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
him- or herself, comprising the attributes that the person thinks he
or she actually possesses, the ideal self is a representation of the
attributes the person would like to have—an image of the person
to be ideally. The ideal self is recognized as important for at least
two reasons. First, the ideal self is a motivator that functions as an
incentive for future behavior, a self “to be approached or avoided”
(Cross & Markus, 1991, p. 232). Second, the ideal self is an
evaluator of the actual self. This view can be traced to James
(1890), who proposed that self-evaluations depend on the degree to
which the self’s actual successes coincide with the self’s aspira-
tions. More recent views have also stressed the importance of
knowing specific self-values and their weightings to understand
self-esteem (e.g., J. Block & Robins, 1993; Harter, 1999). In
addition, building on the ideas of Horney (1950) and Rogers
(1961), numerous studies have provided compelling evidence for
the role of discrepancies between ideal and actual views of the self
in accounting for emotional distress, in particular, depressive dis-
orders (e.g., Moretti & Higgins, 1999; Ogilvie & Clark, 1992;
Tangney, Niedenthal, Covert, & Hill-Barlow, 1998). Considering
the attention that has been given to the ideal self as an ingredient
of self-esteem, emotional distress, and the self-system, surprisingly
little is known about its developmental origins. Do children differ
in their ideal self? If so, how do their unique ideals for the self
come about? Why does John aspire to be primarily successful with
girls, whereas Paul aspires to be successful in his studies? Why
does Mary aspire to be assertive and independent, whereas Ellen
places much more importance on being altruistic? In the present
Marcel Zentner and Olivier Renaud, Department of Psychology, Uni-
versity of Geneva, Geneva, Switzerland.
This research was supported by a grant from the Jacobs Foundation. We
thank Sylvette Delaloye, Emmanuelle Dupont, Florence Dokic, Sabrina
Cappuccio, and Laure Bezençon for assisting in these studies. Our thanks
extend to Tatjana Aue for insightful comments on versions of this article.
Correspondence concerning this article should be addressed to Marcel
Zentner, Department of Psychology (FAPSE), University of Geneva, 40,
Bd. du Pont d’Arve, CH-1205, Geneva, Switzerland. E-mail:
marcel.zentner@pse.unige.ch
research, we propose to address these questions by adopting an
intergenerational approach. More specifically, we analyzed an
intergenerational arc that leads from the parents’ ideal self to the
ideals that parents hold for their children to the ideals that children
believe their parents hold for them to the children’s own ideal self.
Ideal-Self Transmission Processes
The transmission of values, ideals, or standards across genera-
tions has often been described in terms of internalization. The
concept of internalization is generally used to refer to a process
whereby parental and societal standards are gradually integrated
into the child’s self-system. Internalization involves the transfor-
mation of a subject’s “real or imagined characteristics of his
environment into inner regulations and characteristics” (Schafer,
1968, p. 188; see also Grolnick, Deci, & Ryan, 1997). Internal-
ization concepts have been propounded in theories such as sym-
bolic interactionism (Cooley, 1902; Mead, 1934), psychoanalysis
(Freud, 1923/1961; Sandler & Rosenblatt, 1962), and social learn-
ing (Aronfreed, 1968; Bandura, 1971). However, as noted by
Cashmore and Goodnow (1985), these theoretical propositions
have often lacked empirical legitimation. More recently, there has
been a reexamination of conceptualizations of internalization,
which have led to new, empirically more accessible formulations.
One influential theory suggests that intergenerational similarity
results from a two-step process of internalization:
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
through inaccurate perception or through rejection. (Grusec & Good-
now, 1994, p. 4)
A number of studies have provided support for this model, indi-
cating that perceptual accuracy and acceptance account for inter-
generational similarity in attitudes (Cashmore & Goodnow, 1985)
and parental values (Knafo & Schwartz, in press).
Although no research to date has examined intergenerational
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
example, research conducted within self-discrepancy theory (Hig-
gins, 1987) has provided evidence for similarity between adoles-
cents’ ideal self and their perceptions of parental ideals. Adoles-
cents were chosen as target population because stable ideal-self
representations are not believed to emerge before adolescence. The
same holds for the ability to consider multiple perspectives on the
self, including the ideals that significant others hold for oneself
(e.g., Keating, 2004; Moretti & Higgins, 1999). Adolescents were
asked to spontaneously generate sets of up to 10 attributes that
described their ideal self and the ideals the adolescents believed
their parents held for them. The results showed an appreciable
match between children’s ideal self and the ideals that they be-
lieved their parents held for them. The extent of match, defined as
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
percentage overlap, ranged from 25% in adolescents to 44% in
young adults (e.g., Moretti & Higgins, 1999). These findings are
consistent with the notion of acceptance as defined by Grusec and
Goodnow (1994) and provide an initial step in the analysis of
intergenerational relationships in the ideal self.
Aims of the Present Investigation
The overarching purpose of this prospective, longitudinal study
was to provide a more extensive examination of family relations in
the ideal self, one that goes beyond the perceptions of the adoles-
cent and also includes the parents’ actual ideals for themselves and
for their children. Within this general scope, particular emphasis
was placed on identifying processes and factors that would account
for parent– child similarity in the ideal self. To this end, elabora-
tions along five lines were required.
First of all, it is necessary to examine whether parent– child
similarity in ideal self actually exists. Even though the answer to
this question may seem trivial, it is not. Many parents like their
children to be intelligent, conscientious, sociable, self-confident,
and so forth, and many parents and children hold these ideals for
themselves. Thus, any profile-based similarities between parent
and child ideals may simply reflect shared societal ideals. Indeed,
previous work found the base-rate similarity in self-ideals to be
high, in the range of .7 (e.g., Harter, 1999; Klohnen & Mendel-
sohn, 1998). Hence, two ideal-self components need to be distin-
guished: a prototypical component, which reflects consensually
shared ideal-self representations in a culture, and a unique com-
ponent, which refers to the distinctive or unique ideals of an
individual. If, as expected, ideals were heavily shared within a
given culture, a first question to be clarified is the extent to which
transmission of unique ideals within families occurs. Although
internalization is occasionally understood as a process relating to
parent– child transmission of unique parental values (e.g., Rohan
& Zanna, 1996), in other contexts, it is seen as a “developmental
process in which children (as well as adolescents and adults)
progressively integrate societal values and proscriptions into a
coherent sense of self” (Grolnick et al., 1997, p. 136). In our view,
internalization encompasses both perspectives. However, these
perspectives need to be clearly distinguished both conceptually
and empirically.
Second, if unique parent– child similarity exists, how does
ideal-self transmission within the family occur? We propose
that this question can be successfully studied from an extension
of the two-step model of internalization mentioned above. Ac-
cording to this model, the similarity between the ideals that
parents want their child to endorse and the child’s ideal self
should be accounted for by (a) accurate perceptions of the
parental ideal and (b) acceptance of the perceived parental
ideal. Grusec and Goodnow (1994) referred to this similarity as
internalization. However, the mere similarity between parental
and children’s viewpoints does not automatically imply inter-
nalization in the traditional sense of a progressive intake of a
parental position by the child. Thus, we prefer to use the more
neutral term assimilation to denote the extent to which a child’s
ideals are similar to the ideals that parents want their child to
endorse. This part of the model is strictly analogous to the
two-step model (see Figure 1, right part). To account for
parent– child congruence in the ideal self, we added an addi-
tional link reflecting the relationship between parents’ ideal self
and their child-related ideals (see Figure 1, left part). We call
this link (parental) ideal-self transposition as a way of describ-
ing that the ideals that parents hold for themselves are, even-
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-
dance between parents’ own values and the values they hold for
their children (Knafo & Schwartz, in press; Whitbeck & Gecas,
1988). Our proposition, then, is that parent– child congruence in
the ideal self should be accounted for by two main component
processes: (a) ideal-self transposition and (b) ideal-self assim-
ilation, the latter reflecting both acceptance and accuracy (see
Figure 1).
Third, we took a closer look at issues of directionality in the
transmission of ideals between parents and children. There is a
broad consensus today that processes of socialization are best
conceptualized in transactional terms (e.g., Bugental & Grusec,
2006; Kuczynski, 2003). Although the above-mentioned lines of
theorizing often conceptualized transmission of parental beliefs
and values as going from parent to child, the possibility that
parents may be receptive to their children’s ideals should also be
considered. For example, a mother may wish her child to endorse
artistic ideals (such as becoming a performing artist), and this wish
may be taken over by the child. However, some mothers may
develop artistic ideals for their child as a reaction to their child’s
artistic motivations. One important parameter to consider in this
context is the consistency of a child’s ideal self. As long as a
child’s ideal self is inconsistent, lasting child-to-parent effects are
unlikely. However, as the ideal self matures and stabilizes, influ-
ences between parent and child ideals may go both ways. Little is
known about ideal-self stability during adolescence, but some
indirect evidence comes from research on values. Although values
and ideal selves are distinct constructs, in a broad sense, they can
both be characterized as representations of desired hypothetical
end states. Values begin to exhibit moderate levels of stability only
from age 13 (Whitbeck & Gecas, 1988), and they appear to
become increasingly consistent thereafter (Rokeach, 1973). These
results are consistent with Piaget’s notion that the capacity to
reason in terms of possibilities and hypotheticals does not emerge
before adolescence (Inhelder & Piaget, 1958; see also Keating,
2004). In accordance with these considerations, we addressed
issues of directionality with an eye on growth in ideal-self consis-
tency.
Fourth, to better understand the processes involved in trans-
mission, we sought to identify factors that affect the extent of
 ORIGINS OF THE IDEAL SELF 559
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

