVAs]), we also ran the equivalent robust analyses (In- Supplementary analyses confirmed these results.3
sightful Corporation, 2002; Wilcox, 1997). Robust methods are To compare the findings from the permutation test with
designed to give very similar results to classical methods when results obtained through a more conventional procedure, we
the observations are (almost) Gaussian, but they are not influ- also ran one-sample t tests comparing familial similarities with
enced by outlying observations, that is, observations that are far the average of the random similarities. It should be noted that,
from the bulk of the data and that are probably not representa- in contrast to the permutation test, the t test requires that the
tive of the general tendency of the observed phenomenon. In the value with which the sample is compared not depend on the
subsequent analyses, results of the robust analyses were very sample. Even though this condition for the t test was not met,
close to those of the conventional analyses, strengthening con- the findings obtained from t tests were similar to the findings
fidence in the findings and their generalizability. Because of obtained through the permutation test. The complete t values
this overlap, we report conventional analyses only.
1
The number of random congruence coefficients is equal to N ⫻ N – 1.
The actual N of both random and familial dyads varied with the type of
Dissociating Familial From Random Parent–Child dyad (mother– child, father– child), with the type of ideal (ideal self of the
Similarity adolescent, perceived parental ideals, actual parental ideals, etc.), and
because a few participants did not provide all of the Q sorts. This yielded
A primary goal of this study was to examine the extent to the following totals: N ⫽ 109 –115 familial mother– child dyads and N ⫽
which familial parent– child concordance in ideals exceeded 11,772–13,110 random mother– child pairs; N ⫽ 62– 68 familial father–
random parent– child concordance, that is, concordance based child dyads and N ⫽ 3,782– 4,556 random father– child pairs.
2
on unrelated, randomly combined parent– child dyads. Familial Kernel density estimation is similar to plotting the data in a histogram and
versus random similarity was dissociated for all types of then smoothing the histogram boxes to a continuous distribution, but it offers
parent– child concordance in ideals in the following order: some advantages over histograms. First, the kernel diagrams provide the
congruence, transposition, assimilation, accuracy, and accep- possibility of superimposing curves for the random and familial similarities
within the same graph and of comparing them precisely. Second, the shape of
tance. In all cases, the procedure was the same: We first
the histogram depends on the starting point of the boxes, whereas this is not the
computed familial parent– child similarity for the actual fami-
case for the kernel estimation. Third, histograms lead to loss of information
lies. For example, with respect to congruence, we computed the because observations are replaced by the central point of the interval in which
correlations between each parent’s ideal-self Q sort and his or they fall (see Silverman, 1986, for details).
her own child’s ideal-self Q sort. We then computed the random 3
To examine the robustness of the present findings, we ran the same
similarity by correlating each parent’s ideal-self Q sort with all tests with two other similarity coefficients: (a) the sum over all items of the
of the other children’s ideal-self Q sorts in the sample.1 This absolute difference (sometimes called the L1 distance) and (b) the sum
yielded two distributions, a familial and a random distribution, over all items of the squared difference (often called the D2 distance).
as is illustrated in Figure 2 for transposition, accuracy, and Moreover, in addition to differences in mean as exemplified in Appendix
acceptance. The distributions are represented with kernel den- B, we also used two supplementary nonparametric tests to examine the
sity estimators, which have some advantages over histograms.2 difference in the familial versus random distributions: the rank statistic and
the Kolmogorov-Smirnov test. Whereas the rank statistic is not influenced
The high random values indicated that ideals are heavily shared
by outliers or by nonnormality, the Kolmogorov-Smirnov statistic tests the
within a culture; thus, henceforth, we refer to random similarity
maximal distance between the distributions. Thus, for each level (congru-
also as culture-level similarity. ence, transposition, assimilation, etc.), we ran nine tests (using each of the
An ideal statistical tool to determine whether the difference three similarity indices and each of the three test statistics) for four cases,
between the familial and cultural distributions is statistically personality and life goal and mothers and fathers, yielding 36 values for
significant is the permutation test (e.g., Efron & Tibshirani, each of the five levels. Using the alternative similarity indices and statistics
1993; Hubert, 1984). Because permutation tests require no yielded results that are very similar to those reported in Table 1.
Figure 2. Kernel density estimators illustrating the distribution of similarity scores of both familial and cultural
random dyads for acceptance. Values on the x-axis are similarity indices (Pearson’s r); values on the y-axis are
standardized frequencies (density). The four graphs in the first column illustrate the distribution of similarity
correlations of both familial and random dyads for transposition (i.e., the similarity between the parental ideal self and
the ideals parents hold for their children), the graphs in the second column illustrate the distribution for perceptual
accuracy (the similarity between perceived and actual parental ideals), and the graphs in the third column illustrate the
distribution for acceptance (the similarity between children’s ideal self and perceived parental ideals).
are reported in Table 1. The last two columns in Table 1 also held for fathers, accuracy: ␤ ⫽ .36, t ⫽ 4.46, p ⬍ .01;
supplement the significance values with two effect-size esti- acceptance: ␤ ⫽ .27, t ⫽ 2.78, p ⬍ .01; interaction: ␤ ⫽ .19, t ⫽
mates, d and r.4 As can be seen, the extent of the familial versus 1.16, p ⫽ .12; R2 ⫽ .33. In summary, both accuracy and accep-
random difference varied appreciably across the various levels tance accounted for a substantial variance in assimilation. The
of concordance: As should be the case according to the current hypothesized interaction effect emerged for the personality do-
model, values for the intermediate processes, those supposed to main, but not the life-goal domain. Although this result could
account for the indirect links (transposition, accuracy, and indicate a domain-specific outcome, it is equally likely due to
acceptance), were higher than the values for the two indirect power limitations, which notoriously restrict the chances of de-
links (congruence and assimilation). Interestingly, effect sizes tecting hypothesized interaction effects in family research (e.g.,
were larger for ideals relating to life goals than for those Whisman & McClelland, 2005). The interaction effect for the
relating to personality traits. personality domain can be seen in the top panel of Figure 3. It
illustrates that acceptance is strongly predictive of assimilation, but
Pathways From Parents’ to Children’s Ideal Self only for adolescents high in perceptual accuracy.
In a second set of analyses, we examined the role of ideal-self

