ArticlePDF Available

How Preschoolers Associate Power with Gender in Male-Female Interactions: A Cross-Cultural Investigation


Abstract and Figures

Interactions between males and females often display a power imbalance. Men tend to adopt more dominant physical postures, lead conversations more, and are more likely to impose their will on women than vice versa. Furthermore, social representations typically associate males with a higher power than females. However, little is known about how those representations emerge in early childhood. The present study investigated whether preschool children from different countries assign more power to males than to females in the context of mixed-gender interactions. In Experiments 1a (n = 148) and 1b (n = 403), which implemented power through body postures, 4–6 year-old children from France, Lebanon, and Norway strongly associated power with a male character. Experiment 2 (n = 160) showed that although both French boys and girls identified themselves more with a dominant than with a subordinate posture, girls were less likely to do so in a mixed-gender context. In Experiment 3 (n = 213), which no longer used body postures, boys from Lebanon and France attributed more decision power and resource control to a male puppet than did girls. By investigating gender representations through interactions, the present study shows that children associate gender and power at an early age.
This content is subject to copyright. Terms and conditions apply.
1 23
Sex Roles
A Journal of Research
ISSN 0360-0025
Sex Roles
DOI 10.1007/s11199-019-01116-x
How Preschoolers Associate Power with
Gender in Male-Female Interactions: A
Cross-Cultural Investigation
Rawan Charafeddine, Imac Maria
Zambrana, Benoit Triniol, Hugo
Mercier, Fabrice Clément, Laurence
Kaufmann, Anne Reboul, et al.
1 23
Your article is protected by copyright and
all rights are held exclusively by Springer
Science+Business Media, LLC, part of
Springer Nature. This e-offprint is for personal
use only and shall not be self-archived in
electronic repositories. If you wish to self-
archive your article, please use the accepted
manuscript version for posting on your own
website. You may further deposit the accepted
manuscript version in any repository,
provided it is only made publicly available 12
months after official publication or later and
provided acknowledgement is given to the
original source of publication and a link is
inserted to the published article on Springer's
website. The link must be accompanied by
the following text: "The final publication is
available at”.
How Preschoolers Associate Power with Gender in Male-Female
Interactions: A Cross-Cultural Investigation
Rawan Charafeddine
&Imac Maria Zambrana
&Benoit Triniol
&Hugo Mercier
&Fabrice Clément
Laurence Kaufmann
&Anne Reboul
&Francisco Pons
&Jean-Baptiste Van der Henst
#Springer Science+Business Media, LLC, part of Springer Nature 2020
Interactions between males and females often display a power imbalance. Men tend to adopt more dominant physical postures,
lead conversations more, and are more likely to impose their will on women than vice versa. Furthermore, social representations
typically associate males with a higher power than females. However, little is known about how those representations emerge in
early childhood. The present study investigated whether preschool children from different countries assign more power to males
than to females in the context of mixed-gender interactions. In Experiments 1a (n= 148) and 1b (n= 403), which implemented
power through body postures, 46 year-old children from France, Lebanon, and Norway strongly associated power with a male
character. Experiment 2 (n= 160) showed that although both French boys and girls identified themselves more with a dominant
than with a subordinate posture, girls were less likely to do so in a mixed-gender context. In Experiment 3 (n= 213), which no
longer used body postures, boys from Lebanon and France attributed more decision power and resource control to a male puppet
than did girls. By investigating gender representations through interactions, the present study shows that children associate
gender and power at an early age.
Keywords Power .Gender .Dominance .Preschoolers .Mixed-gender interactions .Cross-cultural comparisons
Because children grow up in a socially stratified world, they
experience situations in which some individuals exert power
and authority on others. Upon observing instances of such
interactions, young children may appreciate that people who
dominate and control others share common features. For in-
stance, they will detect that adults often hold a position of
power when interacting with them. Similarly, as they begin
to observe peer interactions, they might also learn that older
children are likely to prevail over younger children in con-
flicts. Experimental work has shown that age and age-related
features, such as body size, are robust cues used by young
children and even infants to predict who is dominant in a
social interaction (Charafeddine et al. 2015; Lourenco et al.
2016; Thomsen et al. 2011). Another recurrent aspect of the
environment that correlates with power is gender. Although
within the family and in the classroom the adults who exercise
authority over children are more often women, in less benev-
olent contexts children are likely to be exposed to dyadic
interactions between same-age individuals where men exer-
cise power over women. The purpose of the present study is to
determine whether early conceptions of male-female interac-
tions carry a power asymmetry, that is, whether preschoolers
consider that a dominant individual is more likely to be male
or female.
Throughout history, human groups have consistently built
gender categories on the basis of sexual morphological traits
and have defined distinct masculine and feminine roles.
Although gender roles vary considerably across societies
Electronic supplementary material The online version of this article
( contains supplementary
material, which is available to authorized users.
*Rawan Charafeddine
Jean-Baptiste Van der Henst
Institut des Sciences Cognitives Marc Jeannerod, Laboratoire
Langage Cerveau Cognition, UMR-5304, 67 Boulevard Pinel,
69500 Bron Cedex, France
University of Oslo, Oslo, Norway
Université de Neuchâtel, 2000 Neuchâtel, Switzerland
Lausanne University, 1015 Lausanne, Switzerland
Sex Roles
Author's personal copy
and history, gender emerges as a social construct that distin-
guishes men from women on a hierarchical basis. Men have
benefited from higher status than women, have controlled
more of the public sphere, and have exercised power over
women (Darmangeat 2009;Rosaldoetal.1974;Whyte
1978). Most explanatory models that account for the origins
of male domination pay particular attention to the sexual di-
vision of labor (SDL) in hunter-gatherer societies and describe
it as the most ancient example of a status-relevant distinction
between men and women. Broadly speaking, biological ex-
planations argue that reproductive constraints prevent women
from carrying out tasks that are described as risky or difficult
and relegate them to the domestic sphere (Bird 1999;
Murdock and Provost 1973). For other approaches, more than
reproductive constraints, it is the exploitation and subjugation
of women for the benefit of men that leads to multiple prohi-
bitions imposed on women (Brightman 1996; Rosaldo et al.
1974; Tabet 1979).
Despite modern political efforts to promote gender equal-
ity and women empowerment in Western industrialized
countries, a hierarchy persists (Ridgeway 2011). Women still
occupy fewer prestigious jobs or leadership positions, control
fewer resources than men, and spend more ime in domestic
(i.e., unpaid and often devalued) work (Bianchi et al. 2000;
Horne et al. 2018; Yavorsky et al. 2015). Asymmetries are
also reflected through power dynamics that govern cross-
gender relationships. Statistics on femicide, rape, sexual ha-
rassment or genital mutilation are dramatic examples of
mens power over women and their bodies (https://www.
At a less extreme but more widespread and systematic lev-
el, people recurrently perform gender inequality in their daily
interactions (Goffman 1977; West and Zimmerman 1987). For
example, in mixed-gender conversations, interruptions and
turn-taking violations are more often made by males
(Anderson and Leaper 1998), even in preschool years
(Esposito 1979). Men are also more likely to communicate
overt orders and imperatives as well as speak in an assertive
manner, whereas women are more likely to engage in indirect
requests and in polite and tentative language (Anderson and
Leaper 1998; Lakoff 1973). Nonverbal behaviors also show
some gender differences relevant to power. Men tend to adopt
more open and expanded body postures and speak louder than
women (Cashdan 1998; Hall et al. 2005; Henley 1977). They
also demonstrate dominant gaze behavior (Dovidio et al.
1988;Ridgeway2013), that is, men are more likely than
women to display direct gaze while speaking and divert gaze
while listening rather than the reverse (dominant gaze ratio;
Dovidio et al. 1988). All such behaviors obviously might be
reinforced by gender stereotypes held by adults and transmit-
ted to children, leading to considering women as less compe-
tent and agentic than men (Fiske et al. 2007; Fiske et al. 2002).
Gender stereotypes also lead to judging dominant behavior
and leadership more negatively when adopted by women than
by men (Eagly et al. 1992).
In addition to asymmetric male-female relationships and
gender stereotypes, individual physical traits related to sex
can influence the perception of who dominates in an interac-
tion. In most primate species including humans, males are
taller, have lower voices, more muscular bodies, and more
dominant faces. Within each sex, these traits correlate with
social status and the outcome of conflicting interactions, and
people often use these traits to predict who is dominant
(Lukaszewski et al. 2016; Puts et al. 2006; Todorov et al.
2015; Toscano et al. 2014). Hence, children are likely to be
exposed to a wide set of stimuli of male-female differences,
including daily interactions, adult stereotypes, and morpho-
logical features that are relevant to assign more power to
Understanding of Social Hierarchies in Early
Recently, growing evidence has shown that preschoolers and
even infants exhibit a fine-grained understanding of hierarchi-
cal relationships. First, childrens identification of hierarchy
relies on the observation of dynamic interactions. Three- to 4-
year-old children judge that individuals who impose their
preference on other individuals or who deny permission to
use resources are the boss or are in charge, and children five
years and older base their power judgments on even more
subtle interactions, such as setting norms for others or being
imitated (Charafeddine et al. 2015; Gülgöz and Gelman 2017;
Over and Carpenter 2015).
Furthermore, children rely on a variety of physical and
nonphysical cues to predict who is dominant. For instance,
erect body postures, being endowed with more resources, or
being older are understood as marks of power (Brey and
Shutts 2015; Charafeddine et al. 2015; Gülgöz and Gelman
2017; Terrizzi et al. 2018). Moreover, before the age of one,
infants expect larger agents or agents with more allies to pre-
vail in conflicts (Pun et al. 2016; Thomsen et al. 2011). It is
also important to note that, like adults, preschoolers are sensi-
tive to physical dominance cues that correlate with differences
between males and females. They thus consider that individ-
uals with more masculine faces, that is, those exhibiting larger
jaws, wider chin, and thicker brows, are stronger and are more
likely to be in charge than people with less masculine faces
(Cogsdill et al. 2014; Keating and Bai 1986; Terrizzi et al.
Another key aspect of childrens ability to represent social
hierarchies lies in the inferences they draw. Studies using
preferential looking technique suggest that fifteen-month in-
fants tend to generalize a dominance relationship between two
agents, represented by geometrical figures, from one situation
Sex Roles
Author's personal copy
to another (Mascaro and Csibra 2012) and that 1013-month-
old infants draw transitive inferences on the basis of dyadic
relations between animal puppets (Gazes et al. 2017). In pre-
school years, children also associate different dimensions of
social hierarchy. When 35-year-olds are presented with a
dominant individual who gives orders and tells another sub-
ordinate individual what to do, they expect the dominant per-
son to have more resources and to win a competitive game
against the subordinate (Charafeddine et al. 2015). A study by
Enright et al. (2017) also suggests that 17-month-old infants
infer dominance from asymmetry in resources. Moreover, pre-
schoolers expect the dominant individual to be better informed
than a subordinate individual (Bernard et al. 2016; Castelain
et al. 2016). In the current study, we will examine whether
inferences children make from power interactions also extend
to gender, that is, whether they expect that the powerful indi-
vidual is more likely to be a girl or a boy in mixed-gender
Early Conceptualization of a Gender
Among the categories that children use to carve up the social
world, gender is one of the most prominent (Bussey and
Bandura 1999; Fagot et al. 2012; Kinzler et al. 2010; Martin
and Halverson Jr 1981; Martin et al. 2002). However, most
studies dedicated to childrens representations of gender have
mainly investigated conceptualizations of individual charac-
teristics related to men and women and did not directly focus
on relational dynamics, such as the imbalance of power be-
tween genders.
Some research addressed gender inequality in the positions
held by men and women in society. For instance, in a study
conducted by Liben et al. (2001), 68-year-old and 1112-
year-old children rated the status of familiar jobs, which were
culturally masculine or feminine, and the status of unknown
fictitious jobs held by women or by men. For familiar jobs,
both age groups judged that masculine jobs were higher in
status than feminine jobs, whereas only the older group gen-
eralized this asymmetry to unknown jobs. In a related study
(Weisgram et al. 2010), which involved children (510 years-
old), adolescents (1117 years-olds) and adults, participants
assigned higher scores of remuneration and power (i.e., mak-
ing rules for others) to masculine jobs than to feminine jobs,
with a developmental pattern indicating that this difference
was greater for adults than for adolescents and greater for
adolescents than for children. Similarly, Neff et al. (2007,
Study 2) interviewed 715-year-old participants about gender
asymmetries in society with questions related to status and
power, such as In general, who do you think has more power
to make important decisions and tell other people what to do,
men or women or do both have the same amount of power?