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,
both theoretical and empirical, to assume that congruence in
parent– child ideals may be related to parenting practices and to
gender. For example, according to Knafo and Schwartz (2003),
parenting affects the accuracy with which parental values are
perceived by children in three ways: It affects the availability of
the parental message, the understandability of the message, and
the child’s motivation to attend to the parental message. More
specifically, these investigators predicted that the parenting
dimension of warmth should contribute to accurate perception,
whereas “autocratic parenting undermines accuracy” (Knafo &
Schwartz, 2003, p. 598). Overall, their prediction was sup-
ported. Parenting is also likely to play an important role in
acceptance for at least two reasons. First, children tend to have
more positive views of their parents if the parents use respon-
sive, supportive, and affectionate parenting strategies (e.g.,
Rohan & Zanna, 1996). A positive parental image, in turn,
should increase the child’s receptivity to accepting parental
ideals. Second, warm and authoritarian parents likely have
different ways of expressing their ideals. The former may be
more inclined to use induction when expressing ideals as if they
were recommendations that need to be openly discussed or
negotiated with the child. A long-standing view in socialization
theory holds that parents’ use of inductive reasoning promotes
internalization of their norms for behavior, especially prosocial
behavior (e.g., Hoffman, 1970). Authoritarian parents, in con-
trast, may be more likely to express their child-related ideals as
if these were obligations. In doing so, these parents risk induc-
ing psychological reactance in their children (Brehm & Brehm,
1981). Consistent with this proposition, it has been found that
authoritarian parenting is associated with dissimilarity in
parent– child value profiles (Balswick & Macrides, 1975; Ro-
han & Zanna). Thus, we anticipated that supportive parenting
would be positively related to assimilation and its components
accuracy and acceptance, whereas restrictive parenting would
be negatively related to them.
A role for gender is suggested by a substantial literature
indicating that boys and girls tend to be socialized in different
ways. A common theme running through this literature is that
girls are encouraged to attend to intimate others’ needs, con-
form to their expectations, and evaluate themselves in terms of
acceptance by their intimates. In contrast, boys are encouraged
to develop an identity that is more distinct from the desires and
wishes of their intimates (see, e.g., J. H. Block, 1983; Cross &
Madson, 1997; Pomerantz, Saxon, & Kenney, 2001, for re-
views). Consistent with this gender pattern in socialization,
adolescent girls score higher on measures of perspective taking,
that is, the tendency to adopt the point of view of others
(Eisenberg, Carlo, Murphy, & van Court, 1995; Eisenberg,
Miller, Shell, McNalley, & Shea, 1991). From this observation,
one would expect girls to perceive parental viewpoints such as
values more accurately than do boys, and indeed, they do (e.g.,
Acock & Bengtson, 1980; Knafo & Schwartz, 2003). In addi-
tion, because conformity with parental expectancies is more
encouraged in girls than in boys, one should also find more
endorsement of parental ideals by girls compared with boys. In
addition to a main effect for child gender, effects of parent
gender had to be considered. For example, mothers spend about
twice as much time with their adolescent children as fathers do
(Montemayor & Brownlee, 1987). This should give maternal
ideals a greater chance than paternal ideals to be perceived
accurately. A final point regarding gender differences can be
 560 ZENTNER AND RENAUD
traced to early theories of internalization and identification,
according to which behavior and value transmission is primarily
from parents to their same-sex rather than to opposite-sex
offspring (e.g., Kohlberg, 1966; Winch, 1962). However, be-
cause findings regarding effects of the gender composition of
the parent– child dyad have been mixed and inconsistent (e.g.,
Troll & Bengtson, 1979), we did not formulate hypotheses
regarding interaction effects of parent and child gender.
Fifth, the present study introduces some methodologically
distinctive features in comparison with earlier work. How to
measure the ideal self has been a matter of controversy in prior
research. Proponents of self-discrepancy theory, for example,
have strongly advocated using an idiographic procedure to
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
assess the ideal self. More specifically, the Selves Question-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
naire was introduced as a means of capturing the ideal self and
perceived parental ideals (Moretti & Higgins, 1999). This in-
strument elicits up to 10 participant-generated self-descriptions
to tap dimensions that may be uniquely important and accessi-
ble to the individual. However, some authors have criticized
this approach because it remains unclear whether the partici-
pant’s choices of attributes are due to their salience and acces-
sibility or simply to limitations in the participant’s working
vocabulary (Tangney et al., 1998). Open-response formats also
restrict the range of statistical techniques that can be used for
fine-grained quantification. At the opposite end of purely idio-
graphic assessments is the conventional variable-centered ap-
proach, which has been used in the great majority of studies on
parent– child resemblance (see Loehlin, 2005, for a review of
this literature). For example, in recent studies on the transmis-
sion of values, the child’s importance rating for a certain value
is the dependent variable, and the parents’ score on the same
value is the predictor (e.g., Albert & Trommsdorff, 2003;
Schönpflug, 2001; see also Larson & Almeida, 1999, for a
similar approach). An important limitation of this approach is
that it is confined to an analysis of parent– child congruence on
single dimensions computed across persons one at a time (e.g.,
collectivism, extrinsic motivation, sociability). In contrast, con-
gruence assessed via the person-centered approach is based on
profile similarities that capture the congruence of the child–
parent dyad across responses on an overarching individual-
differences domain (such as life goals, personality, attitudes). In
other words, by using a person-centered approach, it is possible
to compare a child and his or her parent in terms of a system of
interrelated attributes rather than in terms of single variables,
each considered separately. A second limitation is that a
variable-centered correlation or regression is computed between
the parents’ and the child’s scores on a specific characteristic
across all participants and thus characterizes the sample rather
than the dyad (see Luo & Klohnen, 2005, for an extensive
analysis of the limitations of the variable-centered approach).
To overcome these limitations, we used the Q sort as a
time-honored method of person-centered assessment. The Q
sort captures the unique ordering of multiple attributes within
the individual being described rather than scores on single
traits. Thus, this method is suited to capturing the intergenera-
tional concordances in systems of ideals. In addition, the Q sort
limits responses that result from a social desirability bias, which
is particularly important when many options are desirable, as
with ideals. Most important, because of response sets, the
anchor points of standard rating scales can have nonequivalent
meanings across participants. As noted by Cronbach (1955)
long ago, this may lead to distorted similarity estimates. In
contrast, the Q sort’s forced-rating format ensures commensu-
rability in the response profiles provided by different individ-
uals (e.g., J. Block, 1971; Ozer, 1993). In the present research,
we used a Q sort based on a comprehensive list of adjectives to
assess ideals about personality traits and life aspirations. Ado-
lescents were selected as target population for the reasons
mentioned earlier, that is, their capacity to form ideal-self
representations and their skills in considering multiple perspec-
tives on the self. We chose to include life aspirations in addition
to personality ideals because of the particular relevance that
they acquire during late adolescence, when consequential life
decisions are being prepared.
Method
Participants
Families were recruited via letters sent to families with one child
in the age range between 15 and 19 years. The addresses were
obtained through the Department of Public Instruction of the city
of Geneva, Switzerland. Families who were interested were asked
to return a self-addressed, prestamped postcard. Each family who
returned a card was then contacted by project staff to determine
eligibility for enrollment in the study. The sample at Time 1
consisted of 120 adolescents (M ⫽ 17.5 years, range 15.0 –19.8
years) and their parents, 115 mothers and 68 fathers (M ⫽ 50.0
years, range 38 –71 years). Of the adolescents, 69% were girls
and 31% boys. Over 90% of the original sample, 109 adoles-
cents and their parents (105 mothers, 62 fathers), participated in
the assessments at Time 2, which took place approximately 7
months later.
Of the adolescents in the sample, 69% were high school stu-
dents, 11% were in a professional training school, and 8% were in
college. Thirty percent of the adolescents came from single-parent
homes in which the adolescent lived with his or her mother in the
vast majority of cases, and 70% of the adolescents were living with
both parents. This distribution is relatively representative of the
entire population of Geneva (21% and 79%; population reference
values provided by OCSTAT, 2002). Although, in the biparental
category, both parents were encouraged to participate, in 20% of
the cases, the father could not participate. The vast majority of
participants were of European descent (95%) and French mother
tongue (80%), which approximates the reference values of the city
population (92% and 70%, respectively). To report family income,
participants were presented with a series of sequential income
ranges from which to choose, resulting in the following distribu-
tion: 18%, 40,000 –70,000; 33%, 70,000 –100,000; 33%, 100,000 –
150,000; 9%, 150,000 –200,000; and 7%, above 200,000 Swiss
francs. (In the current period, 1 dollar is equal to 1.3 Swiss francs,
and 1 euro corresponds to 1.5 Swiss francs). In comparison, the
median family income of the city population is between 90,000
and 100,000 Swiss francs. With respect to educational background
of the parents, 42% held an advanced high school degree, 14% had
obtained a degree from an advanced professional training school,
and 38% held a university degree. In the city population, the
respective percentages are 33%, 13%, and 23%. Hence, our sample
was biased toward higher education levels, but it was relatively
 ORIGINS OF THE IDEAL SELF
representative for ethnicity, language, family composition, and
income.
Procedure and Measures
Time 1. At Time 1, families were invited to two sessions, both
held at the University of Geneva. Separated by a minimum of 4 days,
each session lasted about 2 hr, although the time needed to complete
the procedures varied across participants. The participants were tested
in groups of 4 –20. Two research assistants were always present to
give an introductory explanation of the tasks, monitor the sessions,
and answer questions while the participants completed the tasks. After
the initial introduction, parents and adolescents were placed in differ-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
ent rooms to prevent communication or exchange of information
This document is copyrighted by the American Psychological Association or one of its allied publishers.
between family members. Across the two sessions, adolescents were
instructed to complete six different Q sorts: ideal self, perception of
maternal ideals, and perception of paternal ideals, each being done for
two content domains, personality traits and life goals. Parents, in turn,
were asked to complete four different Q sorts: ideal self and the ideals
they held for their child, each Q sort being done for personality traits
and life goals. To avoid carryover, parents provided their self-related
Q sorts in one session and their child-related Q sorts in the other.
Adolescents provided their ideal-self Q sort and one of the two
perceived parental ideals Q sorts in one session and the second Q sort
relating to perceived parental ideals in the second session.
The personality Q sort was based on Block’s Q set for use
with lay people (see J. Block, 1961/1978, Appendix H). It
contains 70 trait-descriptive items, which are unambiguous and