Person-centered regressions. Recall that previous research transposition (similarity between parents’ ideal self and the ideals
suggested that assimilation of parental viewpoints is best described they hold for their children) and of assimilation (similarity be-
in terms of a two-step model involving accurate perception and tween parents’ child-directed ideals and the child’s ideal self) and
acceptance of the parental position. Thus, in a first set of analyses, their interaction in accounting for parent– child similarity in the
we examined the role of perceptual accuracy (similarity between ideal self at Time 1 and, for life aspirations, also at Time 2. On the
parents’ actual ideals for their child and the child’s perception of
basis of the previous findings, we used assimilation scores as
these ideals) and acceptance (similarity between the child’s per-
shorthand for the combined effects of accuracy and acceptance.
ception of parental ideals and the child’s ideal self) in accounting
Using the identical computations as before, we found that, with
for assimilation (direct similarity between parents’ ideals for their
respect to ideals for personality, assimilation contributed signifi-
children and the child’s ideal self). Concurrent analyses were run
cantly to the variance in mother– child ideal-self similarity; in turn,
for both content domains, life goals and personality, whereas
effects of transposition and of the Transposition ⫻ Assimilation
longitudinal analyses were restricted to life goals (see Method
interaction did not reach significance, transposition: ␤ ⫽ .09, t ⫽
section, above). As has been exemplified in research on internal-
1.52, p ⫽ .13; assimilation: ␤ ⫽ .70, t ⫽ 11.15, p ⬍ .01;
ization of values (Knafo & Schwartz, in press), in operational
terms, this aim can be translated into a main effect for accuracy in interaction: ␤ ⫽ .06, t ⫽ 1.35, p ⫽ .17; R2 ⫽ .65. However, all
predicting assimilation, a main effect for acceptance, and an in- three predictors contributed to father– child ideal-self similarity,
teraction of Accuracy ⫻ Acceptance. The interaction reflects the transposition: ␤ ⫽ .44, t ⫽ 5.64, p ⬍ .01; assimilation: ␤ ⫽ .64,
contingency between accuracy and acceptance predicted by the t ⫽ 7.50, p ⬍ .01; interaction: ␤ ⫽ .16, t ⫽ 2.97, p ⬍ .01; R2 ⫽
model. Thus, acceptance should be predictive of assimilation, but .72. With respect to life aspirations, all three predictors contributed
only for those adolescents high in perceptual accuracy; for those to mother– child ideal-self similarity, transposition: ␤ ⫽ .21, t ⫽
low in perceptual accuracy, the effect of acceptance should disap- 3.15, p ⬍ .01; assimilation: ␤ ⫽ .64, t ⫽ 9.80, p ⬍ .01; interaction:
pear or at least diminish sharply. Multiple regressions tested the ␤ ⫽ .14, t ⫽ 2.34, p ⫽ .02; R2 ⫽ .58. A similar pattern emerged
independent and interactive effects of accuracy and acceptance on for father– child ideal-self similarity in life aspirations, transposi-
assimilation, separately for mothers and fathers, at Times 1 and 2.
4
Accuracy of perception and acceptance scores were subjected to Because we could not find guidelines on how to compute effect sizes
Fisher’s r-to-z transformation and subsequently centered around for the permutation test, we relied on the t test for effect-size estimations.
their mean (Aiken & West, 1991). Thereafter, the interaction term This seems reasonable because the t test and the permutation test are both
was computed by multiplying accuracy by acceptance. related to the difference in the means of the familial similarities and the
random similarities. Thus, we used Cohen’s d, using the standard deviation
For personality characteristics, accuracy, acceptance, and their
of the familial similarities as denominators, because the familial similari-
interaction all contributed to the variance in assimilation of moth-
ties are the reference values. The formula for the effect size is as follows:
ers’ child-related ideals, accuracy: ␤ ⫽ .47, t ⫽ 7.12, p ⬍ .01; [mean(familial) ⫺ mean(random)]/SD(familial). When computing the con-
acceptance: ␤ ⫽ .44, t ⫽ 6.71, p ⬍ .01; interaction: ␤ ⫽ .16, t ⫽ fidence intervals (CIs) around d, we found a range of d equal to 0.15– 0.20
2.50, p ⫽ .01; total R2 ⫽ .48.5 Similarly, all three predictors also (or r ⫽ .07–.10) on each side of the value. Because the CI for d increased
contributed to the assimilation of fathers’ child-related ideals, by about 50% for the individual genders (because of the restricted N), we
accuracy: ␤ ⫽ .62, t ⫽ 8.19, p ⬍ .01; acceptance: ␤ ⫽ .22, t ⫽ refrained from reporting d and r values for girls and boys separately.
2.68, p ⫽ .01; interaction: ␤ ⫽ .13, t ⫽ 2.06, p ⫽ .04; R2 ⫽ .71. However, we tested whether the d values for girls and boys were signifi-
For life goals, both accuracy and acceptance predicted assimilation cantly different from each other and found no differences with the excep-
of maternal ideals, accuracy: ␤ ⫽ .61, t ⫽ 8.17, p ⬍ .01; accep- tion of the difference for acceptance of personality ideals for fathers ( p ⫽
.02) and a marginal difference for acceptance of life-goal-related ideals for
tance: ␤ ⫽ .20, t ⫽ 2.63, p ⬍ .01. The interaction term failed to
mothers ( p ⫽ .06). Computations of the significance of the differences in
reach significance, ␤ ⫽ .08, t ⫽ 1.10, p ⫽ .28; R2 ⫽ .48. A similar
d were based on a stratified procedure in which resampling was done
pattern was found for fathers, accuracy: ␤ ⫽ .62, t ⫽ 6.71, p ⬍ .01; separately within the families of girls and boys (see Efron & Tibshirani,
acceptance: ␤ ⫽ .25, t ⫽ 2.81, p ⬍ .01; interaction: ␤ ⫽ .09, t ⫽ 1993; details regarding these computations are available from us).
1.03, p ⫽ .31; R2 ⫽ .56. Time 1 accuracy and acceptance also 5
In all current regression analyses, the degrees of freedom for the
predicted assimilation of maternal life goals at Time 2, accuracy: denominator at Time 1 ranged from 102 to 105 for mothers and from 57 to
␤ ⫽ .41, t ⫽ 4.46, p ⬍ .01; acceptance: ␤ ⫽ .27, t ⫽ 2.78, p ⬍ .01; 61 for fathers. At Time 2, degrees of freedom ranged from 95 to 99 for
interaction: ␤ ⫽ .05, t ⫽ .54, ns; R2 ⫽ .28. The predictive relations mothers and from 57 to 58 for fathers.
Table 1
Significance Levels and Effect Sizes (Cohen’s d and r) Concerning the Degree to Which Familial Similarity in Different Parent–Child
Ideals Exceeds Random (⫽ Cultural) Similarity in the Same Ideals
Mothers Fathers
Effect size Effect size

Type of Cultural Familial Permutation Cultural Familial Permutation
concordance M M test t value d r M M test t value d r
Personality
Congruence .63 .65 .00007 1.83 0.17 .09 .59 .61 .02046 1.27 0.16 .08
Boys .58 .61 .55 .56
Girls .65 .67 .60 .63
Transposition .69 .80 .00001 13.74 1.28 .54 .64 .75 .00001 5.29 0.64 .30
Boys .68 .80 .63 .73