Whereas the youngest children perceived gender equality in
power and status, the oldest participants attributed a higher
rank to men. However, in those studies, the weaker perception
of gender inequality observed for younger children might re-
sult from a more limited experience with gender imbalance in
a societal context, which requires considering women and
men, power and status, as abstract societal entities. Younger
children may have clearer expectations of gender asymmetries
in situations involving concrete and observable interactions, in
which an individual gives orders to another individual, pre-
vails in conflicts, or monopolizes more resource.
Also relevant to the current study are psychological traits
that children associate with females and males. Past research
has shown that preschoolers have knowledge of gender roles
and adult gender stereotypes (Cowan and Hoffman 1986;
Fagot et al. 1992; Kuhn et al. 1978; Leinbach et al. 1997;
Picariello et al. 1990; Williams et al. 1975), which could in-
fluence their representation of power in mixed-gender interac-
tions. For example, preschoolers tend to associate fear, weak-
ness, and softness with female characters and anger, strength,
and hardness with male characters (Birnbaum and Chemelski
1984; Birnbaum et al. 1980; Cowan and Hoffman 1986).
Preschoolers also expect unknown female characters to adopt
more pro-social behaviors than male characters (Clément et al.
2014). Finally, children also hold beliefs on aggressive behav-
ior that differ according to the gender of the aggressor. In
particular, when preschoolers are asked to describe how boys
and girls are likely to harm another child, they tend to see boys
as the perpetrators of physical aggression (e.g., hitting, throw-
ing things at someone) and see girls as the perpetrators of
relational aggression (e.g., denying friendship, not playing
with another child) (Giles and Heyman 2005). Such represen-
tations are consistent with actual field observations on gen-
dered styles of aggression (Crick et al. 1997; Crick and
Grotpeter 1995; Ostrov and Keating 2004).
Childrens conceptions of a gender hierarchy have also
been examined in the context of status differences between
social groups (Mandalaywala et al. 2019; Olson et al. 2012).
This research compares how children from preschool years
make sense of race and gender as cues to group-based hier-
archies, and it uses wealth-to-group matching tasks to mea-
sure perception of status. Whereas children match the ste-
reotypically high-status race group (i.e., faces of White peo-
ple) with a high level of wealth (e.g., high-value houses), as
compared to the lower status racial group (i.e., faces of
Black people), they do not do so for gender. In particular,
they did not match more male targets as compared to female
targets with higher value resources. Rather, children some-
times show a slight tendency to associate their own gender
with higher value goods (Olson et al. 2012). However, when
the wealth asymmetry more directly involves competition
for resources relevant to children (e.g., having lots of toys
and new clothes, always getting to pick the tasks that
Sex Roles
Author's personal copy
everyone else plays at recess and the snacks that everyone
else eats at snack time), preschoolers assign a higher status
to male targets (Mandalaywala et al., 2019).
The prior research shows asymmetries in childrens con-
ceptions of male and female categories, but it does not directly
address the representation of mixed-gender interactions and
power relations. Understanding how children conceptualize
mixed-gender relationships with respect to power is important
not only for research on cognitive development but also for
understanding the perpetuation of gender inequalities and
their justifications in societies. If young children consider
boys to be more likely to dominate girls than the other way
around, they may act in conformity with such an asymmetry
or believe that it is appropriate.
The Current Experiments
The present studies examine four issues. The first relates to
preschoolersexpectations about the gender of the individual
who displays more or less power in a dyadic interaction.
Because in mixed-gender interactions involving adults, men
are more likely to dominate women (Carli 1999), and because
preschoolers are good at understanding power relationships
from interactions and from physical features, they might asso-
ciate power with male individuals. An important question is
the developmental pathway of this association. Although past
research has shown that even infants use differences between
individuals to infer who is dominant in an interaction, some
asymmetries relevant to power are not always understood by
young children. Indeed, although 34-year-old children used
resource control, goal achievement, age, and denying permis-
sion as cues to power, it is only between 5 and 7 years of age
that children consider an individual who sets norms for others,
or who is imitated by others, to hold a powerful position
(Gülgöz and Gelman 2017; Over and Carpenter 2015).
The second issue is the potential influence of participants
gender. There is converging evidence that children view their
own gender and members of their gender categories positive-
ly. Observations of young children in daycare facilities
showed that shortly before their third birthday, children affil-
iate more with same-gender peers (Fishbein and Imai 1993; La
Freniere et al. 1984; Maccoby and Jacklin 1987). When pre-
schoolers are explicitly asked to rank their peers according to
whom they prefer, they tend to attribute higher scores to same-
gender children (Hayden-Thomson et al. 1987; Kuhn et al.
1978; Sebanc et al. 2003). Moreover, preschoolers tend to
align their preferences for novel objects with those of same-
gender peers (Shutts et al. 2009) and are more likely to en-
dorse the testimony of same-gender informants (Ma and
Woolley 2013; Terrier et al. 2016). Children also show a bias
for their own gender in implicit attitude test (Dunham et al.
2016), and they think that their own gender is judged positive-
ly by other people (Halim et al. 2013).
In addition to the own-gender bias, preschoolers show pos-
itive attitudes toward dominant individuals (Hawley 1999).
They trust the testimony of a dominant individual more than
that of a subordinate (Bernard et al. 2016; Castelain et al.
2016), and 34-year-olds distribute resources in ways that
benefit dominant over subordinate individuals (Charafeddine
et al. 2016). Hence, viewing ones own gender positively may
modulate the association between being male and power. That
is, girls may be less likely to consider a dominant individual to
be a male than boys. A strong version of this hypothesis would
be that girls consider that a dominant character is a female to
the same extent as boys consider that a dominant character is a
The third issue is how self-perception interacts with the
male-power association. Preschoolers tend to view their own
dominance rank as high (Mandalaywala et al. 2019; Omark
and Edelman 1975; Sluckin and Smith 1977; Strayer et al.
1978), and in a western culture such as France, they tend to
identify themselves more with the dominant than with the
subordinate figure (Charafeddine et al. 2019). A self-
enhancement bias could thus override, or at least diminish,
the male-power association in girls. That is, boys may see
themselves as dominants when they imagine themselves
interacting with girls, but girls may also see themselves as
dominants when they imagine interacting with boys.
The fourth issue concerns the influence of the cultural con-
text on the male-power association. Research has shown that
culture modulates the content of gender stereotypes (Williams
and Best 1982) and that the social environment can affect
preschoolersattitudes toward gender (Shutts et al. 2017). In
particular, in their study, which involved 25 cultures, Williams
and Best (1982) observed a greater differentiation of stereo-
types in Muslim countries, especially in Pakistan, than in non-
Muslim countries. Other studies have also reported more pro-
nounced stereotypes among Italian than Dutch children
(Zammuner 1982) and more pronounced male stereotypes
among South African than among U.S. children (Albert and
Porter 1986). One may anticipate that the male-power associ-
ation is stronger or emerges earlier in countries where gender
inequalities are more prevalent. To examine this issue,
Experiment1 involved French, Lebanese, and Norwegian par-
ticipants and Experiment 3 compared French and Lebanese
The current work follows a methodology typical of devel-
opmental research on gender stereotypes. It consists of pre-
senting participants with a specific behavior and asking them
to guess whether this behavior is more likely to be performed
by a girl or a boy (Kuhn et al. 1978; Williams et al. 1975). The
experimental procedure in Experiment 1 obtained ethical ap-
proval from the Norwegian ministry of education (Norwegian
Centre for Research Data, #30095). The procedures in
Sex Roles
Author's personal copy
Experiments 1, 2, and 3 were approved by the French ministry
of education (Inspection Académique and Coordination
Académique Recherche-Développement Innovation et
Expérimentation). In the absence of a national ethical board
in Lebanon, the Lebanese ministry of education and higher
education gave an approval in principle to conduct the
Experiment 1a
A combination of nonverbal (i.e., dominance and subordina-
tion physical postures) and verbal cues (statements expressing
power) were used to convey a power asymmetry between two
nongendered individuals. Then, participants had to guess the
gender of the dominant and the gender of the subordinate.
Facial expressions, body postures, and gestures are common
indicators of dominance, such as pointing, touching, staring,
frowning or maintaining an erect posture (Hall et al. 2005;
Henley 1977; Weisfeld and Beresford 1982). Adults quickly
use such features to determine power positions (Hall et al.
2005; Schmid-Mast and Hall 2004). However, the extent to
which a power asymmetry is perceived through postures is
modulated by the gender of individuals adopting such pos-
tures. Henley and her colleagues found that when women
display dominance gestures (e.g., pointing at another individ-
ual, invadingspace) toward men, such gestures were less often
interpreted as signaling dominance than when they were
displayed by men toward women (Henley 1977; Henley and
Harmon 1985). Experiments 1a and 1b examine whether a
similar effect holds for children.
Developmental research has shown that preschoolers are
able to infer power from nonverbal features. For instance,
Keating and Bai (1986) found that children tend to associate
nonsmiling and lowered-brow faces with the dominant posi-
tion (see also, Cogsdill et al. 2014;Terrizzietal.2018).
Similar results were obtained with body postures and gestures
(Charafeddine et al. 2015; Charafeddine et al. 2019). Children
saw a picture with two characters displaying either dominance
or subordination features: the dominant character had their
head up and an erect body and displayed a pointing gesture
(Henley and Harmon 1985) at the subordinate character who
had their head down and who exhibited a constriction posture.
Charafeddine et al. (2015) found that 35-year-old children
judge that the former character was the dominant, and
Charafeddine et al. (2019) reported that this tendency was
stronger for 5- than for 3-year-old children. The ability to
attribute power to body postures has also been reported by
Brey and Shutts (2015) and Terrizzi et al. (2018), who show
that older preschoolers are more sensitive to these cues than
younger preschoolers. In Experiments 1a and 1b, we use a
similar dominance-subordination interaction picture with
nongendered characters as in Charafeddine et al. (2015,
2019). Here, participants first had to decide who was domi-
nant and who was the subordinate and subsequently had to
guess who was the girl and who was the boy.
Initially 149 French preschoolers participated in this experi-
ment. They were tested in two public nursery schools located
in middle SES neighborhoods in the city of Lyon. Parental
consent was obtained for each child. One child was excluded
from the analysis for failing to answer the gender-attribution
questions. The final set of participants included 148 children
= 56.4 mo., SD = 9.03) from three age groups: 34 3-
year-olds (M
= 42.9 mo., SD = 3.56; 12 girls and 22 boys),
52 4-year-olds (M
= 55.3 mo., SD = 3.18; 25 girls and 27
boys), and 62 5-year-olds (M
= 64.7 mo., SD = 3.28; 31
girls and 31 boys).
Materials and procedure
Three female experimenters interviewed the participants; all
of them followed the same standardized procedure. As in
studies by Charafeddine et al. (2015,2019), the first part of
the task consisted of deciding who was dominant and subor-
dinate by answering two comprehension questions. Children
saw a drawing of two nongendered characters interacting with
each other, one displaying a dominance posture and the other
displaying a subordination posture (see Fig. 1; the original
color version of this figure is available in the online
supplement). They were told that one character was saying:
You have to do everything I say! Do what I want![Tu dois
faire tout ce que je dis! Fais comme je veux!] (power state-
ment) and the other was replying: Ok! I will do what you
want[Daccord je fais ce que tu veux] (obedience statement).
Children then had to match postures with each statement. The
spatial positions (left vs. right) of the characters were
counterbalanced across participants. When participants
Fig. 1 Two cha ra cte rs d isp la ying po sture s of domi na nce a nd
subordination (see the online supplement for the original colored
version of the picture)
Sex Roles
Author's personal copy
matched the dominance posture with the power statement and
the subordination posture with the subordination statement,
their answer was considered as correct. In the second part of
the task, participants were told that the picture actually
depicted a girl and a boy and were asked to indicate who
was who.
The specific instructions were as follows:
Look! You see? There are two children. Do you see
them? One of the children tells the other: You must
do everything I say! Do whatever I want!And the other
child is saying: OK! I will do what you want.