easy to understand. For the present version, items that are of
particular significance in an adolescent population were added.
In addition, some of the original items were replaced for the set
to approximate the item pool of the most current model of
personality, the five-factor model. These changes produced 78
items, which are listed in Table A1 of Appendix A. The life
aspiration Q sort was based on the comprehensive set of 38 life
goals recently proposed by Roberts and Robins (2000). The
authors demonstrated that this set of goals is not redundant with
personality, as defined by the five-factor model. The items are
listed in Table A2 of Appendix A.
In the ideal-self Q sort, parents and children were instructed
to sort the cards according to the person they would like to be
or become ideally. With this instruction in mind, they were
asked to sort the 78 personality attributes into nine categories
ranging from 1 (least desired) to 9 (most desired). The partic-
ipants were required to sort the cards according to a predeter-
mined discrimination-maximizing distribution (i.e., 14 items
into the middle category and 8 items in the remaining eight
categories). The same procedure was then repeated for the 38
life goals, which, because of their lesser number, had to be
placed into seven categories (i.e., eight items in the middle
category and five in the remaining six categories; see J. Block,
1961/1978, for details on Q-sort formatting). To describe their
child-directed ideals for personality, parents were instructed to
let themselves dream and to imagine the ideal future character-
istics of their child. We decided to ask parents about their ideal
future child because pilot work suggested that rating an ideal
present child would induce excessive comparisons with the
current child. From this general instruction, parents were asked
561
to sort the cards, as described, into piles ranging from “those
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-
mulated in analogous terms ranging from “life goals you wish
your child would least (1) to those you wish your child would
most (7) pursue in his/her future life.” Finally, for perceptions
of parent ideals, adolescents were asked to sort the cards
ranging from traits/life goals “my mother/father would like me
to have the least to those my mother/father would like me to
have the most.”
Parenting dimensions were assessed with the revised Block
Child Rearing Practices Report (CRPR; Rickel & Biasatti, 1982),
administered to both parents and adolescents after completion of
the Q sorts. The revised CRPR includes two dimensions, nurtur-
ance and restrictiveness. Nurturance relates to flexible and warm
child-rearing practices, such as encouraging independence and
showing support and affection (e.g., “I express affection for my
child, for example, by holding and hugging my child”; “I encour-
age my child to be independent from me”; “I encourage my child
to be curious, to explore and to question things”). Restrictiveness
relates to parental attitudes such as expecting respect, protective-
ness, restriction of showing feelings, and belief in physical pun-
ishment (e.g., “I do not allow my child to get angry with me,” “I
teach my child to keep control of his/her feelings at all times,” “I
do not allow my child to question my decisions”). Dekovic,
Janssens, and Gerris (1991) found self-reports based on this in-
strument to be significantly related to observations of actual pa-
rental behaviors. On the basis of the original report, we used the 19
most discriminating items, 8 for nurturance and 11 for restrictive-
ness (see Rickel & Biasatti, 1982). For mother reports, coefficient
alpha was .76 for nurturing and .77 for restrictiveness. In the child
reports of maternal parenting, the respective alphas were .79 and
.77. For father reports, coefficient alpha was .68 for nurturing and
.79 for restrictiveness. The respective alphas for child reports of
father parenting were .79 and .77. To increase the reliability of the
parenting measure, we computed a composite score for each parent
for each parenting style from the two adolescent ratings and the
parent’s self-rating.
Time 2. Approximately 7 months after the initial assessments,
families were invited to the university to participate in the follow-
up. The purpose of the assessments at Time 2 was fourfold: first,
to assess the ideal-self stability between Time 1 and Time 2;
second, to examine whether the concurrent findings would also be
evident longitudinally; third, to run cross-lagged analyses between
parents’ and children’s ideals; and finally, to gather additional data
that were unrelated to the purposes of this study. So as not to
overtax the attentional capacities of our participants, only a selec-
tion of the tasks administered at Time 1 was repeated at Time 2.
Specifically, adolescents were asked to complete two Q sorts, one
for their ideal self and one for perceived maternal ideals— both
relating to life goals only. Both parents were asked to complete
two life-goal Q sorts, one regarding the ideals they held for their
child and one relating to their ideal self. Thus, in comparison to
Time 1, analyses at Time 2 were restricted to life aspirations. In
addition, because perception of the parental ideals was collected
only for mothers at Time 2, perceptual accuracy and acceptance
could be assessed only for mother– child dyads.
 562 ZENTNER AND RENAUD
Results
Data Analytic Strategy
Units of analysis were Q correlations, that is, profile-based
similarity scores computed within dyads. For example, the con-
gruence index between the parents’ ideal-self template and their
ideal-child template was derived by computing the correlation
between the numerical value assigned to each item in the parental
ideal-self Q sort and the value assigned to each item in the parent’s
ideal-child Q sort. This index quantitatively expresses the degree
to which the parent’s ideal self is congruent with the ideals that the
parent holds for his or her child (J. Block, 1961/1978; Klohnen &
Mendelsohn, 1998; Zentner, 2005). Using this similarity index, the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
subsequent analyses exemplify how permutation analysis can be
This document is copyrighted by the American Psychological Association or one of its allied publishers.
used for distinguishing true from random similarity between re-
lated individuals.
Aside from the conventional analyses based on Gaussianity
and least squares (regressions, t tests, analyses of variance
[ANOVAs]), we also ran the equivalent robust analyses (In-
sightful Corporation, 2002; Wilcox, 1997). Robust methods are
designed to give very similar results to classical methods when
the observations are (almost) Gaussian, but they are not influ-
enced by outlying observations, that is, observations that are far
from the bulk of the data and that are probably not representa-
tive of the general tendency of the observed phenomenon. In the
subsequent analyses, results of the robust analyses were very
close to those of the conventional analyses, strengthening con-
fidence in the findings and their generalizability. Because of
this overlap, we report conventional analyses only.
Dissociating Familial From Random Parent–Child
Similarity
A primary goal of this study was to examine the extent to
which familial parent– child concordance in ideals exceeded
random parent– child concordance, that is, concordance based
on unrelated, randomly combined parent– child dyads. Familial
versus random similarity was dissociated for all types of
parent– child concordance in ideals in the following order:
congruence, transposition, assimilation, accuracy, and accep-
tance. In all cases, the procedure was the same: We first
computed familial parent– child similarity for the actual fami-
lies. For example, with respect to congruence, we computed the
correlations between each parent’s ideal-self Q sort and his or
her own child’s ideal-self Q sort. We then computed the random
similarity by correlating each parent’s ideal-self Q sort with all
of the other children’s ideal-self Q sorts in the sample.1 This
yielded two distributions, a familial and a random distribution,
as is illustrated in Figure 2 for transposition, accuracy, and
acceptance. The distributions are represented with kernel den-
sity estimators, which have some advantages over histograms.2
The high random values indicated that ideals are heavily shared
within a culture; thus, henceforth, we refer to random similarity
also as culture-level similarity.
An ideal statistical tool to determine whether the difference
between the familial and cultural distributions is statistically
significant is the permutation test (e.g., Efron & Tibshirani,
1993; Hubert, 1984). Because permutation tests require no
particular assumptions concerning statistical distributions (ex-
cept for the assumption of independent observations), they are
increasingly applied even in the context of traditional statistical
tests such as correlations, t tests, and ANOVAs (Good, 1994;
May & Hunter, 1993). In the current context, we used the
permutation test to examine whether, for each of the thousands
of possible random combinations (permutations), the test sta-
tistic was equal to the statistic based on the true sample. A
primer on permutation analysis, including a simple example
that illustrates how the test was applied in the current context,
is provided in Appendix B. To determine whether the difference
between each of the distributions was statistically significant,
we used 100,000 permutations to maximize precision (1,000
permutations are usually considered sufficient). As reported in
Table 1 (see under Permutation test), the differences in mean
between familial (diagonal) similarities and random (off-
diagonal) similarities were consistently significant for mothers
and fathers in both content domains, personality and life goals.
Supplementary analyses confirmed these results.3
To compare the findings from the permutation test with
results obtained through a more conventional procedure, we
also ran one-sample t tests comparing familial similarities with
the average of the random similarities. It should be noted that,
in contrast to the permutation test, the t test requires that the
value with which the sample is compared not depend on the
sample. Even though this condition for the t test was not met,
the findings obtained from t tests were similar to the findings
obtained through the permutation test. The complete t values
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
dyad (mother– child, father– child), with the type of ideal (ideal self of the
adolescent, perceived parental ideals, actual parental ideals, etc.), and
because a few participants did not provide all of the Q sorts. This yielded
the following totals: N ⫽ 109 –115 familial mother– child dyads and N ⫽
11,772–13,110 random mother– child pairs; N ⫽ 62– 68 familial father–
child dyads and N ⫽ 3,782– 4,556 random father– child pairs.
2
Kernel density estimation is similar to plotting the data in a histogram and
then smoothing the histogram boxes to a continuous distribution, but it offers
some advantages over histograms. First, the kernel diagrams provide the
possibility of superimposing curves for the random and familial similarities
within the same graph and of comparing them precisely. Second, the shape of
the histogram depends on the starting point of the boxes, whereas this is not the
case for the kernel estimation. Third, histograms lead to loss of information
because observations are replaced by the central point of the interval in which
they fall (see Silverman, 1986, for details).
3
To examine the robustness of the present findings, we ran the same
tests with two other similarity coefficients: (a) the sum over all items of the
absolute difference (sometimes called the L1 distance) and (b) the sum
over all items of the squared difference (often called the D2 distance).
Moreover, in addition to differences in mean as exemplified in Appendix
B, we also used two supplementary nonparametric tests to examine the
difference in the familial versus random distributions: the rank statistic and
the Kolmogorov-Smirnov test. Whereas the rank statistic is not influenced
by outliers or by nonnormality, the Kolmogorov-Smirnov statistic tests the
maximal distance between the distributions. Thus, for each level (congru-
ence, transposition, assimilation, etc.), we ran nine tests (using each of the
three similarity indices and each of the three test statistics) for four cases,
personality and life goal and mothers and fathers, yielding 36 values for
each of the five levels. Using the alternative similarity indices and statistics
yielded results that are very similar to those reported in Table 1.
 ORIGINS OF THE IDEAL SELF 563
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