Girls .69 .80 .64 .76

Assimilation .63 .66 .00001 2.60 0.25 .12 .63 .65 .01082 1.36 0.17 .09
Boys .59 .62 .57 .56
Girls .65 .68 .65 .69
Accuracy .67 .71 .00001 4.72 0.44 .21 .65 .69 .00005 2.81 0.35 .17
Boys .63 .69 .64 .66
Girls .68 .71 .66 .70
Acceptance .61 .73 .00001 8.77 0.83 .38 .61 .74 .00001 8.95 0.89 .41
Boys .55 .69 .56 .72
Girls .64 .74 .63 .75
Life goals
Congruence .40 .48 .00001 4.64 0.43 .21 .34 .42 .00001 3.93 0.48 .24
Boys .33 .38 .28 .34
Girls .44 .53 .37 .45
Transposition .45 .68 .00001 16.13 1.53 .61 .36 .63 .00001 12.26 1.49 .59
Boys .40 .64 .34 .62
Girls .47 .69 .38 .63
Assimilation .40 .52 .00001 7.14 0.67 .32 .37 .47 .00001 4.95 0.61 .29
Boys .33 .45 .29 .44
Girls .44 .55 .40 .49
Accuracy .43 .57 .00001 10.40 0.98 .44 .36 .51 .00001 6.57 0.81 .38
Boys .37 .54 .32 .48
Girls .46 .58 .38 .52
Acceptance .40 .66 .00001 15.90 1.48 .59 .38 .62 .00001 12.96 1.27 .54
Boys .31 .59 .28 .55
Girls .45 .70 .42 .65
Note. N ⫽ 109 –115 familial mother– child dyads (N ⫽ 11,772–13,110 random mother– child pairs); N ⫽ 62– 68 familial father– child dyads (N ⫽
3,782– 4,556 random father– child pairs). Congruence ⫽ average similarity between parents’ and children’s ideal self; Transposition ⫽ average similarity
between parents’ ideal self and their ideals for their children; Assimilation ⫽ average similarity between parental ideals for their children and their children’s
ideal self; Accuracy ⫽ average similarity between parental ideals for their children and the perception of these ideals by the children; Acceptance ⫽ average
similarity between parent ideals as perceived by the child and the child’s ideal self.
tion: ␤ ⫽ .19, t ⫽ 1.68, p ⫽ .09; assimilation: ␤ ⫽ .57; t ⫽ 5.33, Directionality analyses. The previous analyses examined po-
p ⬍ .01; interaction: ␤ ⫽ .19, t ⫽ 1.97, p ⫽ .05; R2 ⫽ .44. As is tential pathways of transmission, including interrelationships be-
illustrated in the bottom panel of Figure 3, the relatively consistent tween processes as predicted by the model. However, these anal-
interaction effect indicates that the effects of assimilation on yses did not address the direction of effects between parent and
parent– child congruence in the ideal self were boosted by high child within each single process. For this reason, we ran cross-
levels of transposition. Initial assimilation, but not initial transpo- lagged analyses between children’s and parents’ ideals at Times 1
sition, predicted congruence of mother– child ideal-self similarity
at Time 2, transposition: ␤ ⫽ ⫺.04, t ⫽ ⫺.46, ns; assimilation:
6
␤ ⫽ .49, t ⫽ 4.88, p ⬍ .01; interaction: ␤ ⫽ ⫺.016, t ⫽ ⫺.16, ns; An alternative way to test the current model is via mediational analy-
R2 ⫽ .27. The longitudinal findings for fathers were similar, ses. Because mediational analyses are variable centered and thus require
that the current life-goal and personality patterns be decomposed into
transposition: ␤ ⫽ .05, t ⫽ .37, p ⬍ .01; assimilation: ␤ ⫽ .50, t ⫽
single variables to be considered one at a time, they were not ideal in the
3.76, p ⬍ .01; interaction: ␤ ⫽ ⫺.20, t ⫽ ⫺1.60, p ⫽ .12; R2 ⫽ present data context. In addition, several variables did not meet the re-
.31. In summary, most of the predictions for the congruence arc quirements for mediation as defined by Baron and Kenny (1986). Still, we
were supported concurrently. Longitudinally, only assimilation did run mediation analyses for variables in which conditions for mediation
contributed significantly to congruence, whereas the effects of analyses were met and found that the analyses supported the current model.
transposition and the interaction did not replicate.6 (These analyses are available from us upon request.)
Figure 3. Illustration of the interaction between accuracy and acceptance (top panel) and between transposition
and assimilation (bottom panel). Both graphs relate to father– child dyads (personality). Scale units of both axes
are r-to-z values.
and 2. In the context of the present person-centered data scheme, ideal-self T1 and the maternal ideal-self T2, suggesting a parent-
in which dyadic similarity correlations are the main units of to-child effect for congruence. Findings for all the other levels are
analysis, this aim translated into comparing the average concor- reported in the following rows. Even though most effects were
dances between the parent ideals at Time 1 and the child ideals at either from parent to child or nonsignificant, a child-to-mother
Time 2 versus the average concordances between the child ideals effect emerged with respect to assimilation (see Table 2, third
at Time 1 and the parent ideals at Time 2. The logic of cross- row). As hypothesized in the introductory section, above, child-
lagged analysis implies that, if the former exceed the latter, the to-parent effects should be more likely for older adolescents than
directionality should be from parent to child, whereas the inverse for younger ones (see also next section). To examine this hypoth-
would suggest a child-to-parent effect. Recall that because, at esis, we recomputed the same analyses for a younger and an older
Time 2, data are restricted to the life-goal domain, these analyses age group based on a median split. Although the results were
were run for life goals only. The results are reported in Table 2. similar on the whole, it is important to note that the only child-
As can be seen in the first row regarding the parent– child to-parent effect that could be identified (assimilation) was re-
similarity in the ideal self (congruence), the mean of the (r-to-z stricted to the older adolescents, t(50) ⫽ ⫺2.40, p ⬍ .05. For the
transformed) similarity correlations between the maternal ideal- younger ones, there was no child-to-parent effect, t(48) ⫽ ⫺1.09,
self T1 and the child’s ideal-self T2 was higher than the inverse, p ⬎ .25. In addition, the child-to-parent effect for perceptual
that is, the mean of the similarity indices between the child’s accuracy, though nonsignificant for the entire sample (see Table 2,
Table 2
Cross-Lagged Correlations in Which the Average Parent T1 to Child T2 Associations Are Contrasted With the Average Child T1 to
Parent T2 Associations
Mother T1/ Child T1/ Father T1/ Child T1/

child T2 mother T2 child T2 father T2
Type of
concordance M SD M SD t p M SD M SD t p
Congruence .49 .19 .44 .18 2.83 ⬍.01 .42 .19 .40 .19 1.41 .17
Transposition .69 .15 .56 .18 8.18 ⬍.01 .64 .16 .57 .18 2.89 ⬍.01
Assimilation .52 .17 .55 .16 ⫺2.44 .02 .48 .18 .48 .17 0.17 .86
Accuracy .57 .14 .59 .13 0.80 .46 — — — — — —
Acceptance .65 .16 .65 .16 0.13 .90 — — — — — —
Note. Values are the original, nontransformed average correlations. Degrees of freedom for the denominator ranged from 95 to 99 in mother– child dyads
and from 57 to 58 in father– child dyads. Dashes indicate that perceived father ideals for accuracy of perception and acceptance were not collected at T2.
T1 ⫽ Time 1; T2 ⫽ Time 2.
fourth row), became marginally significant when analyses were suggest that the ideal self reaches a significant stage of matu-
restricted to the older age group, t(50) ⫽ ⫺1.89, p ⫽ .07. These ration between 17 and 18 years of age. This, in turn, offers a
findings suggest that the likelihood of child-to-parent effects in- possible explanation for the increased likelihood of child-to-
creases with the child’s age. parent effects in late adolescence.
Ideal-self consistency. One possible explanation for the age–
directionality link is that older adolescents’ ideal self is tem- Moderators of Parent–Child Concordance: Parenting and
porally more stable than the ideal self of younger adolescents,
Gender
which would put the older adolescents in a better position to
influence their parents’ ideals. Before testing the hypothesis of Parenting practices. Correlations between the aggregated par-
a positive relationship between age and ideal-self consistency, enting dimensions, nurturance and restrictiveness, with r-to-z
we first computed the average stability of adolescents’ ideal self transformed scores for accuracy of perception, acceptance, and
by correlating each adolescent’s Q sort at Time 1 with his or her assimilation of parental ideals are reported in Table 3. As hypoth-
Q sort at Time 2 and then averaging across all of the stability esized, nurturance was positively related to perceptual accuracy
coefficients. We obtained an average stability of .77 (SD ⫽ .13, and acceptance and also directly to assimilation. The significant
range .33–.94). It should be emphasized that this stability is associations of the three concordance variables with restrictive-
ipsative, that is, it denotes how much, on average, a given ness, though limited to mothers, were also in the anticipated
person’s configuration of attributes changes across time. This is negative direction. To examine the proportion of variance ac-
in contrast to the more widely used rank-order stability that counted for by parenting styles, we ran a series of regressions in
indicates how much the position of an individual changes which the joint contribution of child and parent reports of parent-
relative to other individuals in the group on a given variable ing to perceptual accuracy, acceptance, and assimilation was ex-
(e.g., anxiety). To make both stabilities comparable, Zentner amined. Between 7% and 27% of the variance in the three out-
(2005, p. 247) introduced a procedure to estimate the rank-order comes could be accounted for by parenting dimensions. These
values corresponding to the ipsative value, which is r ⫽ .80 in findings are summarized in the rows beneath mother and father
the current context. This is a relatively high level of stability parenting of Table 3.
that is similar to the stability of personality traits in late ado- In addition to the concurrent analyses, we also examined whether
lescence (Roberts & DelVecchio, 2000). parenting predicted accuracy of perception, acceptance, and assimi-
Subsequently, we tested the hypothesis of a positive age– lation at Time 2, while controlling for the respective autocorrelations.
stability relationship by correlating adolescents’ age with their As usual, these longitudinal analyses were limited to life goals.
(r-to-z transformed) stability correlations from T1 to T2. We Consistent with the hypotheses, initial mother restrictiveness was
found a trend indicating that the older the adolescents, the more negatively related to perceptual accuracy and to acceptance at T2
consistent their ideal self, r(109) ⫽ .18, p ⫽ .06. To examine (partial rs ⫽ ⫺.22, ⫺.25, p ⬍ .05, p ⫽ .01, respectively), but not to
whether this association was linear, we created three age groups T2 assimilation (partial r ⫽ ⫺.09, ns). Initial maternal nurturance was
based on tertiles. Remarkably, the correlation for the youngest marginally predictive of T2 acceptance (partial r ⫽ .16, p ⫽ .11), but
group (15.00 –16.85 years) was appreciable (r ⫽ .40, p ⫽ .01), it did not predict perceptual accuracy (partial r ⫽ .10, ns) or assimi-
whereas for the middle group (16.90 –18.10 years) and the lation (partial r ⫽ .04, ns). The only predictive relationships that could
oldest group (18.15–19.80 years), the correlation was immate- be computed for father– child dyads, namely, T2 assimilation from
rial (rs ⫽ .18 and ⫺.17, respectively; both ns). To formally test fathers’ nurturance and restrictiveness at T1, did not reach signifi-
whether age predicted ideal-self stability differentially across cance (nurturance: partial r ⫽ .20, p ⫽ .14; control: partial r ⫽ .15,
the three age groups, we ran an analysis of covariance with age ns). Overall, then, parenting contributed modestly but consistently to
group as factor and age as the covariate. Indicative of the accuracy of perception, acceptance, and assimilation in the concurrent
differential relation between age and stability across age groups analyses. In addition, the partial correlations from T1 to T2 also
is the interaction between age group and the covariate, which suggest that parenting practices may be driving forces behind some
was significant, F(2, 103) ⫽ 3.79, p ⫽ .03. These findings parent– child concordances in ideals.
Table 3
Relationship of Parenting Practices to Perceptual Accuracy, Acceptance, and Assimilation of Parental Ideals
Accuracy of perception Acceptance Assimilation