These instructions were followed by two comprehen-
sion questions (with their order counterbalanced):
Who is saying Do what I want?’”;Who is saying
OK! I will do what you want?’” The procedure then
continued with the statement, In fact, in this picture
there is a girl and a boy,followed by two gender attri-
bution questions (with their order counterbalanced):
Could you tell me who is the girl?;And who is the
All through the present paper, the reported binomial tests are
two-tailed and compare the relevant rates to chance (50%).
Comprehension questions
Fully 124 of 148 (83.78%) correctly matched the erect body
posture with the power statement and the subordinate posture
with the obedience statement (p<.001, 95% CI [76.84,
89.32]). There was no significant effect of age group (3-
year-olds: 82.35%, 4-year-olds: 82.69%; 5-year-olds:
85.48%), χ
(2, n= 148) = .23, p= .892 or participantsgender
(girls: 85.29% and boys: 82.50%), χ
(1, n=148)=.21,
p= .646.
Gender assignment
A logistic regression was run on all the data to assess the
effects of age, gender of the participants, and comprehension
questions (incorrect (0), correct (1)) on gender assignment.
The gender associated with the dominant figure was coded
in a binary manner: female (0) and male (1). The Chi-square
test on the produced model revealed that the answer to the
comprehension question was the only significant factor
(p< .001; Akaike Information Criterion [AIC] = 176.35).
Children who incorrectly answered the comprehension ques-
tions were at chance: 37.5% associated the erect body posture
with a male character (p= .307, 95% CI [40.59, 81.2]). In
contrast, 75% of the children who correctly answered those
questions considered that the dominant character was a boy
(p< .001, 95% CI [66.42, 82.33]). The difference between the
two types of children was significant (37.5% vs. 75%), χ
n= 148) = 13.2, p< .001. Childrens age was not significant in
the model (p= .070).
Furthermore, we analyzed gender assignment among the
124 children who correctly answered the first two comprehen-
sion questions. The male-power association was significantly
above chance for the 4-year-old group (74.42%; p= .002,
95% CI [58.83, 86.48]) and the 5-year-old group (83.02%;
p< .001, 95% CI [70.2, 91.93]), but not for the 3-year-old
group (60.71%, p= .345, 95% CI [40.58, 78.49]). However,
the comparison between age groups was still not significant,
(2, n= 124) = 4.87, p= .087. Finally, childrens gender was
not a significant factor (p= .912). Among the 124 children
who correctly answered the comprehension questions, the
male-power association was above chance level for both girls
(77.59%; p< .001, 95% CI [64.73, 87.49]) and boys (72.73%;
p< .001, 95% CI [60.36, 82.97]). The difference between
male and female participants was not significant, χ
(1, n=
124) = .39, p= .533.
In line with earlier studies, preschoolers correctly interpreted
dominance postures (Brey and Shutts 2015; Charafeddine
et al. 2015,2019; Keating and Bai 1986; Terrizzi et al.
2018). More importantly, the results indicated that 4- and 5-
year-old children expected the dominant character to be a male
in the context of mixed-gender interactions. This finding ech-
oes results with adults showing that when dominance gestures
are performed by men, they are more likely to be interpreted as
conveying dominance than when performed by women
(Henley and Harmon 1985). However, the male-power asso-
ciation was not found among 3-year-old children. This non-
finding suggests that this association emerges later than asso-
ciations linking power to other dimensions such as age, re-
sources, physical superiority or getting onesway
(Charafeddine et al. 2015; Gülgöz and Gelman 2017). The
hypothesis of an influence of childrens gender was not sup-
ported. Girls were not biased toward their own gender and
exhibited the same male-power association as did boys.
Although participantsgender had no effect, another factor
that might modulate the male-power association is culture.
Experiment 1b examines this factor by comparing Lebanese
and Norwegian children on the same task.
Experiment 1b
Gender inequality largely differs across cultures and nations. In
environments where the level of gender inequality is relatively
Sex Roles
Author's personal copy
low, children might perceive gender relationships as less un-
equal than in countries where it is higher. Two well-known
indexes provide worldwide measures of gender inequality:
The Global Gender Gap Index (GGGI), developed by the
World Economic Forum, which measures the gender gap in
the areas of politics, economics, education and health, and the
Gender Inequality Index (GII), established by the United
Nations Development Program, which focuses on reproductive
health, empowerment, and the labor market. The GGGI indi-
cates that Frances rank ranged from 57th to 15th (median rank:
17) over the 20122016 period, and the GII indicates that
Frances rank ranged from 19th to 9th (median rank: 12.5) over
the 20122015 period. Using the same procedure as
Experiment 1a, Experiment 1b examines the male-power as-
sociation with preschool children from two countries that differ
in terms of gender inequalityNorway and Lebanon.
Norwegian policies largely promote gender equality not only
in welfare and job occupations but also in the area of education,
as shown by government guidelines in kindergarten (Rossholt
(Parker et al. 2009Over 20122016, Norway was ranked 2nd or
3rd according to the GGGI and ranged from the 9th to 5th rank
according to the GII over the 20122015 period. In contrast, in
Lebanon, legal and social discriminations based on gender are
much more common (Human Rights Watch 2015;Mansourand
Karam 2012). Over 20122016, LebanonsrankontheGGGI
ranged from 138th to 120th, and over 20122015, Lebanons
rank on the GII ranged from 83rd to 78th.
In Experiment 1b, we expected the male-power association
to be lower among Norwegian than among Lebanese children.
Moreover, the developmental course of the male-power asso-
ciation might depend on the culture in which the child is
reared. If a cultural environment allows for recurrent manifes-
tations of male dominance, then children might grasp the
male-power association earlier. In addition, it is known that
children become more egalitarian with age, especially after
five years of age (Fehr et al. 2008; Lane and Coon 1972;
LoBue et al. 2011; Rochat et al. 2009). In societies where
gender equality is promoted, older children could thus be
more likely to integrate gender into their representation of
social equality than younger children. To examine this issue,
Experiment 1b included not only 3-, 4- and 5-year-old chil-
dren but also 67-year-old children.
Participants were 417 Norwegian and Lebanese children.
They were tested in their schools or day care facilities after
written consent was collected from their parents. In Norway,
152 children participated in the experiment, but 5 children
from the youngest age group were excluded from the analysis
for failing to answer the gender-attribution questions. The fi-
nal Norwegian sample was composed of 147 children (M
62.8 mo., SD = 13.24, 69 girls and 78 boys) and was divided
into four age groups: 22 3-year-olds (M
= 43.1 mo., SD =
13.24; 11 girls and 11 boys), 48 4-year-olds (M
= 54.9 mo.,
SD = 3.43; 19 girls and 29 boys), 41 5-year-olds (M
= 66.9
mo., SD = 3.3, 17; girls and 24 boys) and 36 67-year-olds
(there were only 3 7-year-olds; M
= 80.7 mo., SD = 4.12; 22
girls and 14 boys). Children came from six childcare centers
and one school. A plurality (41%) of the children went to
school or kindergarten located in middle-class neighborhoods.
One childcare center was private and thus comprised children
from more prosperous families (46% of the participants), and
13% of the participants attended a childcare center located in a
low-income neighborhood.
In Lebanon, 263 children participated in the experiment,
but seven children were excluded from the analysis for failing
to answer the gender-attribution questions. The final Lebanese
sample included 256 children aged 3 to 6 years-old divided
into four age groups (M
= 60.29 mo., SD = 14.68; 126 girls
and 130 boys): 80 3-year-olds (M
= 43.1 mo., SD = 3.29; 40
girls and 40 boys), 63 4-year-olds (M
= 56.2 mo., SD =
3.71; 26 girls and 37 boys), 56 5-year-olds (M
= 69 mo.,
SD = 3.81; 28 girls and 28 boys), and 57 6-year-olds (M
80.4 mo., SD = 3.19, 33; girls and 24 boys). Participants
attended three private schools in Beirut and one in South
Lebanon. Two of the Beirut schools were located in predom-
inantly Christian neighborhoods: one of these schools had
tuition fees in the high average of Lebanese schools (23% of
the participants), whereas the other had low tuition fees (24%
of the participants). The third Beirut school was located in a
mainly Muslim (Sunni) neighborhood and had tuition fees in
the high average (38% of the participants). The last school in
south Lebanon was located in a popular Muslim (Shiite) area
and had average tuition fees (15% of the participants). The
proportion of girls did not significantly differ across the two
groups (46.94% in the Norwegian group and 49.21% in the
Lebanese group), χ
(1, n= 403) = .19, p= .654.
Materials and procedure
The material and procedure were the same as in Experiment
1a. The dialogue was translated into Norwegian and Arabic.
Back-translation into French yielded minimal or no diver-
gence from the original text. One Norwegian and one
Lebanese female experimenters conducted the experiment in
their respective countries using a common standardized exper-
imental procedure.
A total of 364 of 403 (90.32%) children correctly answered
the comprehension questions (p< .001, 95% CI [87, 93.03]).
Sex Roles
Author's personal copy
No significant effect of gender (girls: 89.74% and boys:
90.87%), χ
(1, n= 301) = .14, p= .703, or culture was found
(Norway: 93.2% and Lebanon: 88.67%), χ
(1, n= 301) =
2.19, p= .139, but there was a significant effect of age, χ
(3, n=403)=25.02, p< .001. The partition of the Chi-
square test (Agresti 2002) shows that the 3-year-old group
significantly differs from the other three groups, χ
(1, n=
403) = 22.09, p< .001, whereas the 4-year-old, 5-year-old and
6-year-old groups did not significantly differ from each other,
(2, n= 301) = 4.53, p= .104. The proportions of correct
associations between postures and statements were signifi-
cantly above chance level for all four age classes.
Specifically, the proportions are as follows: 78.43% of the 3-
year-olds (p< .001, 95% CI = [69.18, 85.96]), 90.99% of the
4-year-olds (p< .001, 95% CI [84.05,95.59]), 94.85% of the
5-year-olds (p< .001, 95% CI [88.37,98.3]), and 97.85% of
the 6-year-olds (p< .001, 95% CI [92.44,97.74]).
To assess the specific ways in which age, gender, and cul-
ture of the children affect the results, a logistic regression was
first run on all the data. For Experiment 1a, the gender asso-
ciated with the dominant figure was coded in a binary manner:
female (0) and male (1). The following factors were included
in the regression: participants country, age in months, gender,
and the answer to the comprehension questions (incorrect (0)
or correct (1)). An age-by-culture interaction was also includ-
ed in the model. The Chi-square test on the produced model
revealed that answers to the comprehension questions
(p= .008) and age (p= .004) are the only two significant fac-
tors (Akaike Information Criterion [AIC] = 497.36). Children
who erred on the comprehension question assigned the male
character to the erect body posture at rates that did not signif-
icantly differ from chance level (48.72%; p= .937, 95% CI
[32.42, 65.22]). Conversely, 70.60% of the children who cor-
rectly understood the postures assigned the male character to
the erect body posture, a rate that was significantly above
chance level (p< .001, 95% CI [65.62, 75.24]). Because the
comprehension of postures significantly affected the associa-
tion between gender and dominance, χ
(1, n= 403) = 7.82,
p= .005, all supplementary analyses were performed on the
subset of 364 children who gave a correct answer to the com-
prehension questions.
Among the children who correctly answered the compre-
hension question, gender attribution was significantly affected
by age group, χ
(3, n= 364) = 8.38, p= .039. Partition of the
Chi-square shows a significant difference between the 3-year-
old children and all the other children, χ
(1, n= 364) = 5.55,
p= .018, but no effect of age for the three age groups between
4 and 6 years, χ
(2, n= 284) =3.014, p= .221. Specifically,
whereas the male-power association was not significant
among the 3-year-old children (60%, p=.093, 95% CI
[48.44, 70.8]), it was significant in the other three groups (4-
year-olds: 68.32%, p< .001, 95% CI [68.3,77.22]; 5- year-
olds: 79.34%, p< .001, 95% CI [69.64,87.08], and 6- year-
olds: 73.63%, p< .001, 95% CI [63.34,82.31]). Contrary to
our expectations, the model did not yield an effect of partici-
pantsgender (p= .599), culture (p= .333), or an interaction
between age and culture (p= .409).