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).
 564 ZENTNER AND RENAUD
are reported in Table 1. The last two columns in Table 1
supplement the significance values with two effect-size esti-
mates, d and r.4 As can be seen, the extent of the familial versus
random difference varied appreciably across the various levels
of concordance: As should be the case according to the current
model, values for the intermediate processes, those supposed to
account for the indirect links (transposition, accuracy, and
acceptance), were higher than the values for the two indirect
links (congruence and assimilation). Interestingly, effect sizes
were larger for ideals relating to life goals than for those
relating to personality traits.
Pathways From Parents’ to Children’s Ideal Self
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Person-centered regressions. Recall that previous research
suggested that assimilation of parental viewpoints is best described
in terms of a two-step model involving accurate perception and
acceptance of the parental position. Thus, in a first set of analyses,
we examined the role of perceptual accuracy (similarity between
parents’ actual ideals for their child and the child’s perception of
these ideals) and acceptance (similarity between the child’s per-
ception of parental ideals and the child’s ideal self) in accounting
for assimilation (direct similarity between parents’ ideals for their
children and the child’s ideal self). Concurrent analyses were run
for both content domains, life goals and personality, whereas
longitudinal analyses were restricted to life goals (see Method
section, above). As has been exemplified in research on internal-
ization of values (Knafo & Schwartz, in press), in operational
terms, this aim can be translated into a main effect for accuracy in
predicting assimilation, a main effect for acceptance, and an in-
teraction of Accuracy ⫻ Acceptance. The interaction reflects the
contingency between accuracy and acceptance predicted by the
model. Thus, acceptance should be predictive of assimilation, but
only for those adolescents high in perceptual accuracy; for those
low in perceptual accuracy, the effect of acceptance should disap-
pear or at least diminish sharply. Multiple regressions tested the
independent and interactive effects of accuracy and acceptance on
assimilation, separately for mothers and fathers, at Times 1 and 2.
Accuracy of perception and acceptance scores were subjected to
Fisher’s r-to-z transformation and subsequently centered around
their mean (Aiken & West, 1991). Thereafter, the interaction term
was computed by multiplying accuracy by acceptance.
For personality characteristics, accuracy, acceptance, and their
interaction all contributed to the variance in assimilation of moth-
ers’ child-related ideals, accuracy: ␤ ⫽ .47, t ⫽ 7.12, p ⬍ .01;
acceptance: ␤ ⫽ .44, t ⫽ 6.71, p ⬍ .01; interaction: ␤ ⫽ .16, t ⫽
2.50, p ⫽ .01; total R2 ⫽ .48.5 Similarly, all three predictors also
contributed to the assimilation of fathers’ child-related ideals,
accuracy: ␤ ⫽ .62, t ⫽ 8.19, p ⬍ .01; acceptance: ␤ ⫽ .22, t ⫽
2.68, p ⫽ .01; interaction: ␤ ⫽ .13, t ⫽ 2.06, p ⫽ .04; R2 ⫽ .71.
For life goals, both accuracy and acceptance predicted assimilation
of maternal ideals, accuracy: ␤ ⫽ .61, t ⫽ 8.17, p ⬍ .01; accep-
tance: ␤ ⫽ .20, t ⫽ 2.63, p ⬍ .01. The interaction term failed to
reach significance, ␤ ⫽ .08, t ⫽ 1.10, p ⫽ .28; R2 ⫽ .48. A similar
pattern was found for fathers, accuracy: ␤ ⫽ .62, t ⫽ 6.71, p ⬍ .01;
acceptance: ␤ ⫽ .25, t ⫽ 2.81, p ⬍ .01; interaction: ␤ ⫽ .09, t ⫽
1.03, p ⫽ .31; R2 ⫽ .56. Time 1 accuracy and acceptance also
predicted assimilation of maternal life goals at Time 2, accuracy:
␤ ⫽ .41, t ⫽ 4.46, p ⬍ .01; acceptance: ␤ ⫽ .27, t ⫽ 2.78, p ⬍ .01;
interaction: ␤ ⫽ .05, t ⫽ .54, ns; R2 ⫽ .28. The predictive relations
also held for fathers, accuracy: ␤ ⫽ .36, t ⫽ 4.46, p ⬍ .01;
acceptance: ␤ ⫽ .27, t ⫽ 2.78, p ⬍ .01; interaction: ␤ ⫽ .19, t ⫽
1.16, p ⫽ .12; R2 ⫽ .33. In summary, both accuracy and accep-
tance accounted for a substantial variance in assimilation. The
hypothesized interaction effect emerged for the personality do-
main, but not the life-goal domain. Although this result could
indicate a domain-specific outcome, it is equally likely due to
power limitations, which notoriously restrict the chances of de-
tecting hypothesized interaction effects in family research (e.g.,
Whisman & McClelland, 2005). The interaction effect for the
personality domain can be seen in the top panel of Figure 3. It
illustrates that acceptance is strongly predictive of assimilation, but
only for adolescents high in perceptual accuracy.
In a second set of analyses, we examined the role of ideal-self
transposition (similarity between parents’ ideal self and the ideals
they hold for their children) and of assimilation (similarity be-
tween parents’ child-directed ideals and the child’s ideal self) and
their interaction in accounting for parent– child similarity in the
ideal self at Time 1 and, for life aspirations, also at Time 2. On the
basis of the previous findings, we used assimilation scores as
shorthand for the combined effects of accuracy and acceptance.
Using the identical computations as before, we found that, with
respect to ideals for personality, assimilation contributed signifi-
cantly to the variance in mother– child ideal-self similarity; in turn,
effects of transposition and of the Transposition ⫻ Assimilation
interaction did not reach significance, transposition: ␤ ⫽ .09, t ⫽
1.52, p ⫽ .13; assimilation: ␤ ⫽ .70, t ⫽ 11.15, p ⬍ .01;
interaction: ␤ ⫽ .06, t ⫽ 1.35, p ⫽ .17; R2 ⫽ .65. However, all
three predictors contributed to father– child ideal-self similarity,
transposition: ␤ ⫽ .44, t ⫽ 5.64, p ⬍ .01; assimilation: ␤ ⫽ .64,
t ⫽ 7.50, p ⬍ .01; interaction: ␤ ⫽ .16, t ⫽ 2.97, p ⬍ .01; R2 ⫽
.72. With respect to life aspirations, all three predictors contributed
to mother– child ideal-self similarity, transposition: ␤ ⫽ .21, t ⫽
3.15, p ⬍ .01; assimilation: ␤ ⫽ .64, t ⫽ 9.80, p ⬍ .01; interaction:
␤ ⫽ .14, t ⫽ 2.34, p ⫽ .02; R2 ⫽ .58. A similar pattern emerged
for father– child ideal-self similarity in life aspirations, transposi-
4
Because we could not find guidelines on how to compute effect sizes
for the permutation test, we relied on the t test for effect-size estimations.
This seems reasonable because the t test and the permutation test are both
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
of the familial similarities as denominators, because the familial similari-
ties are the reference values. The formula for the effect size is as follows:
[mean(familial) ⫺ mean(random)]/SD(familial). When computing the con-
fidence intervals (CIs) around d, we found a range of d equal to 0.15– 0.20
(or r ⫽ .07–.10) on each side of the value. Because the CI for d increased
by about 50% for the individual genders (because of the restricted N), we
refrained from reporting d and r values for girls and boys separately.
However, we tested whether the d values for girls and boys were signifi-
cantly different from each other and found no differences with the excep-
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
mothers ( p ⫽ .06). Computations of the significance of the differences in
d were based on a stratified procedure in which resampling was done
separately within the families of girls and boys (see Efron & Tibshirani,
1993; details regarding these computations are available from us).
5
In all current regression analyses, the degrees of freedom for the
denominator at Time 1 ranged from 102 to 105 for mothers and from 57 to
61 for fathers. At Time 2, degrees of freedom ranged from 95 to 99 for
mothers and from 57 to 58 for fathers.
 ORIGINS OF THE IDEAL SELF 565