Accuracy, acceptance, and assimilation
correlations with: Personality Life goals Personality Life goals Personality Life goals
Mother parenting
Mother nurturance .18* .23* .36** .12 .26** .23*
Mother restrictiveness ⫺.05 ⫺.23* ⫺.19* ⫺.27** ⫺.16 ⫺.24*
Variance explained by mother parenting (in R2) R2 ⫽ .12 R2 ⫽ .09 R2 ⫽ .15 R ⫽ .10
2
R2 ⫽ .11 R2 ⫽ .13
F test for significance of R2 F ⫽ 3.19** F ⫽ 2.41* F ⫽ 4.10** F ⫽ 2.55* F ⫽ 2.79* F ⫽ 3.56**
Father parenting
Father nurturance .25* .16 .36** .35** .19† .32*
Father restrictiveness .02 ⫺.03 .08 ⫺.02 ⫺.15 ⫺.07
Variance explained by father parenting (in R2) R2 ⫽ .18 R2 ⫽ .07 R2 ⫽ .27 R2 ⫽ .14 R2 ⫽ .15 R2 ⫽ .16
F test for significance of R2 F ⫽ 2.94* F ⫽ 1.09 F ⫽ 4.53** F ⫽ 2.02† F ⫽ 2.21† F ⫽ 2.63*

Note. Degrees of freedom for the numerator are always 4. Degrees of freedom for the denominator ranged from 92 to 98 for mother– child dyads and from
49 to 54 for father– child dyads.
†
p ⱕ .10. * p ⱕ .05. ** p ⱕ .01.
Gender differences. To examine the effects of child and parent indicates that children’s ideal self was more similar to their moth-
gender with respect to the various concordances in parent– child er’s than to their father’s ideal self. A main effect of child gender
ideals, we used a mixed effects model ANOVA with family as a indicates that, compared with boys’ ideal self, that of girls was
random factor, child gender as a nesting fixed factor, and parent more similar to their parents’ ideal self, again with respect to both
gender as a crossed fixed factor, with an interaction between the domains, personality ideals and life aspirations. Interactions be-
two fixed factors. (This analysis is similar to a repeated measures tween both factors, which would indicate that the effect of child
ANOVA, with the advantage of allowing one to examine the gender was qualified by parent gender, were not found. In inter-
effects of child gender, parent gender, and their interaction based preting the findings, it is important to keep in mind that the extent
on all available dyads. The traditional repeated measures ANOVA, to which familial values exceeded random values did not differ
in contrast, restricts the sample to families in which both parents significantly between boys and girls (see footnote 4). Therefore,
participated, thereby eliminating all of the single-parent families. higher values for girls cannot simply reflect a higher inclination to
For the sake of comparison, we also ran the latter analysis and adopt cultural ideals. Overall, the findings support the hypothesis
found that, on the whole, the results did not change.) The complete that girls tend to develop aspirations that are more similar to their
results are reported in Table 4. A main effect of parent gender parents’ ideals than do boys.
Table 4
F Ratios and Significance Levels Examining Gender Differences in Various Types of Similarities
Between Parents’ and Child Ideals
Main effect: child gender Main effect: parent gender Interaction

Type of
concordance F p Direction F p Direction F p
Congruence
Personality 4.33 .04 g⬎b 6.89 .01 m⬎f 0.01 .90
Life goals 14.32 ⬍.01 g⬎b 8.15 ⬍.01 m⬎f 2.32 .13
Transposition
Personality 0.18 .67 7.99 ⬍.01 m⬎f 0.23 .63
Life goals 0.80 .37 3.29 .07 m⬎f 0.67 .42
Assimilation
Personality 12.65 ⬍.01 g⬎b 1.21 .27 1.44 .23
Life goals 3.88 .05 g⬎b 2.54 .12 (m⬎f) 1.48 .23
Accuracy
Personality 2.80 .10 (g⬎b) 1.46 .23 0.01 .95
Life goals 2.65 .11 (g⬎b) 5.04 .03 m⬎f 0.02 .89
Acceptance
Personality 2.29 .13 (g⬎b) 0.49 .49 0.04 .85
Life goals 8.89 ⬍.01 g⬎b 7.21 ⬍.01 m⬎f 0.01 .94
Note. g⬎b indicates greater girl–parent compared with boy–parent similarity. m⬎f indicates greater mother–
child compared with father– child similarity. Parentheses are used when results merely approach significance.
Degrees of freedom for the numerator are always 1. Degrees of freedom for the denominator ranged between 112
and 120 for child gender and between 53 and 60 for parent gender and the interaction, except for acceptance,
where they ranged between 94 and 97.
Discussion the effect size r for transposition, accuracy, and acceptance led
to an r of .16 (.61 ⫻ .44 ⫻ .59), a value that is relatively close
Summary of Key Findings to the effect size r ⫽ .21 for the direct similarity in mothers’
By investigating a broad range of potential parental sources of and children’s ideal self. This interpretation concurs with the
adolescents’ ideal self and by tracing its development from mid- to findings for additive and interactive effects of transposition,
late adolescence, the current findings contribute to the literature in accuracy, and acceptance in suggesting that, as predicted,
several ways. First, our study provides the strongest evidence yet, parent– child similarity in the ideal self is accounted for by
after controlling for culture-level similarity, that specific similarity these component processes. Although these processes received
between parents’ and children’s ideal selves exists. Second, we much support in the current research, it is also important to keep
extended the two-step model of internalization to provide a frame in mind that an appreciable part of the variance could not be
for understanding ideal-self transmission processes within the fam- explained by these processes, pointing to the need to consider
ily. Consistent with the proposed extension, we found that three additional sources of parent– child congruence in future re-
transmission steps account for parent– child similarity in the ideal search.
With the exception of the direct similarity in the ideal self
self: transfer of parental ideal self to ideals for the children,