Among children who correctly answered the comprehen-
sion questions, girls (69.71%, p<.001, 95% CI [62.33,
66,42]) were as likely as boys (71.43%, p< .001, 95% CI
[64.42,77.75]) to assign the male character to the dominant
posture, χ
(1, n= 364) = .13, p= .719. Similarly, Norwegians
(71.53%, p< .001, 95% CI [63.19,78.91]) were as likely as
Lebanese (70.04%, p< .001, 95% CI [63.63, 75.92]) to make
this association, χ
(1, n= 364) = .09, p= .763. There was no
significant male-power association among the 3-year-old chil-
dren from the two countries (Norway: 52.63%, p= .824, 95%
CI [28.86, 75.55] and Lebanon: 62.29%, p= .0721, 95% CI
[48.96, 74.39]. In both countries, the other age groups were
significantly above chance level (all ps < .035).
As in Experiment 1a, 4- to 6-year-old children, but not 3-year-
olds, associated the dominant posture with the male character.
However, in the current experiment, 3-year-olds were less
likely to infer dominance from postures than older children.
Although this developmental effect was not observed in
Experiment 1a or in an earlier study using the same stimuli
(Charafeddine et al. 2015), it was found in another study using
the same stimuli (Charafeddine et al. 2019), as well as in a
study using different stimuli (Terrizzi et al. 2018). In particu-
lar, Terrizzi et al. (2018) found that 3-year-old children were
above chance when using postural information to judge
strength (Who is stronger?) but not when they had to judge
authority (Who is in charge?). Moreover, whereas 4- and 5-
year-old children were good at matching powerful faces with
powerful postures, 3-year-old children responded at a chance
level (Terrizzi et al. 2018). In Experiments 1a and 1b, children
were more likely to assign the male character to the dominant
posture when they first correctly matched a dominance pos-
ture with a dominance statement. Given that 3-year-olds were
less aware that postures can convey power than older children,
they might be less likely to use dominance postures to assess
the gender of the powerful. In other words, their understand-
ing of postures might not be mature enough to allow infer-
ences from power to gender. Although in Experiment 1a,
French 3-year-olds performed as well as older children in
the comprehension questions, it might still be the case that
their understanding of dominance postures is weaker than
older childrens on other measures.
Second, contrary to our expectations, the male-power as-
sociation did not vary across cultures. Explicit cultural norms,
such as gender equality prevalent in Norway, did not prevent
our participants from thinking that dominants were more like-
ly to be boys. Hence, although Norway and Lebanon differ
Sex Roles
Author's personal copy
with respect to womens empowerment, such cultural differ-
ences may not be strong enough to modify the belief in male
dominance. This may be due to a discrepancy between the
official politics of gender equality and the reality of asymmet-
ric mixed-gender relationships to which young children are
exposed in their environment. However, other cultures, age
groups, and other dominance asymmetries may need to be
explored further. Measures of gender inequality considered
here, namely, the GGGI and the GII, do not explicitly take
into account situations in which gender inequality manifests
itself through situations in which one individual gives orders
and tells someone what to do to another individual. In
Experiment 3, we use resource asymmetry as another expres-
sion of dominance, which is more in line with the areas eval-
uated by the GGGI and the GII.
Another striking feature of our results is that in both
experiments, the girlsanswers did not significantly differ
from the boysanswers, which suggests that the girls and
boys perceived gender inequalities in similar ways. This
result does not support the hypothesis that the positive
viewing of their own gender would lead girls to consider
that a dominant individual is more likely to be a girl. It
could even be possible that girls might judge other girls
who are adopting a dominant behavior more negatively
because such behavior would violate gender norms
associated with females. In keeping with this
interpretation, Sebanc et al. (2003)foundthat,inthecon-
text of a competitive task, high-ranking preschool girls who
controlled more resources and who adopted a dominant
behavior in doing so (e.g., issuing more commands) were
less accepted by peers, particularly by girls, than lower-
ranking girls, whereas high-ranking boys were more ac-
cepted by peers. In Experiments 1a and 1b, a positive gen-
der bias thus may have yielded the same patterns of re-
sponse among boys and girls, but for different reasons.
Although girls may not see their own gender category as
the dominant one, research shows that they do, however, hold
high opinions, as do boys, of their own dominance status. This
pattern was originally found in early work on dominance per-
ception in which participants had to compare themselves with
each of their peers for toughness (e.g., Who is the toughest?
You o r K e v in? ;EdelmanandOmark1973;Omarkand
Edelman 1975; Sluckin and Smith 1977; Strayer et al.
1978). Preschoolers largely disagreed with their peers because
they frequently answered that the tougher individual was
themselves and thus tended to overrate their status. More
recently, using a rope task that measures subjective status
through power and wealth, Mandalaywala et al. (2019) found
that boys and girls, from 3.5- to 6.5-years-old, viewed their
own status as particularly elevated because they scored almost
at ceiling on this task.
Charafeddine et al. (2019)alsoobservedevidenceof
such optimistic views in a self-identification task. They
presented to French and Japanese preschoolers the same
picture as in Experiments 1 and 2 and found that French
preschoolers, but not Japanese, were more likely to identify
with the dominant character, with no difference between
girls and boys. In that experiment, the gender of the other
character on the picture was not specified so that partici-
pants were not lead to imagine a dominance interaction
between themselves and a different-gender child.
However, in studies in which school-age children had to
compare themselves for toughness, girl-boy dyads elicited
a relatively high level of agreement among children (1st to
4th grade). The boy tended to be recognized as the tougher
individual by both the boy and the girl of the dyad, whereas
same-gender dyads produced a lower level of agreement
(Edelman and Omark 1973; Omark and Edelman 1975).
Experiment 2 addresses this issue with preschoolers and
through the same power relation presented in Experiments
1a and 1b. It thus investigates whether the male-power as-
sociation observed among boys and girls affects their per-
ceptions of their own power status.
Experiment 2
In Experiment 2, participants saw the same picture as in
Experiments 1a and 1b, but instead of assigning a gender to
each character, they had to identify themselves with one of the
two characters and were informed that the other character was
either a girl or a boy. Hence, the other character could be of the
same gender as themselves (same-gender condition) or could
be of a different gender (different-gender condition). In the
same-gender condition, we predicted that girls and boys
would identify more with the dominant character, and we
did not expect a difference between girls and boys because
both previously exhibited a similar tendency in an analogous
sample (Charafeddine et al. 2019).
In the different-gender conditions, three predictions from
three alternative hypotheses were tested. First, if the self-
enhancement bias trumps the male-power association, then
both girls and boys would identify with the dominant charac-
ter and would do so to a similar extent. Second, if the male-
power association trumps the self-enhancement bias, then
boys would identify with the dominant character to the same
extent as girls would identify with the subordinate one. Third,
if both factors operate, boys would be more likely than girls to
identify with the dominant character, but girls would identify
with the subordinate to a lesser extent than boys would iden-
tify with the dominant one. Given that 3-year-old children did
not associate power with a male character in Experiments 1a
and 1b and given that they are less likely to infer power from
postures, we included only 4- and 5-year-old children in
Experiment 2.
Sex Roles
Author's personal copy
Fully 160 French children participated in this experiment
= 60.9 mo., SD = 6.47) and were divided into two age
groups: 4-year-olds (80 children, M
= 55.3 mo., SD = 3)
and 5-year-olds (80 children, M
= 66.6 mo., SD = 3.28).
The 4-year-old group was composed of 40 girls (M
= 54.4
mo., SD = 3.23) and 40 boys (M
= 56.1 mo., SD = 2.42),
and the 5-year-old group was composed of 40 girls (M
66.3 mo., SD = 2.98) and 40 boys (M
= 66.8 mo., SD =
3.57). Children were tested in two private (52% of children)
and three public (48% of children) schools in Lyons suburban
area. The two private schools included children of middle to
high SES families. The public schools were located in three
different neighborhoods of the same city: one in high middle-
class district (21.2%), one in a mixed neighborhood (17%),
and one in a low SES neighborhood (10%).
Materials and procedure
The material was the same as in Experiments 1a and 1b. The
experiment had two parts. The first part was identical to
Experiment 1 and included the same two comprehension
questions. In the second part, participants had to identify
themselves with the dominant or subordinate character.
Children were told that one of the characters was themselves
and that the other character was a girl or a boy. Compared to
Experiments 1a and 1b, the experimenter provided more de-
tails to allow the children to better project themselves in a
situation that is similar to their daily interactions with other
The instructions began:
See? I brought this picture to show it to you, can you see
what's on it? Those are children can you see them?
Show them to me. Well what are these children doing:
are they swimming? Are they dancing? Are they
playing? Are they talking? You know what? Those chil-
dren are talking together, and I can hear what they're
saying to each other. Do you want me to tell you?
Yes? Well: One of the children tells the other: You
must do everything I say! Do whatever I want!And the
other child is saying: OK! I will do what you want.
The experimenter next asked the same two comprehension
questions (with their order counterbalanced: Comprehension
questions (order counterbalanced): Who is saying Do what I
want?’”;Who is saying OK! I will do what you want?’” In
contrast to the previous studies which focused on identifying
the charactersgender, the experimenter continued: Yo u
know what? You are on this picture! In fact, on this picture
there is you and a boy [girl].The experimenter concluded by
asking the two counterbalanced self-identification questions:
Now tell me: Which one in this picture is you and which one
is the [other] boy [girl].(The adjective otherwas added in
the same-gender condition.) The order in which the participant
(you) and the other child were mentioned was
counterbalanced across participants. Half of the children were
randomly assigned to the same-gender condition, and the oth-
er half was assigned to the different-gender condition.
Participants were interviewed by the same French male exper-
imenter who was unaware of the experiments hypotheses.
Fully 146 children correctly matched postures with domi-
nance and subordination statements (91.25%, p< .001, 95%
CI [85.75,95.13]). This proportion did not significantly differ
according to childrens age, χ
(1, n= 160) = 1.956, p= .161,
or gender, χ
(1, n= 160) = .0783, p= .779. The following
analyses only include the 146 children who correctly matched
postures with dominance and subordination statements.
In the same-gender condition, children identified more with
the dominant than with the subordinate character (78.57%, p
<.001, 95% CI [67.13, 87.48]). This effect was found both for
girls (75%, p= .004, 95% CI [57.79, 87.88]) and boys
(82.35% p< .001, 95% CI [65.47, 93.23]), and there was no
significant difference between girls and boys, χ
(1, n=
70) = .209, p= .647. There was also no significant difference
between the two age groups, χ
(1, n= 70) = .063, p= .802. In
the different-gender condition, children identified more with
the dominant than with the subordinate character (75%,
p< .001, 95% CI [63.74, 84.22]). The identification with the
dominant character was significant for boys (86.68%,
p< .001, 95% CI [71.91, 95.58]) but not for girls (63.15%,
p= .143; 95% CI [45.99, 78.18]). Girls identified with the
dominant character to a lesser extent than boys, χ
(1, n=
76) = 4.491, p= .034, CramersV= .24. However, the propor-
tion of girls who identified with the subordinate character was
significantly lower than the proportion of boys who identified
with the dominant character, χ
(1, n= 76) =18.066, p< .001,
CramersV= .538. Finally, there was no significant difference
between the two age groups, χ
(1, n= 76) = .158, p= .691.
Three conclusions can be drawn from our results. First, in the
same-gender condition, girls and boys showed a similar self-
enhancement bias. This suggests that being dominant in an
interaction was viewed as a desirable property and echoes
other results indicating that preschoolers value the high-
power position in various types of tasks (Bernard et al.
2016; Castelain et al. 2016; Charafeddine et al. 2016;
Thomas et al. 2018). By identifying themselves with the
Sex Roles
Author's personal copy
high-power character, children exhibit a high opinion of them-
selves. This has been widely documented in the domain of
personality judgments and perception of achievements
(Boseovski 2010;HarterandPike1984; Stipek and Iver
(Charafeddine et al. 2019; Mandalaywala et al. 2019; Omark
and Edelman 1975; Sluckin and Smith 1977; Strayer et al.