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
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Boys .68 .80 .63 .73

This document is copyrighted by the American Psychological Association or one of its allied publishers.

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,
p ⬍ .01; interaction: ␤ ⫽ .19, t ⫽ 1.97, p ⫽ .05; R2 ⫽ .44. As is
illustrated in the bottom panel of Figure 3, the relatively consistent
interaction effect indicates that the effects of assimilation on
parent– child congruence in the ideal self were boosted by high
levels of transposition. Initial assimilation, but not initial transpo-
sition, predicted congruence of mother– child ideal-self similarity
at Time 2, transposition: ␤ ⫽ ⫺.04, t ⫽ ⫺.46, ns; assimilation:
␤ ⫽ .49, t ⫽ 4.88, p ⬍ .01; interaction: ␤ ⫽ ⫺.016, t ⫽ ⫺.16, ns;
R2 ⫽ .27. The longitudinal findings for fathers were similar,
transposition: ␤ ⫽ .05, t ⫽ .37, p ⬍ .01; assimilation: ␤ ⫽ .50, t ⫽
3.76, p ⬍ .01; interaction: ␤ ⫽ ⫺.20, t ⫽ ⫺1.60, p ⫽ .12; R2 ⫽
.31. In summary, most of the predictions for the congruence arc
were supported concurrently. Longitudinally, only assimilation
contributed significantly to congruence, whereas the effects of
transposition and the interaction did not replicate.6
Directionality analyses. The previous analyses examined po-
tential pathways of transmission, including interrelationships be-
tween processes as predicted by the model. However, these anal-
yses did not address the direction of effects between parent and
child within each single process. For this reason, we ran cross-
lagged analyses between children’s and parents’ ideals at Times 1
6
An alternative way to test the current model is via mediational analy-
ses. Because mediational analyses are variable centered and thus require
that the current life-goal and personality patterns be decomposed into
single variables to be considered one at a time, they were not ideal in the
present data context. In addition, several variables did not meet the re-
quirements for mediation as defined by Baron and Kenny (1986). Still, we
did run mediation analyses for variables in which conditions for mediation
analyses were met and found that the analyses supported the current model.
(These analyses are available from us upon request.)
 566 ZENTNER AND RENAUD
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

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,
in which dyadic similarity correlations are the main units of
analysis, this aim translated into comparing the average concor-
dances between the parent ideals at Time 1 and the child ideals at
Time 2 versus the average concordances between the child ideals
at Time 1 and the parent ideals at Time 2. The logic of cross-
lagged analysis implies that, if the former exceed the latter, the
directionality should be from parent to child, whereas the inverse
would suggest a child-to-parent effect. Recall that because, at
Time 2, data are restricted to the life-goal domain, these analyses
were run for life goals only. The results are reported in Table 2.
As can be seen in the first row regarding the parent– child
similarity in the ideal self (congruence), the mean of the (r-to-z
transformed) similarity correlations between the maternal ideal-
self T1 and the child’s ideal-self T2 was higher than the inverse,
that is, the mean of the similarity indices between the child’s
ideal-self T1 and the maternal ideal-self T2, suggesting a parent-
to-child effect for congruence. Findings for all the other levels are
reported in the following rows. Even though most effects were
either from parent to child or nonsignificant, a child-to-mother
effect emerged with respect to assimilation (see Table 2, third
row). As hypothesized in the introductory section, above, child-
to-parent effects should be more likely for older adolescents than
for younger ones (see also next section). To examine this hypoth-
esis, we recomputed the same analyses for a younger and an older
age group based on a median split. Although the results were
similar on the whole, it is important to note that the only child-
to-parent effect that could be identified (assimilation) was re-
stricted to the older adolescents, t(50) ⫽ ⫺2.40, p ⬍ .05. For the
younger ones, there was no child-to-parent effect, t(48) ⫽ ⫺1.09,
p ⬎ .25. In addition, the child-to-parent effect for perceptual
accuracy, though nonsignificant for the entire sample (see Table 2,
 ORIGINS OF THE IDEAL SELF 567

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
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

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.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

T1 ⫽ Time 1; T2 ⫽ Time 2.

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

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
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

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

This document is copyrighted by the American Psychological Association or one of its allied publishers.