children’s perception, and acceptance of these parental ideals. between parents and their children, the size of the effects for the
Third, we found that most processes accounting for parent– child intergenerational similarities was substantial compared with the
resemblance in the ideal self are meaningfully related to parenting size of effects typically found in the social sciences (Meyer et al.,
practices. Although these findings indicate that parental ideals and 2001). With respect to the variations in effect sizes across all levels
socialization practices contribute to resemblance between parent and domains reported in Table 1, two points should be made. First,
and child ideals, the findings do not rule out the possibility of effect sizes for transposition and acceptance were consistently
child-to-parent effects. Indeed, tentative evidence suggests that higher compared with those for congruence, assimilation, and
such reverse effects become more likely as the ideal self stabilizes perceptual accuracy. One explanation is that, in the case of trans-
in late adolescence. Fourth, compared with boys’ ideal self, that of position and acceptance, responses to be correlated were derived
girls was more similar to the ideals that parents held for their from the same person (the parent in the case of transposition and
children and for themselves. Finally and in a methodological the child in the case of acceptance). In contrast, correlations for the
perspective, the current research exemplifies how permutation latter three links were based on responses from different individ-
analysis can be used for distinguishing unique from culture-level uals within the family. It is known that correlations based on
similarity between related individuals. In the following sections, responses from the same informant tend to be inflated because the
we review the findings in more detail and consider their implica- systematic error variance is aligned (Meyer et al., 2001). Given
tions for the understanding of ideal-self development. that transposition and acceptance concern subjective processes
within an individual, it would not only have been difficult but also
inappropriate to assess these processes by relying on responses
Ideal-Self Transmission: A Telephone Game Effect?
from different individuals. It should also be noted that because our
One of the key findings of this study is that, when the effects analyses are concerned with relationships between variables rather
of cultural similarity are statistically removed, the actual sim- than absolute magnitudes, the validity of all the subsequent anal-
ilarity between parents’ and children’s ideal selves was slightly yses and conclusions is not affected by this point. However, it is a
above .20 for life goals and slightly below .10 for personality point to be kept in mind when interpreting the relative magnitude
traits. This finding is consistent with an extensive literature on of effect sizes across the various processes reported in Table 1.
parent– child resemblance, indicating that parent– child correla- Second and unexpectedly, effect sizes for parent– child concor-
tions are between .05 and .20 for personality traits and between dances in life aspirations were consistently higher than in person-
.15 and .30 for attitudes (see Loehlin, 2005). Contrary to folk ality ideals. One explanation for this difference is that, during late
wisdom, then, the apple appears to fall far from the tree. At adolescence, parents and adolescents are sensitized to the impor-
first, this finding seems to question the notion of the family as tance of making the right life choices, prompting frequent parent–
a major agency of transmission of ideals. However, one of the child discussions about life goals. Consequently, the child is mo-
distinctive aspects of this research is that it offers an explana- tivated to pay attention to parental messages in this domain, and in
tion for this generation gap. Our explanation is best described in turn, parental ideals about life goals are accessible to the child. In
terms of a metaphor. In the telephone game, a message is passed contrast, parental ideals for personality may simply be less visible
from one player to another but is progressively altered in and less important to the child than ideals for life goals. Accep-
transit. In the context of our study, the message is the parental tance of parental values, in turn, is linked to the legitimacy that
ideal self, and the players are the three steps, transposition, adolescents grant to parental authority in a given domain (Knafo &
perceptual accuracy, and acceptance. In transposition, the mes- Schwartz, in press; see also Smetana, 2000). Because of the more
sage is altered because not everything that parents wish for private character of ideals regarding personality, adolescents may
themselves do they also wish for their children. Subsequently, consider their parents’ authority as less legitimate when parents try
the message is altered again because what parents wish for their to influence them regarding personality than when they try to
children is perceived by the latter with imperfect accuracy. A influence them with life plans. A second, altogether different
final transmutation of the message occurs because parental explanation relates to the higher interindividual variability in ide-
ideals, even if accurately perceived, are not always accepted by als for life goals compared with personality (see Table 1). As a
the child. To add quantitative substance to this metaphorical consequence, the extent to which familial similarity coefficients
reasoning, consider the transmission in ideals for mother– child can depart from random similarity is much more limited for traits
dyads regarding life goals (see Table 1). The multiplication of than for life goals, thereby restraining effect sizes. This explana-
tion is conceptually related to the finding that observers are more supervision of girls is that parents convey their standards and
accurate when attributes are evaluatively neutral rather than highly expectations more emphatically to their daughters than to their
desirable or undesirable (Funder, 1995). sons (e.g., Pomerantz et al., 2001). As a result, girls may
become more invested in decoding parental ideals, leading to
Conservative Consequences of Liberal Parenting higher perceptual accuracy of these ideals. Furthermore, en-
dorsement of these ideals should also be enhanced, considering
Consistent with our hypotheses, we found that the more parent- girls’ tendency to evaluate themselves in terms of acceptance by
ing was characterized by attributes such as warmth, support, and their intimates.
encouragement of the child’s independence, the more similar was
the child’s ideal self to the ideals that parents wanted him or her to Ideal-Self Development
endorse. Conversely, the more that parenting was characterized by
lack of nurturance and by attributes such as control, protectiveness, A distinctive finding of the current research is that age predicts
and restriction of the child’s independence, the more the child’s ideal-self stability for adolescents until about age 17, after which
ideal self tended to deviate from the parents’ ideals. Paradoxically, the ideal self levels off at traitlike levels of stability. In our view,
then, those parents who seemed inclined to accept or even to this suggests that, by the end of adolescence, the ideal self has
actively promote their children’s aspirations to differ from their reached an important point of maturation. This finding ties up
own tended to have children whose aspirations were similar to nicely with the previously reported growth in the stability of values
theirs. In contrast, parents who seemed least inclined to accept across adolescence (Rokeach, 1973), and it is also in concert with
deviations from their socialization ideals tended to have rebellious more recent work showing that, from age 18, life goals exhibit
children, that is, children whose aspirations were dissimilar to traitlike levels of stability (Roberts, O’Donnell, & Robins, 2004).
theirs. It is important to note that we could also discern some Consistent with previous theorizing, these data concur in suggest-
predictive relationships suggesting that parenting, in addition to ing that adolescence is a sensitive period for the development of
being a correlate of parent– child concordances in ideals, may be a desirable hypothetical end states such as ideals, values, and life
driving force behind changes in such concordances. goals (e.g., Inhelder & Piaget, 1958; Moretti & Higgins, 1999).
One explanation for this apparent paradox is that because Thus, particular attention should be directed toward an understand-
nurturant, supportive parenting is related to a positive attitude ing of the influences impinging on this significant period of
of children toward the parents, such as perceptions of parental ideal-self development.
trustworthiness, fairness, and lack of hypocrisy (Rohan & The current research examined two aspects of parental influ-
Zanna, 1996), it should enhance the child’s motivation to attend ences, parenting practices and parents’ own ideals (for themselves
to and endorse parental ideals. Conversely, the less positive this and for their children). In terms of the role of parenting, we found
attitude is, the more likely the child’s motivation to differ from some evidence suggesting that parenting practices may be not only
his or her parents and to nourish his or her ideal self through a correlate of concordances in parent– child ideals but also a
influences outside the family environment will be. A second predictor of changes in concordances. Even though the short
possibility is that restrictive parents may express their ideal period of the study makes it difficult to extrapolate to longer
standards in ways that imply a higher degree of “oughtness.” As periods of time, it is tempting to think about the role of parenting
is known from work on reactance theory, attempts to restrict an within a broader developmental perspective. It is of particular
individual’s freedom often produce a reactive boomerang ef- relevance to remember in this context that parents who are sup-
fect, that is, an increase in the restricted behavior (Brehm & portive of their child during adolescence have, in most cases, also
Brehm, 1981). In the present context, this effect translates into been supportive before adolescence (e.g., McNally, Eisenberg, &
children of restrictive parents trying to defend their freedom by Harris, 1991). Hence, one way to think about parenting influences
rejecting those ideals that parents try to force upon them. This over time is that, during childhood, a warm and supportive parent
interpretation is consistent with previous reports of a relation- is likely to have fostered a positive parental image in the child that,
ship between authoritarian parenting and parent– child dissim- by the time the child becomes capable of forming consistent
ilarity (Rohan & Zanna, 1996). ideal-self representations, infuses his or her ideal self. Thus, the
In terms of gender differences, the present findings support association between parenting practices and parent– child concor-
the hypothesis that boys tend to develop aspirations that are dance in ideals, beyond reflecting a present state, may also reflect
more distinct from their parents’ aspirations than do girls. This a past history.
pattern held for both content domains, personality and life As for the direction of effects between parents’ and children’s
aspirations. The extent to which this outcome can be attributed ideals, parent-to-child effects tended to be slightly more frequent
to gender-specific patterns of socialization is a fascinating than child-to-parent effects. However, the presence of a couple of
question that deserves further scrutiny. An extensive literature child-to-parent effects reminds us of the importance of acknowl-
suggests that parents monitor the behavior of their daughters edging the possibility of bidirectional influences. More important,
more closely than the behavior of their sons (e.g., J. H. Block, reverse influences from child to parents were observed only for
1983; Pomerantz et al., 2001). One immediate consequence of adolescents who were 17 years or older. Although ideal-self sta-
this heightened parental monitoring of girls is that they tend to bility is one likely reason for this outcome, an additional consid-
spend more time in close proximity to the family than do boys, eration is that, as adolescents become young adults, their ideals
who, in turn, tend to be more encouraged to explore the non- progressively transmute into tangible realities, such as moving to
familial environment. As a result, girls may orient more than a specific destination, entering a given profession or field of study,
boys toward the family when developing ideal-self representa- or choosing particular longer term partners. At this juncture, it
tions. Another possible manifestation of the heightened parental becomes increasingly difficult for parents to overlook their chil-
dren’s aspirations, and as a consequence, parents may become also highlights various issues that remain to be investigated. We
more receptive to their children’s ideals. hope that it can serve as a point of departure from which further
Another way of looking at the current findings is that they research into these issues can be developed.
define boundaries of parental influence. More specifically, because
parental restrictiveness and lack of support were negatively related
to parent– child resemblance in ideals, we can hypothesize that the
References
less support children receive from their parents, the more they will Acock, A. C., & Bengtson, V. L. (1980). Socialization and attribution
be inclined to construct their ideal self based on the nonfamilial processes: Actual versus perceived similarity among parents and youth.
environment such as the school, peer relations, and the media. Journal of Marriage and the Family, 40, 501–515.
Furthermore, being a boy also reduces the likelihood of parental Aiken, L., & West, S. (1991). Multiple regression: Testing and interpreting
influences on ideal-self development. A third basis for predicting interactions. New York: Academic Press.
parental influences (or lack thereof) can be drawn from evidence that Albert, I., & Trommsdorff, G. (2003, August). Intergenerational transmis-
parent-to-child effects become less likely as the adolescent grows sion of family values. Paper presented at the 11th European Conference
older. Thus, the adolescent most receptive to parental influence on his on Developmental Psychology, Milan, Italy.
Aronfreed, J. (1968). Conduct and conscience. New York: Academic Press.