1978). Their unrealistic self-evaluations may result from a
confusion between the actual self and the desired self
(Harter 2006) and from the need to maintain or enhance self-
esteem. Some authors argue that positive illusions about the
self are beneficial to mental health because they help individ-
uals deal with negative self-threatening experiences (Taylor
and Brown 1988,1994). In the case of status, Barkow
(1975) postulated a fundamental motive for status that leads
individuals, by ignoring negative status-related information
and focusing on positive information, to evaluate the self as
higher than others. Developmental psychologists also claimed
that it may be adaptive for a young child to overrate his
position within a group and help him or her to face the initially
strange and possibly frightening world of peer-group rela-
tions(Omark et al. 1980, p. 175).
Second, the self-enhancement bias showed a different pat-
tern in the different-gender condition: It was only observed
among boys who identified more with the dominant character
than did girls. It seems therefore that the male-power associa-
tion interfered with childrens self-representation.As shown in
Experiments 1a and 1b, this association was present among
girls, and in Experiment 2 it has presumably modulated their
self-perception in the different-gender condition. This finding
echoes the results of past studies with older participants in
which dyads of children compared themselves for toughness
(Edelman and Omark 1973;OmarkandEdelman1975).
School-age children from girl-boy dyads disagreed less on
who the tougher was than children from same-sex dyads be-
cause girls were less likely to see themselves as tougher than a
boy in a girl-boy dyad. However, toughness may not be the
best dimension to characterize dominance. Pickert and Wall
(1981) actually found that children interpret this notion in
terms of aggressive behavior (e.g., a tough person is stronger
and meaner than others and often fights with others; Pickert
and Wall 1981, p. 78) and found that getting ones own way
was defined more as indicating power or dominance. When
evaluating boystoughness, girls might have thus taken into
consideration their higher aggressiveness and focused less on
relational aspects (Edwards and Whiting 1980; Fabes et al.
2003; La Freniere et al. 1984). In contrast, in our experiments,
dominance was conveyed through a combination of gestures
and verbal statements indicating that one individual prevailed
and gave orders to another regardless of toughness.
Third, the male-power association did not fully override the
self-enhancement bias in the different-gender condition, be-
cause girls did not identify more with the subordinate than
with the dominant character. The comparison of Experiment
2 with Experiment 1 shows how the representation of others
can differ from self-representation. In Experiment 1, girls and
boys associated the dominant posture with a male character,
whereas in the different-gender condition of Experiment 2,
girls no longer made this association. The asymmetry between
childrens expectations about others and about themselves is
relatively common in the literature on gender stereotypes.
Adults and children produce more stereotypical answers when
judging others than when judging themselves, especially
when the stereotype content is negative (Cowan and
Hoffman 1986; Williams and Best 1982).
The self-enhancement bias observed in the current experi-
ment raises questions about the absence of an own-gender bias
among girls in Experiments 1a and 1b because own-gender
favoritism and, more generally, in-group favoritism relies on
the motivation for a positive self-concept (Tajfel and Turner
1979). Moreover, own-gender favoritism is sometimes higher
for girls than for boys (Dunham et al. 2016; Yee and Brown
1994). A possible reasonfor the absence of an own-gender bias
in Experiments 1a and 1b lies in the methodology. Children
were not presented with a picture of a boy or a girl butwith two
nongendered and barely human silhouettes and were informed,
only verbally, that one was a girl and the other was a boy. This
might have been insufficient to activate gender representations
relevant to self-perception and might have hindered identifica-
tion with the characters. Moreover, although giving orders is a
behavior that is common to both dominant boys and dominant
girls (Charlesworth and Dzur 1987; Sebanc et al. 2003), phys-
ical expression of dominance through body postures might be
more common among boys. In Experiment 3, we no longer use
nongendered characters. A girl puppet and boy puppet are first
shown to children and then hidden during the test phase.
Additionally, asymmetry is no longer expressed through body
postures but through verbal statements only.
Experiment 3
In Experiment 3, participants still had to determine which of a
male or female character was in a powerful position, but they
initially saw each of these characters through a girl or a boy
puppet. A dialogue took place between the two puppets be-
hind a concealing board during which one puppet held more
power than the other. Children had to guess which puppet (the
girl or the boy) occupied the powerful position and which one
occupied the subordinate position. Children faced two situa-
tions. One was similar to Experiments 12, as the powerful
puppet imposes her/his will to the other while the subordinate
puppet complies. The second situation involved a puppet who
had more resources than the other. We choose this situation
because preschoolers and even infants consider asymmetry in
resources as a cue to power (Charafeddine et al. 2015; Enright
et al. 2017; Gülgöz and Gelman 2017). Moreover, using
Sex Roles
Author's personal copy
resource asymmetry offers a way to make cross-cultural com-
parisons that fit more with the gender inequality indexes used
in Experiment 1b, namely, the GGGI and the GII, because
these indexes rely on measures of economic power. In
Experiment 3, we compare a group of French children with
a group of Lebanese children. We therefore expect that
Lebanese children would be more likely than French children
to associate higher resources with the boy puppet. Pilot testing
of this paradigm with 3-year-old children showed that children
of this age group had difficulty associating the dialogues with
invisible speakers as assessed by the experiments preliminary
questions. This piloting led us to conduct the experiment with
only 4- and 5-year-old children.
Fully 213 children (M
= 59.5 mo., SD = 6.97) from France
and Lebanon participated in this experiment. All children
returned written parental consent to their teachers prior to
testing. The proportion of girls did not significantly differ
across the two groups (47.22% of the French sample (n=
51) and 57.14% of the Lebanese sample (n= 60)), χ
(1, n=
213) = 2.09 p= .147. In France, 108 children (M
= 60.4
mo., SD = 6.98) were divided into two age groups: 4-year-
olds (54 children, M
= 54.1 mo., SD = 3.05) and 5-year-
olds (54 children, M
= 66.6 mo., SD = 3.18). The 4-year-
old group was composed of 24 girls (M
= 54.7 mo., SD =
2.86) and 30 boys (M
= 53.7 mo., SD = 3.17), and the 5-
year-old group was composed of 27 girls (M
= 67.5 mo.,
SD = 2.85) and 27 boys (M
= 65.7 mo., SD = 3.29).
Children came from two public schools in Lyons periphery
and one public school in its suburban area. The inhabitants of
these neighborhoods are mixed between medium and low
SES families.
In Lebanon, 105 children (M
= 58.7 mo., SD = 6.89)
were divided into two age groups: 4-year-olds (55 children,
= 52.8 mo., SD = 3.41) and 5-year-olds (50 children,
= 65.1 mo., SD = 2.76). The 4-year-old group was com-
posed of 30 girls (M
= 52.7 mo., SD = 3.37) and 25 boys
= 53 mo., SD = 3.54), and the 5-year-old group was
composed of 30 girls (M
= 64.8 mo., SD = 2.51) and 20
boys (M
= 65.6 mo., SD = 3.11). Children attended a pri-
vate Evangelical mission school in the city of Tyre (South
Lebanon governate). Students in this school are of mixed re-
ligious backgrounds, although a majority is of Muslim Shiite
background and predominantly middle-income families.
Materials and procedure
Children heard a sequence of dialogues between a girl and a
boy puppet that took place behind a board. Throughout the
dialogues, participants had to guess which puppet said what.
The main experimenter was standing behind the board with a
puppet in each hand (the position of the puppets was
counterbalanced across participants). Another experimenter
was seated beside the child. In Lebanon, the experimenters
were both women, whereas in France, two mixed-gender pairs
ran the experiments, alternating their roles. All experimenters
followed the same standardized procedure. The experimenter
who stood behind the board was the one who played the sce-
nario and asked the questions. The other experimenter ensured
that the child was paying attention during the dialogues and
took note of the childrens answers. Participants were told that
the two puppets, which were designated as two children, will
talk to each other and will play behind the board. They were
informed that it would then not be possible to see them. To
assess whether participants understood the procedure, the pup-
pets disappeared behind the board during the instruction phase
and asked the participant the first preliminary questions: Can
you see us now? But can you hear us?Before each dialogue
was played, the puppets were visible and started speaking. At
some point, the puppets and the main experimenter disap-
peared behind the board to state the critical utterances. When
the dialogue was over, they reappeared, and the experimenter
asked the child who was saying what.
Six dialogues were presented in two phases: a training
phase and an experimental phase. The goal of the training
phase was to familiarize children with the method of asking
questions about the puppetsutterances. The training phases
included four dialogues presented in a fixed order, each
followed by a question of the type Who said what?
During the first two dialogues, each puppet provided explicit
information about her/his gender. In the first one, as the pup-
pets were behind the board together with the experimenter,
one said, Iamgirland the other said Iamboy
(counterbalanced order). Then, the puppets and the experi-
menter appeared above the board, and children had to answer
the question Who said, I am a girland who said, I am a
Following the same procedure, in the second dialogue, a
puppet said, My name is Sami/Sam(Samifor Lebanese
children and Samfor French children, which are male
names in both languages), and the other puppet said, My
name is Lea(a female name in both languages). The next
two dialogues were not explicit about the puppetsgender
but referred to typical preferences associated with girls or
boys, which are known to children (Cunningham and
Macrae 2011;LoBueandDeLoache2011;Weisgram
et al. 2014). In addition to familiarizing children with the
task, these dialogues served to activate childrens knowl-
edge about gender differences in attitudes and behaviors. In
particular, during the third dialogue, the puppets said that
they like color pencils. Then, one of them said Ireallylike
pink,whereas the other said I really like blue
Sex Roles
Author's personal copy
(counterbalanced order). During the fourth dialogue, in
which the puppets said that they liked to play with their
toys, a puppet said, I like playing with Barbie dolls,and
the other said I like playing with motorcycles
(counterbalanced order).
In the testing phase, two dialogues involved power-related
situations and were presented in a counterbalanced order. In
the hide-and-seek dialogue, a conflict arises, and one of the
puppets imposes its will to the other who complies. The dia-
logue, with the puppets still visible and talking to the child,
was as follows:
Now we are going to play hide-and-seek. Do you like to
play hide-and-seek? We love to play hide-and-seek. We
are going to play hide-and-seek (this sentence was re-
peated twice, and the puppets disappeared behind the
[Puppets behind the board]
Puppet 1: I count.
Puppet 2: Oh no, I count, I like to count.
Puppet 1: So what? I decide. I like to count too, and I
decide. I choose who counts and who doesnt. I choose
Puppet 2: Ah ok, so I choose nothing.
Puppet 1: Yes, thats it. And then when this game is over,
I will choose what other games we will play. I choose
Puppet 2: Oh, OK, I choose nothing.
The experimenter then asks the child (in counterbalanced or-
der): Who did say I choose everything, I decide everything’”
and Who did say I choose nothing; I decide nothing?In the
resource asymmetry dialogue, puppets announce their
amounts of money in a context where they want to buy ice
cream. The dialogue (with the puppets still visible and talking
to the child) was as follows:
We are going to buy ice creamHum, here comes the
ice-cream seller, we have to catch up with him. The ice-
cream seller! The ice-cream seller! We want to buy ice
[Puppets disappear behind the board]:
Puppet 1: I have 10 pennies in my pocket, I have a lot of
Puppet 2: I have 3 pennies in my pocket, I dont have a
lot of money.
Puppet 1: I have a lot of money.
Puppet 2: I dont have a lot of money.
In French, we used the word souswhich is a casual and
generic word for money (uncountable usage) and monetary
units (countable usage). In Lebanon, we used 1000 pound
units (3000 pounds versus 10,000 pounds), which is the unit
that is most frequently used in the country. After the dialogue,
the experimenter asked the child (in counterbalanced order):
Who said, I have a lot of money’” and Who said, I dont
have a lot of money?’”
In the training phase, all children correctly answered the ques-
tions about the puppets genders and names and thus under-
stood the task consisting of assigning the puppetsgender to
the puppetsstatements. Moreover, 92% (n= 196) of the chil-
dren indicated that the girl puppet liked pink and the boy
puppet liked blue. Finally, 98% (n= 209) of children indicated
that the girl puppet liked to play with Barbie dolls whereas the
boy puppet liked to play with the motorbike. These results
indicate that children displayed standard gender stereotypes
regarding toys and colors.