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
gender with respect to the various concordances in parent– child
ideals, we used a mixed effects model ANOVA with family as a
random factor, child gender as a nesting fixed factor, and parent
gender as a crossed fixed factor, with an interaction between the
two fixed factors. (This analysis is similar to a repeated measures
ANOVA, with the advantage of allowing one to examine the
effects of child gender, parent gender, and their interaction based
on all available dyads. The traditional repeated measures ANOVA,
in contrast, restricts the sample to families in which both parents
participated, thereby eliminating all of the single-parent families.
For the sake of comparison, we also ran the latter analysis and
found that, on the whole, the results did not change.) The complete
results are reported in Table 4. A main effect of parent gender
indicates that children’s ideal self was more similar to their moth-
er’s than to their father’s ideal self. A main effect of child gender
indicates that, compared with boys’ ideal self, that of girls was
more similar to their parents’ ideal self, again with respect to both
domains, personality ideals and life aspirations. Interactions be-
tween both factors, which would indicate that the effect of child
gender was qualified by parent gender, were not found. In inter-
preting the findings, it is important to keep in mind that the extent
to which familial values exceeded random values did not differ
significantly between boys and girls (see footnote 4). Therefore,
higher values for girls cannot simply reflect a higher inclination to
adopt cultural ideals. Overall, the findings support the hypothesis
that girls tend to develop aspirations that are more similar to their
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.
 ORIGINS OF THE IDEAL SELF
Discussion
Summary of Key Findings
By investigating a broad range of potential parental sources of
adolescents’ ideal self and by tracing its development from mid- to
late adolescence, the current findings contribute to the literature in
several ways. First, our study provides the strongest evidence yet,
after controlling for culture-level similarity, that specific similarity
between parents’ and children’s ideal selves exists. Second, we
extended the two-step model of internalization to provide a frame
for understanding ideal-self transmission processes within the fam-
ily. Consistent with the proposed extension, we found that three
transmission steps account for parent– child similarity in the ideal
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
self: transfer of parental ideal self to ideals for the children,
This document is copyrighted by the American Psychological Association or one of its allied publishers.
children’s perception, and acceptance of these parental ideals.
Third, we found that most processes accounting for parent– child
resemblance in the ideal self are meaningfully related to parenting
practices. Although these findings indicate that parental ideals and
socialization practices contribute to resemblance between parent
and child ideals, the findings do not rule out the possibility of
child-to-parent effects. Indeed, tentative evidence suggests that
such reverse effects become more likely as the ideal self stabilizes
in late adolescence. Fourth, compared with boys’ ideal self, that of
girls was more similar to the ideals that parents held for their
children and for themselves. Finally and in a methodological
perspective, the current research exemplifies how permutation
analysis can be used for distinguishing unique from culture-level
similarity between related individuals. In the following sections,
we review the findings in more detail and consider their implica-
tions for the understanding of ideal-self development.
Ideal-Self Transmission: A Telephone Game Effect?
One of the key findings of this study is that, when the effects
of cultural similarity are statistically removed, the actual sim-
ilarity between parents’ and children’s ideal selves was slightly
above .20 for life goals and slightly below .10 for personality
traits. This finding is consistent with an extensive literature on
parent– child resemblance, indicating that parent– child correla-
tions are between .05 and .20 for personality traits and between
.15 and .30 for attitudes (see Loehlin, 2005). Contrary to folk
wisdom, then, the apple appears to fall far from the tree. At
first, this finding seems to question the notion of the family as
a major agency of transmission of ideals. However, one of the
distinctive aspects of this research is that it offers an explana-
tion for this generation gap. Our explanation is best described in
terms of a metaphor. In the telephone game, a message is passed
from one player to another but is progressively altered in
transit. In the context of our study, the message is the parental
ideal self, and the players are the three steps, transposition,
perceptual accuracy, and acceptance. In transposition, the mes-
sage is altered because not everything that parents wish for
themselves do they also wish for their children. Subsequently,
the message is altered again because what parents wish for their
children is perceived by the latter with imperfect accuracy. A
final transmutation of the message occurs because parental
ideals, even if accurately perceived, are not always accepted by
the child. To add quantitative substance to this metaphorical
reasoning, consider the transmission in ideals for mother– child
dyads regarding life goals (see Table 1). The multiplication of
569
the effect size r for transposition, accuracy, and acceptance led
to an r of .16 (.61 ⫻ .44 ⫻ .59), a value that is relatively close
to the effect size r ⫽ .21 for the direct similarity in mothers’
and children’s ideal self. This interpretation concurs with the
findings for additive and interactive effects of transposition,
accuracy, and acceptance in suggesting that, as predicted,
parent– child similarity in the ideal self is accounted for by
these component processes. Although these processes received
much support in the current research, it is also important to keep
in mind that an appreciable part of the variance could not be
explained by these processes, pointing to the need to consider
additional sources of parent– child congruence in future re-
search.
With the exception of the direct similarity in the ideal self
between parents and their children, the size of the effects for the
intergenerational similarities was substantial compared with the
size of effects typically found in the social sciences (Meyer et al.,
2001). With respect to the variations in effect sizes across all levels
and domains reported in Table 1, two points should be made. First,
effect sizes for transposition and acceptance were consistently
higher compared with those for congruence, assimilation, and
perceptual accuracy. One explanation is that, in the case of trans-
position and acceptance, responses to be correlated were derived
from the same person (the parent in the case of transposition and
the child in the case of acceptance). In contrast, correlations for the
latter three links were based on responses from different individ-
uals within the family. It is known that correlations based on
responses from the same informant tend to be inflated because the
systematic error variance is aligned (Meyer et al., 2001). Given
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
from different individuals. It should also be noted that because our
analyses are concerned with relationships between variables rather
than absolute magnitudes, the validity of all the subsequent anal-
yses and conclusions is not affected by this point. However, it is a
point to be kept in mind when interpreting the relative magnitude
of effect sizes across the various processes reported in Table 1.
Second and unexpectedly, effect sizes for parent– child concor-
dances in life aspirations were consistently higher than in person-
ality ideals. One explanation for this difference is that, during late
adolescence, parents and adolescents are sensitized to the impor-
tance of making the right life choices, prompting frequent parent–
child discussions about life goals. Consequently, the child is mo-
tivated to pay attention to parental messages in this domain, and in
turn, parental ideals about life goals are accessible to the child. In
contrast, parental ideals for personality may simply be less visible
and less important to the child than ideals for life goals. Accep-
tance of parental values, in turn, is linked to the legitimacy that
adolescents grant to parental authority in a given domain (Knafo &
Schwartz, in press; see also Smetana, 2000). Because of the more
private character of ideals regarding personality, adolescents may
consider their parents’ authority as less legitimate when parents try
to influence them regarding personality than when they try to
influence them with life plans. A second, altogether different
explanation relates to the higher interindividual variability in ide-
als for life goals compared with personality (see Table 1). As a
consequence, the extent to which familial similarity coefficients
can depart from random similarity is much more limited for traits
than for life goals, thereby restraining effect sizes. This explana-
 570 ZENTNER AND RENAUD
tion is conceptually related to the finding that observers are more
accurate when attributes are evaluatively neutral rather than highly
desirable or undesirable (Funder, 1995).
Conservative Consequences of Liberal Parenting
Consistent with our hypotheses, we found that the more parent-
ing was characterized by attributes such as warmth, support, and
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
endorse. Conversely, the more that parenting was characterized by
lack of nurturance and by attributes such as control, protectiveness,
and restriction of the child’s independence, the more the child’s
ideal self tended to deviate from the parents’ ideals. Paradoxically,
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
then, those parents who seemed inclined to accept or even to
actively promote their children’s aspirations to differ from their
own tended to have children whose aspirations were similar to
theirs. In contrast, parents who seemed least inclined to accept
deviations from their socialization ideals tended to have rebellious
children, that is, children whose aspirations were dissimilar to
theirs. It is important to note that we could also discern some
predictive relationships suggesting that parenting, in addition to
being a correlate of parent– child concordances in ideals, may be a
driving force behind changes in such concordances.
One explanation for this apparent paradox is that because
nurturant, supportive parenting is related to a positive attitude
of children toward the parents, such as perceptions of parental
trustworthiness, fairness, and lack of hypocrisy (Rohan &
Zanna, 1996), it should enhance the child’s motivation to attend
to and endorse parental ideals. Conversely, the less positive this
attitude is, the more likely the child’s motivation to differ from
his or her parents and to nourish his or her ideal self through
influences outside the family environment will be. A second
possibility is that restrictive parents may express their ideal
standards in ways that imply a higher degree of “oughtness.” As
is known from work on reactance theory, attempts to restrict an
individual’s freedom often produce a reactive boomerang ef-
fect, that is, an increase in the restricted behavior (Brehm &
Brehm, 1981). In the present context, this effect translates into
children of restrictive parents trying to defend their freedom by
rejecting those ideals that parents try to force upon them. This
interpretation is consistent with previous reports of a relation-
ship between authoritarian parenting and parent– child dissim-
ilarity (Rohan & Zanna, 1996).
In terms of gender differences, the present findings support
the hypothesis that boys tend to develop aspirations that are
more distinct from their parents’ aspirations than do girls. This
pattern held for both content domains, personality and life
aspirations. The extent to which this outcome can be attributed
to gender-specific patterns of socialization is a fascinating
question that deserves further scrutiny. An extensive literature
suggests that parents monitor the behavior of their daughters
more closely than the behavior of their sons (e.g., J. H. Block,
1983; Pomerantz et al., 2001). One immediate consequence of
this heightened parental monitoring of girls is that they tend to
spend more time in close proximity to the family than do boys,
who, in turn, tend to be more encouraged to explore the non-
familial environment. As a result, girls may orient more than
boys toward the family when developing ideal-self representa-
tions. Another possible manifestation of the heightened parental
supervision of girls is that parents convey their standards and
expectations more emphatically to their daughters than to their
sons (e.g., Pomerantz et al., 2001). As a result, girls may
become more invested in decoding parental ideals, leading to
higher perceptual accuracy of these ideals. Furthermore, en-
dorsement of these ideals should also be enhanced, considering
girls’ tendency to evaluate themselves in terms of acceptance by
their intimates.
Ideal-Self Development
A distinctive finding of the current research is that age predicts
ideal-self stability for adolescents until about age 17, after which
the ideal self levels off at traitlike levels of stability. In our view,
this suggests that, by the end of adolescence, the ideal self has
reached an important point of maturation. This finding ties up
nicely with the previously reported growth in the stability of values
across adolescence (Rokeach, 1973), and it is also in concert with
more recent work showing that, from age 18, life goals exhibit
traitlike levels of stability (Roberts, O’Donnell, & Robins, 2004).
Consistent with previous theorizing, these data concur in suggest-
ing that adolescence is a sensitive period for the development of
desirable hypothetical end states such as ideals, values, and life
goals (e.g., Inhelder & Piaget, 1958; Moretti & Higgins, 1999).
Thus, particular attention should be directed toward an understand-
ing of the influences impinging on this significant period of
ideal-self development.
The current research examined two aspects of parental influ-
ences, parenting practices and parents’ own ideals (for themselves
and for their children). In terms of the role of parenting, we found
some evidence suggesting that parenting practices may be not only
a correlate of concordances in parent– child ideals but also a
predictor of changes in concordances. Even though the short
period of the study makes it difficult to extrapolate to longer
periods of time, it is tempting to think about the role of parenting
within a broader developmental perspective. It is of particular
relevance to remember in this context that parents who are sup-
portive of their child during adolescence have, in most cases, also
been supportive before adolescence (e.g., McNally, Eisenberg, &
Harris, 1991). Hence, one way to think about parenting influences
over time is that, during childhood, a warm and supportive parent
is likely to have fostered a positive parental image in the child that,
by the time the child becomes capable of forming consistent
ideal-self representations, infuses his or her ideal self. Thus, the
association between parenting practices and parent– child concor-
dance in ideals, beyond reflecting a present state, may also reflect
a past history.
As for the direction of effects between parents’ and children’s
ideals, parent-to-child effects tended to be slightly more frequent
than child-to-parent effects. However, the presence of a couple of
child-to-parent effects reminds us of the importance of acknowl-