or her ideal self will be a girl who has not yet reached her 18th
Balswick, J. O., & Macrides, C. (1975). Parental stimulation for adolescent
birthday and whose parents adopt supportive educational practices. rebellion. Adolescence, 10, 253–266.
Bandura, A. (1971). Psychological modeling: Conflicting theories. Chi-
Caveats and Limitations cago: Aldine-Atherton.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator distinction in
This study has several limitations. First, the sample of families social psychological research: Conceptual, strategic, and statistical consid-
came from a Western European background. The extent to which erations. Journal of Personality and Social Psychology, 51, 1173–1182.
the findings can be replicated with families from other cultures Block, J. (1971). Lives through time. Berkeley, CA: Bancroft Books.
remains to be investigated. Second, because of the relatively short Block, J. (1978). The Q-sort method in personality assessment and psy-
duration of the study and the lack of certain measures at Time 2, chiatric research. Palo Alto, CA: Consulting Psychologists Press. (Orig-
inal work published 1961)
the developmental and causal inferences that can be drawn are
Block, J., & Robins, R. (1993). A longitudinal study of consistency and
limited. We should also note that our focus on adolescence as a change in self-esteem from early adolescence to early adulthood. Child
period of ideal-self development does not imply that precursors to Development, 64, 909 –923.
individuals’ ideal self could not be identified at earlier phases of Block, J. H. (1983). Differential premises arising from differential socializa-
development. Indeed, we feel that examining childhood precursors tion of the sexes: Some conjectures. Child Development, 54, 1335–1354.
to the ideal self in adolescence is a fascinating subject that needs Brehm, S. S., & Brehm, J. W. (1981). Psychological reactance: A theory
much more attention than it has hitherto received. Third, the of freedom and control. New York: Academic Press.
current work focused on gender, parental ideals, and parental Bugental, D. B., & Grusec, J. E. (2006). Socialization processes. In W.
practices as sources of adolescents’ ideal-self development. Future Damon (Series Ed.) & N. Eisenberg (Vol. Ed.), Handbook of child
research is necessary to understand other likely influences, such as psychology: Vol. 3. Social, emotional, and personality development (6th
ed., pp. 366 – 428). New York: Wiley.
additional parent and child characteristics (e.g. degree of child–
Cashmore, J. A., & Goodnow, J. J. (1985). Agreement between genera-
parent identification), extrafamilial factors (e.g., school, peers, tions: A two-process approach. Child Development, 56, 493–501.
media), and genetic influences. A final cautionary note applies to Cooley, C. H. (1902). Human nature and the social order. New York:
our measures of ideals. We opted for the Q sort in the context of Scribner.
this research for the advantages described in the introductory Cronbach, L. J. (1955). Processes affecting scores on “understanding of
section. However, there are also some disadvantages. Although the others” and “assumed similarity.” Psychological Bulletin, 52, 177–193.
forced distribution has the critical advantage of making ratings Cross, S., & Madson, L. (1997). Models of the self: Self-construals and
truly comparable across participants, the cost of forcing is that the gender. Psychological Bulletin, 122, 5–37.
overall level of ideals is identical across participants. In other Cross, S., & Markus, H. (1991). Possible selves across the life span.
words, our data do not allow us to determine whether some Human Development, 34, 230 –255.
Dekovic, M., Janssens, J. M., & Gerris, J. R. (1991). Factor structure and
individuals have particularly high or low levels of ideals. Simi-
construct validity of the Block Child Rearing Practices Report (CRPR).
larly, the extent to which participants’ ideals were chronically Psychological Assessment, 3, 182–187.
accessible (see Higgins, 1987) was not assessed in this research. Efron, B., & Tibshirani, R. (1993). An introduction to the bootstrap.
London: Chapman & Hall.
Concluding Comment Eisenberg, N., Carlo, G., Murphy, B., & van Court, P. (1995). Prosocial
development in late adolescence. Child Development, 66, 1179 –1197.
Following James’s (1902) pronouncement that “the man’s inte- Eisenberg, N., Miller, P. A., Shell, R., McNalley, S., & Shea, C. (1991).
rior is a battleground for what he feels to be two deadly hostile Prosocial development in adolescence: A longitudinal study. Develop-
selves, one actual, the other ideal” (p. 135), a century of research mental Psychology, 27, 849 – 857.
has amply documented the importance of ideal-self representations Freud, S. (1961). The ego and the id. In J. Strachey (Ed. & Trans.), The
standard edition of the complete psychological works of Sigmund Freud
for psychological functioning, in particular for emotion regulation
(Vol. 19, pp. 12–59). London: Hogarth Press. (Original work published 1923).
and self-esteem. However, little is known about the provenience Funder, D. (1995). On the accuracy of personality judgment: A realistic
and development of the ideal self. How does an ideal self come approach. Psychological Review, 102, 652– 670.
about? What are its determinants? And how can its developmental Good, P. (1994). Permutation tests: A practical guide to resampling
trajectory be characterized? The current research presents one of methods for testing hypotheses. New York: Springer-Verlag.
the most comprehensive studies of these questions to date, but it Grolnick, W., Deci, E., & Ryan, M. (1997). Internalization within the
family: The self-determination theory perspective. In J. E. Grusec & self-regulation: Developmental antecedents and functional conse-
L. Kuczynski (Eds.), Parenting and children’s internalization of quences. Review of General Psychology, 3, 188 –223.
values: A handbook of contemporary theory (pp. 135–161). New OCSTAT. (2002). Office Cantonal de la Statistique (OCSTAT): Commu-
York: Wiley. nications statistiques [Statistics Bureau of the Canton of Geneva: Sta-
Grusec, J. E., & Goodnow, J. J. (1994). Impact of parental discipline tistical communications]. Geneva, Switzerland: OCSTAT Publishing.
methods on the child’s internalization of values: A reconceptualization Ogilvie, D. M., & Clark, M. D. (1992). The best and the worst of it: Age
of current points of view. Developmental Psychology, 30, 4 –19. and sex differences in self-discrepancy research. In R. P. Lipka & T. M.
Harter, S. (1999). The construction of the self: A developmental perspec- Brinthaupt (Eds.), The self: Definitional and methodological issues (pp.
tive. New York: Guilford Press. 186 –222). Albany: State University of New York Press.