We now turn to the testing phase. In the hide-and-seek
situation, 58.21% (n= 126) of children indicated that it was
the boy puppet who said I choose everythingand the girl
puppet who said Ichoosenothing(p=.0196, 95% CI
[51.28,64.91]). There was no significant difference between
Lebanese and French children (58.09% vs. 58.33%), χ
n=213)=.001, p= .971, or between 4- and 5-year-olds
(57.79% vs. 58.65%), χ
[1, n= 213] = .002, p= .899, but
there was a significant effect of participantsgender because
boys were more likely to attribute the dominant position to the
boy puppet than were girls (73.52% vs. 44.14%), χ
(1, n=
213) = 18.87, p< .001, CramersV= .29. Among boys, the
tendency to attribute the dominant position to the boy puppet
significantly differed from chance (p< .001) but this was not
the case among girls (p= .254). More specifically, in Lebanon,
41.66% of girls associated the dominant position with the boy
puppet (p= .245, 95% CI [44.88,70.93]), and 80% of the boys
did so (p< .001, 95% CI [65,90]). In France, 47.06% of girls
associated the dominant position with the boy puppet
(p= .779, 95% CI [38.46,67.07]) and 68.42% of the boys
did so (p= .007, 95% CI [55.76, 80.09]). In France, where
experimenters could be male or female, the gender of the
experimenter had no significant effect: 55.81% of the children
who heard a male voice made the male-power association
Sex Roles
Author's personal copy
versus 60% of the children who heard a female voice, χ
n= 108) = .187, p=.666.
In the resource asymmetry situation, the results showed a
similar pattern. Overall, 60.56% (n= 129) of the children in-
dicated that it was the boy puppet who said, I have a lot of
moneyand the girl puppet who said, I dont have a lot of
money(p= .002, 95% CI [53.65,67.17]). There was no sig-
nificant difference between Lebanese and French children
(57.14% vs. 63.89%), χ
(1, n= 213) = 1.014, p= .314, or
between 4- and 5-year-olds (57.8% vs. 63.46%), χ
(1, n=
213) = .715, p= .398, but there was a significant effect of par-
ticipantsgender because boys were more likely to attribute
the dominant position to the boy puppet than were girls
(72.55% vs. 49.55%), χ
(1, n=213)=18.869, p<.001,
CramersV= .23. The tendency to attribute the dominant po-
sition to the boy puppet significantly differed from chance
among boys (p< .001, 95% CI [62.82, 80.92]) but not among
girls (p=.925, 95% CI [39.92, 59.20]. More specifically, in
Lebanon, 48.33% of the girls (p=.897, 95% CI
[38.39,64.77]) and 68.89% of the boys (p= .016, 95% CI
[53.35, 81.81]) judged that the boy puppet had more money.
In France, 50.98% of the girls (p=.912, 95% CI
[36.59,65.24]) and 75.44% of the boys (p< .001, 95% CI
[62.24, 85.87]) judged that the boy puppet was richer.
Again, the gender of the experimenter had no effect: 58.14%
of the children who heard a male voice made the male-power
association versus 67.69% of the children who heard a female
voice, χ
(1, n= 108) = 1.023, p= .312).
The results show different response patterns for boys and girls
in both scenarios. While a majority of boys associated the boy
character with the dominant position, girlsresponses were at
chance levels. In light of previous findings indicating that
boys and girls have the same ability to detect hierarchical
relations (Brey and Shutts 2015; Charafeddine et al. 2015;
Gülgöz and Gelman 2017), it is implausible that girls an-
swered randomly. The most likely interpretation is that the
male-power association was dampened by an own-gender bias
that led some girls to attribute the dominant position to the girl
Overall the results of Experiment 3 indicate that preschool
children use gender as a cue to power, not only when one
agent imposes their will on another but also when one indi-
vidual controls more resources. This result contrasts with re-
search reporting that children do not use gender to predict a
wealth asymmetry (Mandalaywala et al. 2019; Olson et al.
2012). In those studies, participants had to assess who be-
tween a female or a male target lives in a high-value house
and who lives in a low-value house. As the authors of these
studies themselves note, children are unlikely to see differ-
ences between males and females with respect to housing
conditions, whereas this dimension is more relevant to predict
status differences between racial groups. Living in a house or
taking advantage of other lifestyle-related goods like cars,
largely concerns family units composed of both male and
female members, which may explain why children do not
relate this type of wealth to a specific gender.
In the current experiment, the asymmetry operates at a
different level. The resources do not concern living conditions
but involve the possibility of achieving a specific goal, namely
the purchase of ice-cream, which is explicitly presented as
relevant to the puppets and which is probably relevant to the
participants as well. The relevance of the goods may influence
how children relate asymmetry of resources to gender. For
example, in Mandalaywala et al. (2019), children (from 3.5
to 6.5 years-old) did not associate gender with wealth in the
house-matching task described previously, but they did so
when the resources were specifically relevant to children
(i.e. having lots of toys and new clothes; always getting to
pick the tasks that everyone else plays at recess and the snacks
that everyone else eats at snack time). Participants associated a
male child target with a higher status than a female child
target, and, as in the current experiment, this association was
stronger for male than female participants. Moreover, in our
task, the asymmetry of resources might be more directly relat-
ed to the notion of power because of the interactive context in
which it appears. Specifically, although the two puppets have
the same goal (to buy ice cream), the richer puppet has more
power to gain access to the goods and therefore to control the
situation. The association between resources and gender in the
current experiment may be influenced by the power dynamics
that stem directly from it. Finally, as in Experiment 1b, we
found no difference between children from the two countries,
even though the resource asymmetry scenario fitted more with
indexes of gender inequality.
General Discussion
Through a series of four experiments, we investigated chil-
drens representations of the link between gender and power
that led to three main findings. First, children were overall
more likely to associate power with a male than with a female
character. This was found in several age groups, in several
countries (France, Lebanon, and Norway), and equally among
girls and boys in Experiments 1a and 1b. Moreover,
Experiment 2 showed that the male-power association was
not limited to third-party relations but also influenced how
children, particularly girls, viewed themselves. The second
important result is the difference between girlsand boys
response patterns in some situations. In the different-gender
condition of Experiment 2, boys identified themselves more
with the dominant character than did girls. Moreover, in
Experiment 3, a large majority of male participants considered
Sex Roles
Author's personal copy
that the puppet who imposed its choice and who controlled
more resources was the boy puppet, whereas this was not the
case for female participants. The third main finding is the lack
of difference between participants from societies that differ in
their level of gender inequality (Experiments 1b and 3).
Early Representation of Male Power
The male-power association observedfor children as young as
4 years-old contrasts with earlier studies in which the youn-
gest children (68-year-olds) did not perceive gender inequal-
ity or perceived it to a lower degree than older children (Liben
et al. 2001; Neff et al. 2007). However, those studies relied on
more complex entities, such as occupational interests or job
remuneration (Liben et al. 2001; Weisgram et al. 2010) or used
a methodology that asked children to consider power and
gender in abstract ways (men and women as general entities;
Neff et al. 2007). In the current experiments, preschoolers had
to consider power and gender through concrete dyadic rela-
tionships between actual characters who displayed explicit
verbal and nonverbal behavior. This methodology, which
had proved to be successful at demonstrating preschoolers
and infantsability to represent dominance and power, prob-
ably facilitated the association between gender and power at
preschool age.
However, Experiments 1a and 1b suggest that the male-
power association does not emerge before age four. This
might be due to less stable representations of power based
on postures for the youngest children (Charafeddine et al.
2019; Terrizzi et al. 2018). More generally, 3-year-old chil-
dren may also be less likely to use gender as a relevant cue to
power. They have poorer experience of mixed-gender interac-
tions than older children, and the cognitive mechanisms that
allow them to make sense of asymmetrical relationships are
less sophisticated (Gülgöz and Gelman 2017,Overand
Carpenter 2015). In addition, the task itself might have been
too difficult for the youngest children because assigning a
specific gender to a non-gendered character is a relatively
abstract process. Finally, another explanation relies on the
predominance of female authority figures in young childrens
educational and domestic environments (i.e., parents and
teachers). Three-year-old children are more dependent on au-
thority figures than older children when they have to make
decisions and obtain relevant information. Given that author-
ity figures are more often women, they might be less likely to
assign higher power to males than older children are.
The Influence of ParticipantsGender
Participantsgender produced mixed results regarding the
male-power association. Whereas Experiments 1a and 1b
showed that girls and boys associated power with a male char-
acter to a similar extent, this was not the case in Experiments 2
and 3. In those latter experiments, the association was stronger
among boys than among girls, who answered at chance levels.
In particular, when implicating the self in the relationship, as
in Experiment 2, girls did not ascribe the powerful position to
a boy character more than to themselves. This similarity most
likely stems from a self-enhancement bias, which was ob-
served in the same gender condition and in other studies
(Charafeddine et al. 2019;EdelmanandOmark1973;
Mandalaywala et al. 2019;OmarkandEdelman1975;
Sluckin and Smith 1977; Strayer et al. 1978). This result there-
fore nuances the view that girls attribute negative value to
female power.
In Experiment 3, although a large majority of boys consid-
ered that the puppet who imposed its choice and who had
more resources was the boy puppet, this was not the case for
girls. This difference can be accounted for by own-gender
favoritism leading some girls to attribute the socially advan-
tageous position to an individual of their own gender. The
gender effect obtained in Experiment 3 echoes that observed
in Experiment 2 because own-gender favoritism partially re-
lies on positive self-views (Tajfel and Turner 1979). In
Experiments 1a and 1b, the absence of own-gender bias might
have resulted from the limited gender-related information that
we provided, whereas in Experiment 3 the gender of the char-
acters was clearly identifiable and gender-related information
was made salient in preliminary questions. The greater sa-
lience of gender may have led a subgroup of the female par-
ticipants to identify with the girl puppet, thus giving her the
advantageous position. However, Experiments 1a and 1b also
involved physical postures that were absent in Experiment 3.
Therefore, it cannot be ruled out that postures hindered the
own-gender bias in girls and strengthened the male-power
association in Experiments 1a and 1b.
Absence of Cultural Differences
Overall, the current experiments included participants from
three different countries but showed no influence of culture.
Although this result does not confirm our prediction, it shows
that the power-masculinity association and the difference be-
tween boys and girls were robust. Our results could reflect that
even in countries where gender equality is fostered at a formal
political level, which is the case in Norway and in France to a
lesser extent, some inter-gender relations still give advanta-
geous positions to men (Baxter and Kane 1995). Moreover,
in countries in which social attitudes emphasize gender equal-
ity, some features of the childrens environment, such as toy
collections, display marked gender roles that are similar to
those in less egalitarian countries (Nelson 2005).
Research has shown that preschoolers from various cultural
backgrounds can have similar gender stereotypes, as reflected
in their toy preferences and play (Turner et al. 1993), as well as
their behavioral and psychological traits (Best et al. 1977;
Sex Roles
Author's personal copy
Williams and Best 1982). For instance, in their wide-range
cross-cultural study, which included 25 countries, Williams
and Best (1982) reported that 5-year-olds associated traits
such as weak and softhearted with female individuals and
aggressiveness and strength with males. Moreover, the devel-
opmental pattern of gender stereotypes was relatively similar
across countries, although the learning rates could differ
(Williams and Best 1982). Typically, in most cultures, stereo-
types emerge before the age of five; they then show a relative-
ly sharp increase during school years and are fully acquired
during adolescence (Williams and Best 1982).
In the current study, children might have been too young to
show an acute sensitivity to the level of gender inequality in
their environment, which could have resulted in intercultural
differences. Given that the power asymmetry between genders
is universal and that physical traits associated with dominance
differ between males and females in a similar way across cul-
tures, children might develop initial and rough representations
that associate power with male figures to the same extent. It
might be only at later ages that gender representations integrate
societal inputs that either foster or discourage power asymme-
try between genders. Indeed, by the end of their elementary
schooling, childrens gender concepts become more flexible
and less normative (Levy et al. 1995; Taylor et al. 2009).
Limitations and Further Research Directions
Although our work is one of the first known to explore how
preschool children represent the relationship between gender
and power in interactive contexts and in different cultures, it
contains several limitations that call for complementary stud-
ies. A first limitation concerns the participantsage range. It
was relatively narrow compared to earlier developmental
work on the representation of gender inequalities, which in-
cluded school-age children, adolescents, and adults (Liben
et al. 2001; Neff et al. 2007; Weisgram et al. 2010). Those
studies found that for more societal dimensions of hierarchy
(e.g., job occupations), the association between high-status
and maleness tends to reinforce with age. In our study we
did not find any difference between the 4 year-olds and 6
7 year-olds, but investigating older children would be helpful
at better characterizing the developmental pathway of power
representations related to gender. On the on hand, because
children expect more equality in relationships as they grow
older (Shaw and Olson 2012), the male-power association
might decline with age. On the other hand, the accumulation
of observations of male-female asymmetric interactions can
result in strengthening their belief in a male power. The age
range issue also concerns younger children. Although we
found no evidence of a male-power association among 3-
year-old children, further research is needed to explore wheth-
er more appropriate stimuli may elicit such an association.