edging the possibility of bidirectional influences. More important,
reverse influences from child to parents were observed only for
adolescents who were 17 years or older. Although ideal-self sta-
bility is one likely reason for this outcome, an additional consid-
eration is that, as adolescents become young adults, their ideals
progressively transmute into tangible realities, such as moving to
a specific destination, entering a given profession or field of study,
or choosing particular longer term partners. At this juncture, it
becomes increasingly difficult for parents to overlook their chil-
 ORIGINS OF THE IDEAL SELF
dren’s aspirations, and as a consequence, parents may become
more receptive to their children’s ideals.
Another way of looking at the current findings is that they
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
less support children receive from their parents, the more they will
be inclined to construct their ideal self based on the nonfamilial
environment such as the school, peer relations, and the media.
Furthermore, being a boy also reduces the likelihood of parental
influences on ideal-self development. A third basis for predicting
parental influences (or lack thereof) can be drawn from evidence that
parent-to-child effects become less likely as the adolescent grows
older. Thus, the adolescent most receptive to parental influence on his
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
or her ideal self will be a girl who has not yet reached her 18th
birthday and whose parents adopt supportive educational practices.
Caveats and Limitations
This study has several limitations. First, the sample of families
came from a Western European background. The extent to which
the findings can be replicated with families from other cultures
remains to be investigated. Second, because of the relatively short
duration of the study and the lack of certain measures at Time 2,
the developmental and causal inferences that can be drawn are
limited. We should also note that our focus on adolescence as a
period of ideal-self development does not imply that precursors to
individuals’ ideal self could not be identified at earlier phases of
development. Indeed, we feel that examining childhood precursors
to the ideal self in adolescence is a fascinating subject that needs
much more attention than it has hitherto received. Third, the
current work focused on gender, parental ideals, and parental
practices as sources of adolescents’ ideal-self development. Future
research is necessary to understand other likely influences, such as
additional parent and child characteristics (e.g. degree of child–
parent identification), extrafamilial factors (e.g., school, peers,
media), and genetic influences. A final cautionary note applies to
our measures of ideals. We opted for the Q sort in the context of
this research for the advantages described in the introductory
section. However, there are also some disadvantages. Although the
forced distribution has the critical advantage of making ratings
truly comparable across participants, the cost of forcing is that the
overall level of ideals is identical across participants. In other
words, our data do not allow us to determine whether some
individuals have particularly high or low levels of ideals. Simi-
larly, the extent to which participants’ ideals were chronically
accessible (see Higgins, 1987) was not assessed in this research.
Concluding Comment
Following James’s (1902) pronouncement that “the man’s inte-
rior is a battleground for what he feels to be two deadly hostile
selves, one actual, the other ideal” (p. 135), a century of research
has amply documented the importance of ideal-self representations
for psychological functioning, in particular for emotion regulation
and self-esteem. However, little is known about the provenience
and development of the ideal self. How does an ideal self come
about? What are its determinants? And how can its developmental
trajectory be characterized? The current research presents one of
the most comprehensive studies of these questions to date, but it
571
also highlights various issues that remain to be investigated. We
hope that it can serve as a point of departure from which further
research into these issues can be developed.
References
Acock, A. C., & Bengtson, V. L. (1980). Socialization and attribution
processes: Actual versus perceived similarity among parents and youth.
Journal of Marriage and the Family, 40, 501–515.
Aiken, L., & West, S. (1991). Multiple regression: Testing and interpreting
interactions. New York: Academic Press.
Albert, I., & Trommsdorff, G. (2003, August). Intergenerational transmis-
sion of family values. Paper presented at the 11th European Conference
on Developmental Psychology, Milan, Italy.
Aronfreed, J. (1968). Conduct and conscience. New York: Academic Press.
Balswick, J. O., & Macrides, C. (1975). Parental stimulation for adolescent
rebellion. Adolescence, 10, 253–266.
Bandura, A. (1971). Psychological modeling: Conflicting theories. Chi-
cago: Aldine-Atherton.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator distinction in
social psychological research: Conceptual, strategic, and statistical consid-
erations. Journal of Personality and Social Psychology, 51, 1173–1182.
Block, J. (1971). Lives through time. Berkeley, CA: Bancroft Books.
Block, J. (1978). The Q-sort method in personality assessment and psy-
chiatric research. Palo Alto, CA: Consulting Psychologists Press. (Orig-
inal work published 1961)
Block, J., & Robins, R. (1993). A longitudinal study of consistency and
change in self-esteem from early adolescence to early adulthood. Child
Development, 64, 909 –923.
Block, J. H. (1983). Differential premises arising from differential socializa-
tion of the sexes: Some conjectures. Child Development, 54, 1335–1354.
Brehm, S. S., & Brehm, J. W. (1981). Psychological reactance: A theory
of freedom and control. New York: Academic Press.
Bugental, D. B., & Grusec, J. E. (2006). Socialization processes. In W.
Damon (Series Ed.) & N. Eisenberg (Vol. Ed.), Handbook of child
psychology: Vol. 3. Social, emotional, and personality development (6th
ed., pp. 366 – 428). New York: Wiley.
Cashmore, J. A., & Goodnow, J. J. (1985). Agreement between genera-
tions: A two-process approach. Child Development, 56, 493–501.
Cooley, C. H. (1902). Human nature and the social order. New York:
Scribner.
Cronbach, L. J. (1955). Processes affecting scores on “understanding of
others” and “assumed similarity.” Psychological Bulletin, 52, 177–193.
Cross, S., & Madson, L. (1997). Models of the self: Self-construals and
gender. Psychological Bulletin, 122, 5–37.
Cross, S., & Markus, H. (1991). Possible selves across the life span.
Human Development, 34, 230 –255.
Dekovic, M., Janssens, J. M., & Gerris, J. R. (1991). Factor structure and
construct validity of the Block Child Rearing Practices Report (CRPR).
Psychological Assessment, 3, 182–187.
Efron, B., & Tibshirani, R. (1993). An introduction to the bootstrap.
London: Chapman & Hall.
Eisenberg, N., Carlo, G., Murphy, B., & van Court, P. (1995). Prosocial
development in late adolescence. Child Development, 66, 1179 –1197.
Eisenberg, N., Miller, P. A., Shell, R., McNalley, S., & Shea, C. (1991).
Prosocial development in adolescence: A longitudinal study. Develop-
mental Psychology, 27, 849 – 857.
Freud, S. (1961). The ego and the id. In J. Strachey (Ed. & Trans.), The
standard edition of the complete psychological works of Sigmund Freud
(Vol. 19, pp. 12–59). London: Hogarth Press. (Original work published 1923).
Funder, D. (1995). On the accuracy of personality judgment: A realistic
approach. Psychological Review, 102, 652– 670.
Good, P. (1994). Permutation tests: A practical guide to resampling
methods for testing hypotheses. New York: Springer-Verlag.
Grolnick, W., Deci, E., & Ryan, M. (1997). Internalization within the
 572 ZENTNER AND RENAUD
family: The self-determination theory perspective. In J. E. Grusec &
L. Kuczynski (Eds.), Parenting and children’s internalization of
values: A handbook of contemporary theory (pp. 135–161). New
York: Wiley.
Grusec, J. E., & Goodnow, J. J. (1994). Impact of parental discipline
methods on the child’s internalization of values: A reconceptualization
of current points of view. Developmental Psychology, 30, 4 –19.
Harter, S. (1999). The construction of the self: A developmental perspec-
tive. New York: Guilford Press.
Higgins, T. E. (1987). Self-discrepancy: A theory relating self and affect.
Psychological Review, 94, 319 –340.
Hoffman, M. L. (1970). Moral development. In P. H. Mussen (Ed.), Carmichael’s
manual of child psychology (Vol. 2, pp. 261–361). New York: Wiley.
Horney, K. (1950). Neurosis and human growth. New York: Norton.
Hubert, L. J. (1984). Statistical applications of linear assignment. Psy-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
chometrika, 49, 449 – 473.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Inhelder, B., & Piaget, J. (1958). The growth of logical thinking from
childhood to adolescence. New York: Basic Books.
Insightful Corporation. (2002). S-PLUS 6 Robust Library user’s guide.
Seattle, WA: Author.
James, W. (1890). The principles of psychology. New York: Holt.
James, W. (1902). The varieties of religious experience. London: Routledge.
Keating, D. (2004). Cognitive and brain development. In R. Lerner & L.
Steinberg (Eds.), Handbook of adolescent psychology (2nd ed., pp.
45– 84). Hoboken, NJ: Wiley.
Klohnen, E. C., & Mendelsohn, G. (1998). Partner selection for personality
characteristics: A couple-centered approach. Personality and Social
Psychology Bulletin, 24, 268 –278.
Knafo, A., & Schwartz, S. (2003). Parenting and adolescents’ accuracy in
perceiving parental values. Child Development, 74, 595– 611.
Knafo, A., & Schwartz, S. (in press). Accounting for parent– child value
congruence: Theoretical considerations and empirical evidence. In U.
Schönpflug (Ed.), Perspectives on cultural transmission. Oxford, En-
gland: Oxford University Press.
Kohlberg, L. (1966). A cognitive-developmental analysis of children’s sex
role concepts and attitudes. In E. Maccoby (Ed.), The development of sex
differences (pp. 82–173). Palo Alto, CA: Stanford University Press.
Kuczynski, L. (Ed.). (2003). Hardbook of dynamics in parent– child rela-
tions. Thousand Oaks, CA: Sage.
Larson, R. W., & Almeida, D. (1999). Emotional transmission in the daily
lives of families: A new paradigm for studying family processes. Journal
of Marriage and the Family, 61, 5–20.
Loehlin, J. C. (2005). Resemblance in personality and attitudes between
parents and their children: Genetic and environmental contributions. In
S. Bowles, H. Gintis, & M. Osborne Groves (Eds.), Unequal chances:
Family background and economic success (pp. 192–207). Princeton, NJ:
Princeton University Press.
Luo, S., & Klohnen, E. (2005). Assortative mating and marital quality in
newlyweds: A couple-centered approach. Journal of Personality and
Social Psychology, 88, 304 –326.
May, R. B., & Hunter, M. A. (1993). Some advantages of permutation
tests. Canadian Psychology, 34, 401– 407.
McNally, S., Eisenberg, N., & Harris, D. H. (1991). Consistency and
change in maternal child-rearing practices and values: A longitudinal
study. Child Development, 62, 190 –198.
Mead, G. H. (1934). Mind, self, and society from the standpoint of a social
behaviorist. Chicago: University of Chicago Press.
Meyer, G. J., Finn, S. E., Eyde, L. D., Kay, G., Moreland, K. L., Dies,
R. R., et al. (2001). Psychological testing and psychological assess-
ment: A review of evidence and issues. American Psychologist, 56,
128 –165.
Montemayor, R., & Brownlee, J. (1987). Fathers, mothers and adolescents:
Gender-based differences in parental roles during adolescence. Journal
of Youth and Adolescence, 16, 289 –302.
Moretti, M., & Higgins, T. E. (1999). Own versus other standpoints in
self-regulation: Developmental antecedents and functional conse-
quences. Review of General Psychology, 3, 188 –223.
OCSTAT. (2002). Office Cantonal de la Statistique (OCSTAT): Commu-
nications statistiques [Statistics Bureau of the Canton of Geneva: Sta-
tistical communications]. Geneva, Switzerland: OCSTAT Publishing.
Ogilvie, D. M., & Clark, M. D. (1992). The best and the worst of it: Age
and sex differences in self-discrepancy research. In R. P. Lipka & T. M.
Brinthaupt (Eds.), The self: Definitional and methodological issues (pp.
186 –222). Albany: State University of New York Press.
Ozer, D. (1993). The Q-sort method and the study of personality develop-
ment. In D. C. Funder, R. D. Parke, C. T. Keason, & K. Widaman (Eds.),
Studying lives through time, personality and development (pp. 147–168).
Washington, DC: American Psychological Association.
Pomerantz, E., Saxon, J. L., & Kenney, G. A. (2001). Self-evaluation: The
development of sex differences. In G. B. Moskowitz (Ed.), Cognitive
social psychology: The Princeton Symposium on the Legacy and Future
of Social Cognition (pp. 59 –73). Mahwah, NJ: Erlbaum.
Rickel, A. U., & Biasatti, L. L. (1982). Modification of the Block Child
Rearing Practices Report. Journal of Clinical Psychology, 38, 129 –134.
Roberts, B. W., & DelVecchio, W. F. (2000). The rank-order consistency
of personality traits from childhood to old age: A quantitative review of
longitudinal studies. Psychological Bulletin, 126, 3–25.
Roberts, B. W., O’Donnell, M., & Robins, R. W. (2004). Goal and
personality trait development in emerging adulthood. Journal of Per-
sonality and Social Psychology, 87, 541–550.
Roberts, B. W., & Robins, R. W. (2000). Broad dispositions, broad
aspirations: The intersection of the Big Five dimensions and major life
goals. Personality and Social Psychology Bulletin, 26, 1284 –1296.
Rogers, C. (1961). On becoming a person. Boston: Houghton Mifflin.
Rohan, M. J., & Zanna, M. P. (1996). Value transmission in families. In C.
Seligman, J. M. Olson, & M. P. Zanna (Eds.), The psychology of values:
The Ontario Symposium (Vol. 8, pp. 253–276). Hillsdale, NJ: Erlbaum.
Rokeach, M. (1973). The nature of human values. New York: Free Press.
Sandler, J., & Rosenblatt, B. (1962). The concept of the representational
world. Psychoanalytic Study of the Child, 17, 128 –145.
Schafer, R. (1968). Aspects of internalization. Madison, CT: International
Universities Press.
Schönpflug, U. (2001). Intergenerational transmission of values: The role of
transmission belts. Journal of Cross-Cultural Psychology, 32, 174 –185.
Silverman, B. W. (1986). Density estimation for statistics and data anal-
ysis. London: Chapman & Hall.
Smetana, J. G. (2000). Middle-class African American adolescents’ and
parents’ conceptions of parental authority and parenting practices: A
longitudinal investigation. Child Development, 71, 1682–1686.
Tangney, J. P., Niedenthal, P. M., Covert, M. V., & Hill-Barlow, D. (1998).
Are shame and guilt related to distinct self-discrepancies? A test of
Higins’s 1987 hypotheses. Journal of Personality and Social Psychol-
ogy, 75, 262–268.
Troll, L., & Bengtson, V. (1979). Generations and the family. In W. R.
Burr, R. Hill, F. I. Nye, & I. L. Reiss (Eds.), Contemporary theories
about the family (Vol. 1, pp. 127–161). New York: Free Press.
Whisman, M. A., & McClelland, G. H. (2005). Designing, testing, and
interpreting interactions and moderator effects in family research. Jour-
nal of Family Psychology, 19, 111–120.
Whitbeck, L. B., & Gecas, V. (1988). Value attributions and value trans-
mission between parents and children. Journal of Marriage and the
Family, 50, 829 – 840.
Wilcox, R. R. (1997). Introduction to robust estimation and hypothesis
testing. San Diego, CA: Academic Press.
Winch, R. F. (1962). Identification and its familial determinants. India-
napolis, IN: Bobbs-Merrill.
Zentner, M. (2005). Ideal mate personality concepts and compatibility in
close relationships: A longitudinal analysis. Journal of Personality and
Social Psychology, 89, 242–256.
 ORIGINS OF THE IDEAL SELF 573