Higgins, T. E. (1987). Self-discrepancy: A theory relating self and affect. Ozer, D. (1993). The Q-sort method and the study of personality develop-
Psychological Review, 94, 319 –340. ment. In D. C. Funder, R. D. Parke, C. T. Keason, & K. Widaman (Eds.),
Hoffman, M. L. (1970). Moral development. In P. H. Mussen (Ed.), Carmichael’s Studying lives through time, personality and development (pp. 147–168).
manual of child psychology (Vol. 2, pp. 261–361). New York: Wiley. Washington, DC: American Psychological Association.
Horney, K. (1950). Neurosis and human growth. New York: Norton. Pomerantz, E., Saxon, J. L., & Kenney, G. A. (2001). Self-evaluation: The
Hubert, L. J. (1984). Statistical applications of linear assignment. Psy- development of sex differences. In G. B. Moskowitz (Ed.), Cognitive
chometrika, 49, 449 – 473. social psychology: The Princeton Symposium on the Legacy and Future
Inhelder, B., & Piaget, J. (1958). The growth of logical thinking from of Social Cognition (pp. 59 –73). Mahwah, NJ: Erlbaum.
childhood to adolescence. New York: Basic Books. Rickel, A. U., & Biasatti, L. L. (1982). Modification of the Block Child
Insightful Corporation. (2002). S-PLUS 6 Robust Library user’s guide. Rearing Practices Report. Journal of Clinical Psychology, 38, 129 –134.
Seattle, WA: Author. Roberts, B. W., & DelVecchio, W. F. (2000). The rank-order consistency
James, W. (1890). The principles of psychology. New York: Holt. of personality traits from childhood to old age: A quantitative review of
James, W. (1902). The varieties of religious experience. London: Routledge. longitudinal studies. Psychological Bulletin, 126, 3–25.
Keating, D. (2004). Cognitive and brain development. In R. Lerner & L. Roberts, B. W., O’Donnell, M., & Robins, R. W. (2004). Goal and
Steinberg (Eds.), Handbook of adolescent psychology (2nd ed., pp. personality trait development in emerging adulthood. Journal of Per-
45– 84). Hoboken, NJ: Wiley. sonality and Social Psychology, 87, 541–550.
Klohnen, E. C., & Mendelsohn, G. (1998). Partner selection for personality Roberts, B. W., & Robins, R. W. (2000). Broad dispositions, broad
characteristics: A couple-centered approach. Personality and Social
aspirations: The intersection of the Big Five dimensions and major life
Psychology Bulletin, 24, 268 –278.
goals. Personality and Social Psychology Bulletin, 26, 1284 –1296.
Knafo, A., & Schwartz, S. (2003). Parenting and adolescents’ accuracy in
Rogers, C. (1961). On becoming a person. Boston: Houghton Mifflin.
perceiving parental values. Child Development, 74, 595– 611.
Rohan, M. J., & Zanna, M. P. (1996). Value transmission in families. In C.
Knafo, A., & Schwartz, S. (in press). Accounting for parent– child value
Seligman, J. M. Olson, & M. P. Zanna (Eds.), The psychology of values:
congruence: Theoretical considerations and empirical evidence. In U.
The Ontario Symposium (Vol. 8, pp. 253–276). Hillsdale, NJ: Erlbaum.
Schönpflug (Ed.), Perspectives on cultural transmission. Oxford, En-
Rokeach, M. (1973). The nature of human values. New York: Free Press.
gland: Oxford University Press.
Sandler, J., & Rosenblatt, B. (1962). The concept of the representational
Kohlberg, L. (1966). A cognitive-developmental analysis of children’s sex
world. Psychoanalytic Study of the Child, 17, 128 –145.
role concepts and attitudes. In E. Maccoby (Ed.), The development of sex
Schafer, R. (1968). Aspects of internalization. Madison, CT: International
differences (pp. 82–173). Palo Alto, CA: Stanford University Press.
Universities Press.
Kuczynski, L. (Ed.). (2003). Hardbook of dynamics in parent– child rela-
Schönpflug, U. (2001). Intergenerational transmission of values: The role of
tions. Thousand Oaks, CA: Sage.
Larson, R. W., & Almeida, D. (1999). Emotional transmission in the daily transmission belts. Journal of Cross-Cultural Psychology, 32, 174 –185.
lives of families: A new paradigm for studying family processes. Journal Silverman, B. W. (1986). Density estimation for statistics and data anal-
of Marriage and the Family, 61, 5–20. ysis. London: Chapman & Hall.
Loehlin, J. C. (2005). Resemblance in personality and attitudes between Smetana, J. G. (2000). Middle-class African American adolescents’ and
parents and their children: Genetic and environmental contributions. In parents’ conceptions of parental authority and parenting practices: A
S. Bowles, H. Gintis, & M. Osborne Groves (Eds.), Unequal chances: longitudinal investigation. Child Development, 71, 1682–1686.
Family background and economic success (pp. 192–207). Princeton, NJ: Tangney, J. P., Niedenthal, P. M., Covert, M. V., & Hill-Barlow, D. (1998).
Princeton University Press. Are shame and guilt related to distinct self-discrepancies? A test of
Luo, S., & Klohnen, E. (2005). Assortative mating and marital quality in Higins’s 1987 hypotheses. Journal of Personality and Social Psychol-
newlyweds: A couple-centered approach. Journal of Personality and ogy, 75, 262–268.
Social Psychology, 88, 304 –326. Troll, L., & Bengtson, V. (1979). Generations and the family. In W. R.
May, R. B., & Hunter, M. A. (1993). Some advantages of permutation Burr, R. Hill, F. I. Nye, & I. L. Reiss (Eds.), Contemporary theories
tests. Canadian Psychology, 34, 401– 407. about the family (Vol. 1, pp. 127–161). New York: Free Press.
McNally, S., Eisenberg, N., & Harris, D. H. (1991). Consistency and Whisman, M. A., & McClelland, G. H. (2005). Designing, testing, and
change in maternal child-rearing practices and values: A longitudinal interpreting interactions and moderator effects in family research. Jour-
study. Child Development, 62, 190 –198. nal of Family Psychology, 19, 111–120.
Mead, G. H. (1934). Mind, self, and society from the standpoint of a social Whitbeck, L. B., & Gecas, V. (1988). Value attributions and value trans-
behaviorist. Chicago: University of Chicago Press. mission between parents and children. Journal of Marriage and the
Meyer, G. J., Finn, S. E., Eyde, L. D., Kay, G., Moreland, K. L., Dies, Family, 50, 829 – 840.
R. R., et al. (2001). Psychological testing and psychological assess- Wilcox, R. R. (1997). Introduction to robust estimation and hypothesis
ment: A review of evidence and issues. American Psychologist, 56, testing. San Diego, CA: Academic Press.
128 –165. Winch, R. F. (1962). Identification and its familial determinants. India-
Montemayor, R., & Brownlee, J. (1987). Fathers, mothers and adolescents: napolis, IN: Bobbs-Merrill.
Gender-based differences in parental roles during adolescence. Journal Zentner, M. (2005). Ideal mate personality concepts and compatibility in
of Youth and Adolescence, 16, 289 –302. close relationships: A longitudinal analysis. Journal of Personality and
Moretti, M., & Higgins, T. E. (1999). Own versus other standpoints in Social Psychology, 89, 242–256.
Appendix A
Q Sorts Used in the Present Research