Another limitation concerns the implementation of power
relationships. In Experiments 1 and 2, the power asymmetry
was conveyed through a combination of verbal statements and
physical postures. However, the physical expression of domi-
nance through facial traits, muscular strength, postures, size, or
voice also correlates with physical differences between males
and females. For instance, expanded and erect body postures,
which typically convey dominance, occupy more physical
space than constricted postures, but this is also the case of male
physical behavior and male body size as compared to female
ones. In Experiments 1 and 2, although the two characters are
of the same size, the dominant character has its arm and its
finger directed toward the subordinate and has its head up. In
contrast, the subordinate occupies less space because its arm is
not visible, and its head is down. This may give the impression
that the subordinate is smaller. Some participants could thus
have based their gender assignment only on perceptual cues
without considering the power difference in the interaction.
However, some results are not consistent with this possibility.
In Experiments 1a and 1b children who failed to answer the
comprehension questions were at chance level when
responding to the gender attribution questions. This finding
suggests that, for these children, the perceptual features were
not sufficient to infer maleness from the erect-body character.
A related point that needs to be explored is the possibility
that some expressions of power are more associated to one
gender than to another. Although in our task the erect-body
posture differs from typical aggressive behavior such as hit-
ting, pushing or threatening, the dominant character may look
aggressive because of its antisocial behavior: the dominant
imposes its will on the subordinate who has no choice but to
obey. Adopting an antisocial behavior is often a consequence
of exercising power (Guinote et al. 2015; Lammers and Stapel
2011). Preschoolers are sensitive to this dimension because
they are more likely to infer power from antisocial and malev-
olent behavior (e.g., mean, unhelpful) than from pro-social and
benevolent behavior (Gülgöz and Gelman 2017; Terrizzi et al.
2019). This aspect of power may guide children in their gender
assignation. In particular, if they tend to view boys as more
physically aggressive than girls (Giles and Heyman 2005),
they might infer that the powerful individual is a boy. Hence,
future research should more systematically explore how chil-
dren associate various expressions of power with gender. In
particular, if preschoolers use aggressiveness to assign power
to boys, they might be less likely to do so when power mani-
fests itself in a more benevolent form. One difficulty, however,
is that preschoolers may have more trouble inferring power
from benevolent behavior (Gülgöz and Gelman 2017).
Another limitation concerns cross-cultural comparisons.
The societies we have chosen may not be the best at bringing
out cross-cultural differences in children of the ages we stud-
ied. The criteria of gender inequality that we took into account
are on a societal level and do not necessarily capture what
Sex Roles
Author's personal copy
happens on a more local level or in specific culturally embed-
ded interactions. Microscale societies that emphasize the im-
portance of women in valued social areas would be worthy of
study, such as matrilineal societies in which property, titles,
land or family names can be passed down through the lineage
of the mother. Interestingly, in women from matrilineal soci-
eties, the attitude toward aspects of social hierarchies, such as
competition or economic resources, resembles that of men
from more patriarchal societies. For instance, it has been ob-
served that among the Khasi, women are more likely to take
part in competitive activities than men (Gneezy et al. 2009).
Similarly, among the Mosuo, in which women are in charge of
economic power, a study found that women shared their mon-
etary resources in the dictator game half the time compared to
men (Gong et al. 2015), whereas the opposite pattern has been
observed in western societies (Eckel and Grossman 1998;
Selten and Ockenfels 1998). Moreover, within western socie-
ties, educational practices, at school or in the family circle,
which promote gender equality and/or gender neutrality, can
also have an impact on childrens representations of gender.
Indeed, gender-neutral pedagogy tends to reduce gender
stereotyping among preschoolers (Shutts et al. 2017) com-
pared to typical pedagogy, whereas school environments that
emphasize gender distinctions increase gender stereotypes and
negative attitudes toward different-gender peers (Hilliard and
Liben 2010).
Finally, future research could use finer ways to code par-
ticipantsgender. In our study, as in most studies on childrens
social cognition, binary gender categories are used that direct-
ly correspond to participantssex. Finer grained results could
be obtained by using childrens felt gender identity as an inter-
subject variable (Bem 1981; Martin et al. 2017).
Practice Implications
A first implication of the current research is that kindergarten
schools and educational institutions not only should be re-
sponsive to gendered divisions based on groups, which typi-
cally associate distinct toys, activities, and linguistic labelling
to distinct genders, but also should consider the relational
dynamics inherent to gender. Not only do children have dif-
ferent expectations about the psychological traits of males and
females, but they also have expectations about the outcomes
of interactions in cross-gender interactions, as demonstrated
by the current study. From the perspective of intervention,
gender inequality should thus also encompass girl-boy power
dynamics and their representation of a gendered power.
A second implication is that the current work can inform
educational practices about the age at which gender-based
power relations begin to play an important role in childrens
social expectations and self-construal. Our studies show that
children understand gender inequality in its relational mani-
festation before they understand gender differences in social
status. By using power relations as a manifestation of gender
inequality, educational programs for gender equality can target
children since their first preschool years. Moreover, interven-
tion programs can use childrens abilities to infer power from a
wide range of cues to present inter-gender interactions that go
against stereotypical male dominance. Finally, thinking from a
first-person perspective seems to incite girls to moderate their
expectations of male power. In this regard, implicating the
self, by inviting children to act in fictitious social power situ-
ations or through explicit identification with third party char-
acters, may prove fruitful.
Gender hierarchy relies on power and produces important
power asymmetries in the social world. The current study used
asymmetric power interactions to examine childrens repre-
sentations of cross-gender relations and to show early sensi-
tivity to a hierarchy between males and females. The social
inferences that children draw from gender categories is a cen-
tral issue for developmental psychology because of its obvi-
ous implication in social prejudices and in childrens concep-
tions of their social selves. As Erving Goffman (1977, p. 301)
put it: In modern industrial society, as apparently in all others,
sex is at the base of a fundamental code in accordance with
which social interactions and social structures are built up, a
code which also establishes the conceptions individuals have
concerning their fundamental human nature.The current
study and the corresponding line of research should contribute
to a more fine-grained understanding of the way gender power
relations are used and reproduced in development.
Acknowledgements Rawan Charafeddine Laboratoire Langage Cerveau
Cognition, Institut des Sciences Cognitives Marc Jeannerod; Imac Maria
Zambrana, Faculty of Educational Science, Department of Special Needs
Education, University of Oslo; Benoît Triniol, Laboratoire Langage
Cerveau Cognition, Institut des Sciences Cognitives Marc Jeannerod;
Hugo Mercier, Laboratoire Langage Cerveau Cognition, Institut des
Sciences Cognitives Marc Jeannerod; Fabrice Clément, Centre de
Sciences Cognitives, Université de Neuchâtel; Laurence Kaufmann,
Institut des Sciences Sociales, Faculté des Sciences Sociales et
Politiques, Lausanne University; Anne Reboul, Laboratoire Langage
Cerveau Cognition, Institut des Sciences Cognitives Marc Jeannerod;
Francisco Pons, Faculty of Social Sciences, Department of Psychology,
University of Oslo; Jean-Baptiste Van der Henst, Laboratoire Langage
Cerveau Cognition, Institut des Sciences Cognitives Marc Jeannerod
We wi s h to tha n k Emil y H e wit t , B éra n g ère Tes t ud, Ma y s san
Charafeddine, Léonore Ferrer Catala and Antoine Danielou for help with
data collection. We also thank the participating children and their parents,
the schools and Inspection Académique de Lyon, in particular Sylvie
Coustier, Stéphane Garapon, Vincent Guili and Michèle Prieur, for mak-
ing this research possible. We thank Gloria Origgi for her valuable input;
Justine Epinat, Thomas Castelain, Audrey Breton, Thomas Charavet-
Gomel and Ira Noveck for their support and two anonymous reviewers
for their helpful comments. This research was supported by a grant from
Fondation de France and by CNRS (Défi Genre) awarded to the last
Sex Roles
Author's personal copy
Agresti, A. (2002). Categorical data analysis (2nd ed.). New York:
Albert, A. A., & Porter, J. R. (1986). Childrens gender role stereotypes:
A comparison of the United States and South Africa. Journal of
Cross-Cultural Psychology, 17(1), 4565.
Anderson, K. J., & Leaper, C. (1998). Meta-analyses of gender effectson
conversational interruption: Who, what, when, where, and how. Sex
Roles, 39(34), 225252.
Barkow, J. H. (1975). Prestige and culture: A biosocial interpretation.
Current Anthropology, 16(4), 553563.
Baxter, J., & Kane, E. W. (1995). Dependence and independence: A
cross-national analysis of gender inequality and gender attitudes.
Gen de r & Soci et y, 9(2), 193215.
Bem, S. L. (1981). Gender schema theory: A cognitive account of sex
typing. Psychological Review, 88(4), 354364.
Bernard, S., Castelain, T., Mercier, H., Kaufmann, L., Van der Henst, J.-
B., & Clément, F. (2016). The boss is always right: Preschoolers
endorse the testimony of a dominant over that of a subordinate.
Journal of Experimental Child Psychology, 152, 307317. https://
Best, D. L., Williams, J. E., Cloud, J. M., Davis, S. W., Robertson, L. S.,
Edwards, J. R., Fowles, J. (1977). Development of sex-trait ste-
reotypes among young children in the United States, England, and
Ireland. Child Development, 48, 13751384.
Bianchi, S. M., Milkie, M. A., Sayer, L. C., & Robinson, J. P. (2000). Is
anyone doing the housework? Trends in the gender division of
household labor. Social Forces, 79(1), 191228.
Bird, R. (1999). Cooperation and conflict: The behavioral ecology of the
sexual division of labor. Evolutionary Anthropology, 8(2), 6575.;2-v.
Birnbaum, D. W., & Chemelski, B. E. (1984). Preschoolersinferences
about gender and emotion: The mediation of emotionality stereo-
types. Sex Roles, 10(78), 505511.
Birnbaum, D. W., Nosanchuk, T. A., & Croll, W. L. (1980). Childrens
stereotypes about sex differences in emotionality. Sex Roles, 6(3),
Boseovski, J. J. (2010). Evidence for rose-colored glasses: An exami-
nation of the positivity bias in young childrens personality judg-
ments. Child Development Perspectives, 4(3), 212218. https://doi.
Brey, E., & Shutts, K. (2015). Children use nonverbal cues to make
inferences about social power. Child Development, 86(1), 276
Brightman, R. (1996). The sexual division of foraging labor: Biology,
taboo, and gender politics. Comparative Studies in Society and
History, 38(4), 687729.
Bussey, K., & Bandura, A. (1999). Social cognitive theory of gender
development and differentiation. Psychological Review, 106(4),
Carli, L. L. (1999). Gender, interpersonal power, and social influence.
Journal of Social Issues, 55(1), 8199.
Cashdan, E. (1998). Smiles, speech, and body posture: How women and
men display sociometric status and power. Journal of Nonverbal
Behavior, 22(4), 209228.
Castelain,T., Bernard, S., Van der Henst, J. B., & Mercier, H. (2016). The
influence of power and reason on young Maya childrens endorse-
ment of testimony. Developmental Science, 19(6), 957966. https://
Charafeddine, R., Mercier, H., Kaufman, L., Clement, F., Berchtold, A.,
Reboul, A., Van Der Henst, J.-B. (2015). How preschoolers use
cues of dominance to make sense of their social environment.
Journal of Cognition and Development, 16(4), 587604. https://
Charafeddine, R., Mercier, H., Clément, F., Kaufmann, L., Reboul, A., &
Van Der Henst, J.-B. (2016). Childrens allocation of resources in
social dominance situations. Developmental Psychology, 52(11),
Charafeddine, R., Yamada, Y., Matsui, T., Sudo, M., Germain, P.,
Bernard, S., Van Der Henst, J.-B. (2019). Cross-cultural differ-
ences in the valuing of dominance by young children. Journal of
Cognition and Culture, 19(34), 256272.