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
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Forgives easily Forthcoming Generous, likes to share with others

This document is copyrighted by the American Psychological Association or one of its allied publishers.

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)
 574 ZENTNER AND RENAUD

Appendix B

Illustrative Example of Permutation Analysis

Consider a table with one row for each child and one column for
each mother (see Figure B1). The value in a cell is the similarity
index between the child (rows) and the mother (columns). The
hypothesis can be restated as follows: Are the values on the
diagonal (similarities between true, or familial, dyads) larger than
the off-diagonal values (similarities between random dyads)? The
difference of means between diagonal and off-diagonal values is
the averaged increase in correlation because of familial links. The
permutation test computes this difference for the original table (see
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
difference for all possible permutations of the rows (highlighted in
Figure B1, right side). The values’ diffmean for the six permuta-
tions are 0.15, 0.025, 0, ⫺0.025, ⫺0.05, and ⫺0.1. The p value is
the proportion of values ⱖ .15, that is, 1/6 ⫽ .17. All permutations
are computed, including the identity permutation, the original
sample. Its value being always equal to the target value, the p value
can never be smaller than 1 over the number of permutations. In
this example, the p value can therefore not be smaller than 1/6 ⫽
.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.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Figure B1. Illustration of permutation analysis.

Received May 19, 2005

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