Table A1
The Adjective Q Set Used in the Present Research to Describe Personality Traits
Absent-minded Adventurous, seeks new and Ambitious, wants to succeed
exciting experiences
Anxious Assertive, says what he/she thinks Authoritarian, dominant
Beautiful, attractive Calm, relaxed Charming
Competent, does well what he/she does Competitive Creative, imaginative
Critical, skeptical, not easily impressed Curious Early riser
Elicits disputes with others Emotional Empathic (easily feels what others feel)
Energetic, active Feels easily guilty Feminine/masculine
Forgives easily Forthcoming Generous, likes to share with others

Refined, has good taste Idealistic Impulsive, acts without thinking

Independent, autonomous Intelligent Irritable, easily angered
Jealous, envious Joyful, merry Lazy
Likable, popular Likes material comfort Likes to help others
Likes to show off Manipulative Moody
Needs approval from others Not sentimental, not easily moved Passionate
Persevering Physically in good shape Plans ahead
Prefers to spend his/her time in a Provocative, shameless Rarely indulges in anything
familiar environment
Reasonable Rebellious, oppositional Reliable, somebody one can count on
Restless Romantic Sad, melancholic
Sarcastic Secretive, mysterious Seductive, in quest of love
Self-confident Self-controlled Selfish
Has sense of humor Sensitive to music and/or poetry Sensitive to own physiological needs
Shy Sincere, frank Sociable
Strict, severe Stubborn Submissive
Takes all his/her time Talks more about himself/herself Tender, affectionate
than about others
Trusts others easily Undecided Understanding, is a good listener
Unsatisfied Well-organized Worried
Note. Items in this table are based on Block’s Q set for use with lay people (see J. Block, 1961/1978, Appendix H). See text for details
on the current adaptation.
Table A2
The Q Set Used to Describe Life Goals
Avoiding hard work Becoming a business executive

Becoming a community leader Becoming an authority on a special subject in my field
Becoming an outstanding athlete Being an accomplished musician
Being in a traditional marriage where the husband is the primary Being in good physical condition
wage earner and the wife primarily cares for the children
Being influential in public affairs Being married to someone who has a career in
addition to my own
Devoting attention to my spiritual life Earning enough to be comfortable, and no more
Feeling a real purpose in life Having a career
Having a high standard of living and wealth Having a high-status career
Having a job that is personally satisfying even if it does not Having a satisfying marriage/relationship
make me rich
Having an easy life Having an exciting lifestyle
Having an influential and prestigious occupation Having children
Having fun Having harmonious relationships with my parents and
my siblings
Having new and different experiences Helping others in need
Living in aesthetically pleasing surroundings Making my parents proud
Owning my own business Participating in religious activities
Preparing myself for graduate school Supporting artistic activities and the fine arts
Taking part in volunteer community and public service Working to promote the welfare of others
Note. Items in this table are based on the set of life goals proposed by Roberts and Robins (2000).
(Appendixes continue)
Appendix B
Illustrative Example of Permutation Analysis
Consider a table with one row for each child and one column for difference for all possible permutations of the rows (highlighted in
each mother (see Figure B1). The value in a cell is the similarity Figure B1, right side). The values’ diffmean for the six permuta-
index between the child (rows) and the mother (columns). The tions are 0.15, 0.025, 0, ⫺0.025, ⫺0.05, and ⫺0.1. The p value is
hypothesis can be restated as follows: Are the values on the the proportion of values ⱖ .15, that is, 1/6 ⫽ .17. All permutations
diagonal (similarities between true, or familial, dyads) larger than are computed, including the identity permutation, the original
the off-diagonal values (similarities between random dyads)? The sample. Its value being always equal to the target value, the p value
difference of means between diagonal and off-diagonal values is can never be smaller than 1 over the number of permutations. In
the averaged increase in correlation because of familial links. The this example, the p value can therefore not be smaller than 1/6 ⫽
permutation test computes this difference for the original table (see .17, and in the tests reported in the text, it cannot be smaller than
Figure B1, left side, diffmean ⫽ 0.15) and computes the same 1/100,000 ⫽ .00001.
Figure B1. Illustration of permutation analysis.
Received May 19, 2005

Revision received October 5, 2006
Accepted October 9, 2006 䡲

Origins of Adolescents' Ideal Self: An Intergenerational Perspective

Uploaded by

Copyright:

Available Formats

Origins of Adolescents' Ideal Self: An Intergenerational Perspective

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Origins of Adolescents' Ideal Self: An Intergenerational Perspective

Uploaded by

Copyright:

Available Formats

Journal of Personality and Social Psychology Copyright 2007 by the American Psychological Association

2007, Vol. 92, No. 3, 557–574 0022-3514/07/$12.00 DOI: 10.1037/0022-3514.92.3.557

Origins of Adolescents’ Ideal Self: An Intergenerational Perspective

Keywords: ideal self, internalization, intergenerational transmission, parenting practices, Q sort

assess the ideal self. More specifically, the Selves Question-

Results particular assumptions concerning statistical distributions (ex-

permutations are usually considered sufficient). As reported in

Table 1 (see under Permutation test), the differences in mean

In a second set of analyses, we examined the role of ideal-self

Effect size Effect size

Boys .68 .80 .63 .73

Girls .69 .80 .64 .76

Mother T1/ Child T1/ Father T1/ Child T1/

Accuracy of perception Acceptance Assimilation

F test for significance of R2 F ⫽ 2.94* F ⫽ 1.09 F ⫽ 4.53** F ⫽ 2.02† F ⫽ 2.21† F ⫽ 2.63*

Main effect: child gender Main effect: parent gender Interaction

self: transfer of parental ideal self to ideals for the children,

Aronfreed, J. (1968). Conduct and conscience. New York: Academic Press.

Q Sorts Used in the Present Research

Forgives easily Forthcoming Generous, likes to share with others

Refined, has good taste Idealistic Impulsive, acts without thinking

Avoiding hard work Becoming a business executive

Illustrative Example of Permutation Analysis

Figure B1. Illustration of permutation analysis.

Received May 19, 2005

You might also like