Charlesworth, W. R., & Dzur, C. (1987). Gender comparisons of pre-
schoolersbehavior and resource utilization in group problem solv-
ing. Child Development, 58(1), 191200.
Clément, F., Harris, P., Bernard, S., Antonietti, J.-P., & Kaufmann, L.
(2014). Rousseaus child. Swiss Journal of Psychology, 73(2),
Cogsdill, E. J., Todorov, A. T., Spelke, E. S., & Banaji, M. R. (2014).
Inferring character from faces. Psychological Science, 25(5), 1132
Cowan, G., & Hoffman, C. D. (1986). Gender stereotyping in young
children: Evidence to support a concept-learning approach. Sex
Roles, 14(34), 211224.
Crick, N. R., & Grotpeter, J. K. (1995). Relational aggression, gender, and
social-psychological adjustment. Child Development, 66(3), 710
Crick, N. R., Casas, J. F., & Mosher, M. (1997). Relational and overt
aggression in preschool. Developmental Psychology, 33(4), 579
Cunningham, S. J., & Macrae, C. N. (2011). The colour of gender
stereotyping. British Journal of Psychology, 102(3), 598614.
Darmangeat, C. (2009). Le communisme primitif nest plus ce quil était :
Aux origines de loppression des femmes [primitive communism is
no longer what it was: The origins of womens oppression].
Toulouse: Smolny.
Dovidio, J. F., Ellyson, S. L., Keating, C. F., Heltman, K., & Brown, C.E.
(1988). The relationship of social power to visual displays of dom-
inance between men and women. Journal of Personality and Social
Psychology, 54(2), 233242.
Dunham, Y., Baron, A. S., & Banaji, M. R. (2016). The development of
implicit gender attitudes. Developmental Science, 19(5), 781789.
Eagly, A. H., Makhijani, M. G., & Klonsky, B. G. (1992). Gender and the
evaluation of leaders: A meta-analysis. Psychological Bulletin,
111(1), 322.
Eckel, C. C., & Grossman, P. J. (1998). Are women less selfish than
men?: Evidence from dictator experiments. The Economic
Journal, 108(448), 726735.
Edelman, M. S., & Omark, D. R. (1973). Dominance hierarchies in young
children. Social Science Information, 12(1), 103110. https://doi.
Edwards, C.P., & Whiting, B. B. (1980). Differential socialization of girls
and boys in light of cross-cultural research. New Directions for Child
Sex Roles
Author's personal copy
and Adolescent Development, 1980(8), 4557.
Enright, E. A., Gweon,H., & Sommerville, J. A. (2017). To the victor go
the spoils: Infants expect resources to align with dominance struc-
tures. Cognition, 164,821.
Esposito, A. (1979). Sex differences in childrens conversation.
Language and Speech, 22(3), 213220.
Fabes, R. A., Martin, C. L., & Hanish, L. D. (2003). Young childrens
play qualities in same-, other-, and mixed-sex peer groups. Child
Development, 74(3), 921932.
Fagot, B. I., Leinbach, M. D., & Oboyle, C. (1992). Gender labeling,
gender stereotyping, and parenting behaviors. Developmental
Psychology, 28(2), 225230.
Fagot, B. I., Rodgers, C. S., & Leinbach, M. D. (2012). Theories of
gender socialization. In T. Eckes & H. Trautner (Eds.), The devel-
opmental social psychology of gender (pp. 79104). Mahwah:
Lawrence Erlbaum Associates Publishers.https://www.
Fehr, E., Bernhard, H., & Rockenbach, B. (2008). Egalitarianism in
young children. Nature, 454(7208), 10791083.
Fishbein, H. D., & Imai, S. (1993). Preschoolers select playmates on the
basis of gender and race. Journal of Applied Developmental
Psychology, 14(3), 303316.
Fiske, S. T., Cuddy, A. J., Glick, P., & Xu, J. (2002). A model of (often
mixed) stereotype content: Competence and warmth respectively
follow from perceived status and competition. Journal of
Personality and Social Psychology, 82(6), 878902. https://doi.
Fiske, S. T., Cuddy, A. J., & Glick, P. (2007). Universal dimensions of
social cognition: Warmth and competence. Trends in Cognitive
Sciences, 11(2), 7783.
Gazes, R. P., Hampton, R. R., & Lourenco, S. F. (2017). Transitive infer-
ence of social dominance by human infants. Developmental Science,
20(2), e12367.
Giles, J. W., & Heyman, G. D. (2005). Young childrens beliefs about the
relationship between gender and aggressive behavior. Child
Development, 76(1), 107121.
Gneezy, U., Leonard, K. L., & List, J. A. (2009). Gender differences in
competition: Evidence from a matrilineal and a patriarchal society.
Econometrica, 77(5), 16371664.
Goffman, E. (1977). The arrangement between the sexes. Theory and
Society, 4(3), 301331.
Gong, B., Yan, H., & Yang, C.-L. (2015). Gender differences in the
dictator experiment: Evidence from the matrilineal Mosuo and the
patriarchal Yi. Experimental Economics, 18(2), 302313. https://
Guinote, A., Cotzia, I., Sandhu, S., & Siwa, P. (2015). Social status
modulates prosocial behavior and egalitarianism in preschool chil-
dren and adults. Proceedings of the National Academy of Sciences,
112(3), 731736.
Gülgöz, S., & Gelman, S. A. (2017). Whos the boss? Concepts of social
power across development. Child Development, 88(3), 946963.
Halim, M. L., Ruble, D. N., & Tamis-LeMonda, C. S. (2013). Four-year-
olds' beliefs about how others regard males and females. British
Journal of Developmental Psychology, 31(1), 128135. https://doi.
Hall, J. A., Coats, E. J., & LeBeau, L. S. (2005). Nonverbal behavior and
the vertical dimension of social relations: A meta-analysis.
Psychological Bulletin, 131(6), 898924.
Harter, S. (2006). The self. In W. Damon, R. M. Lerner, & N. Eisenberg
(Eds.), Handbook of child psychology: Social, emotional, and per-
sonality development (Vol. 3, 6th ed., pp. 505-570). New Jersey:
Harter, S., & Pike, R. (1984). The pictorial scale of perceived competence
and social acceptance for young children. Child Development,
55(6), 19691982.
Hawley, P. H. (1999). The ontogenesis of social dominance: A strategy-
based evolutionary perspective. Developmental Review, 19(1), 97
Hayden-Thomson, L., Rubin, K. H., & Hymel, S. (1987). Sex preferences
in sociometric choices. Developmental Psychology, 23(4), 558562.
Henl ey, N . (19 77 ). Body politics: Power, sex, and nonverbal
communication. New Jersey: Prentice Hall.
Henley, N. M., & Harmon, S. (1985). The nonverbal semantics of power
and gender: A perceptual study. In S. L. Ellyson & J. F. Dovidio
(Eds.), Power, dominance, and nonverbal behavior (pp. 151164).
New York: Springer.
Hilliard, L. J., & Liben, L. S. (2010). Differing levels of gender salience in
preschool classrooms: Effects on childrens gender attitudes and
intergroup bias. Child Development, 81(6), 17871798. https://doi.
Horne, R. M., Johnson, M. D., Galambos, N. L., & Krahn, H. J. (2018).
Time, money, or gender? Predictors of the division of household
labour across life stages. Sex Roles, 78(11), 731743. https://doi.
Human Rights Watch. (2015). Unequal and unprotected: Womens rights
under Lebanons religious personal status laws.
Keating, C. F., & Bai, D. L. (1986). Childrens attributions of social
dominance from facial cues. Child Development, 57(5), 1269
Kinzler, K. D., Shutts, K., & Correll, J. (2010). Priorities in social cate-
gories. European Journal of Social Psychology, 40(4), 581592.
Kuhn, D., Nash, S. C., & Brucken, L. (1978). Sex role concepts of two-
and three-year-olds. Child Development, 49(2), 445451. https://
La Freniere, P., Strayer, F. F., & Gauthier, R. (1984). The emergence of
same-sex affiliative preferences among preschool peers: A
developmental/ethological perspective. Child Development, 55(5),
Lakoff, R. (1973). Language and womans place. Language in Society,
2(1), 4579.
Lammers, J., & Stapel, D. A. (2011). Power increases dehumanization.
Group Processes & Intergroup Relations, 14(1), 113126. https://
Lane, I. M., & Coon, R. C. (1972). Reward allocation in preschool chil-
dren. Child Development, 43(4), 13821389.
Leinbach, M. D., Hort, B. E., & Fagot, B. I. (1997). Bears are for boys:
Metaphorical associations in young childrens gender stereotypes.
Cognitive Development, 12(1), 107130.
Levy, G. D., Taylor, M. G., & Gelman, S. A. (1995). Traditional and
evaluative aspects of flexibility in gender roles, social conventions,
moral rules, and physical laws. Child Development, 66(2), 515531.
Liben, L. S., Bigler, R. S., & Krogh, H. R. (2001). Pink and blue collar
jobs: Childrens judgments of job status and job aspirations in
Sex Roles
Author's personal copy
relation t o se x of w or ke r. Journal of Experimental Child
Psychology, 79(4), 346363.
LoBue, V., & DeLoache, J. S. (2011). Pretty in pink: The early develop-
ment of gender-stereotyped colour preferences. British Journal of
Developmental Psychology, 29(3), 656667.
LoBue, V., Nishida, T., Chiong, C., DeLoache, J. S., & Haidt, J. (2011).
When getting something good is bad: Even three-year-olds react to
inequality. Social Development, 20(1), 154170.
Lourenco, S. F., Bonny, J. W., & Schwartz, B. L. (2016). Children and
adults use physical size and numerical alliances in third-party judg-
ments of dominance. Frontiers in Psychology, 6, 20502060.
Lukaszewski, A. W., Simmons, Z. L., Anderson, C., & Roney, J. R.
(2016). The role of physical formidability in human social status
allocation. Journal of Personality and Social Psychology, 110(3),
Ma, L., & Woolley, J. D. (2013). Young childrens sensitivity to speaker
gender when learning from others. Journal of Cognition and
Development, 14(1), 100119.
Maccoby, E. E., & Jacklin, C. N. (1987). Gender segregation in child-
hood. Advances in Child Development and Behavior, 20, 239
287. .
Mandalaywala, T. M., Rhodes, M., & Tai, C. (2019). Children's use of
race and gender as cues to social status.
Mansour, Z., & Karam, S. (2012). School-related gender based violence
(SRGBV) in Lebanon.Beirut.UNESCORegionalBureaufor
Education in the Arab States.
Martin, C. L., & Halverson Jr., C. F. (1981). A schematic processing
model of sex typing and stereotyping in children. Child
Development, 52(4), 11191134.
Martin, C. L., Ruble, D. N., & Szkrybalo, J. (2002). Cognitive theories of
early gender development. Psychological Bulletin, 128(6), 903
Martin, C. L., Andrews, N. C., England, D. E., Zosuls,K., & Ruble, D. N.
(2017). A dual identity approach for conceptualizing and measuring
children's gender identity. Child Development, 88(1), 167182.
Mascaro, O., & Csibra, G. (2012). Representation of stable social domi-
nance relations by human infants. Proceedings of the National
Academy of Sciences, 109(18), 68626867.
Murdock, G. P., & Provost, C. (1973). Factors in the division of labor by
sex: A cross-cultural analysis. Ethnology, 12(2), 203225. https://
Neff, K. D., Cooper, C. E., & Woodruff, A. L. (2007). Childrens and
adolescentsdeveloping perceptions of gender inequality. Social
Development, 16(4), 682699.
Nelson, A. (2005). Childrens toy collections in SwedenA less gender-
typed country? Sex Roles, 52(12), 93102.
Olson, K. R., Shutts, K., Kinzler, K. D., & Weisman, K. G. (2012).
Children associate racial groups with wealth: Evidence from South
Africa. Child Development, 83(6), 18841899.
Omark, D. R., & Edelman, M. S. (1975). A comparison of status hierar-
chies in young children: An ethological approach. Social Science
Information, 14(5), 87107.
Omark, D. R., Strayer, F. F., & Freedman, D. G. (1980). Dominance
relations: An ethological viewofhumanconflictandsocial
interaction. New York: Garla