A ship is sinking. Who should save the day and rescue others? Should women be saved first? People’s answers to these questions are often influenced by prescriptive gender role attitudes (e.g., who should do what), yet we know little about how and when these attitudes first appear. Moreover, attitudes about which gender is heroic and which gender is helpless might lay the foundation for benevolent sexism (Glick and Fiske 1996; Glick and Hilt 2000), an insidious and harmful type of sexism that emphasizes patronizing and paternalistic attitudes toward women. Protective paternalism, a subset of benevolent sexist attitudes, encourages men to act as paternalistic rescuers to women, who in turn, require men’s assistance (Glick and Fiske 1996; Glick and Hilt 2000). Benevolent sexism has been linked to decreased cognitive performance, self-esteem, and feelings of competency for women (Dardenne et al. 2007; Dumont et al. 2010), and, in mixed-gender interactions, it has been linked to men helping women in ways that foster dependence on men (Shnabel et al. 2016). Further, benevolent sexism has been found to be associated with decreased support for gender equality (Barreto and Ellemers 2005; Becker and Wright 2011). Thus, understanding more about the origins of these specific gender role attitudes is important.

Benevolent Sexism: Who Is Heroic and Who Is Helpless?

Attitudes about who should engage in heroism and who should be dependent and require special assistance are incorporated within a broader system of sexism that posits that individuals hold ambivalent attitudes toward women. Within this model of ambivalent sexism are hostile sexism, or antagonistic attitudes toward women, and benevolent sexism. Benevolent sexism describes attitudes that men should provide paternalistic protection and care for women who adhere to traditional gender roles. Additionally, it posits that women possess special qualities, like superior morality, that complement men’s qualities and reinforces men’s duty to provide special care to these “good” women. Tying together these attitudes is a component of heterosexual, romantic dependency between women and men (Glick and Fiske 1996). Attitudes that men should provide care, or be heroes, and women should be given special treatment and care, or be helpless, reflect paternalistic components of benevolent sexism.

Only a handful of developmental studies have directly examined benevolent sexism, and this research has been limited to preadolescents and adolescents (e.g., Ferragut et al. 2013; Garaigordobil and Aliri 2012; Silván-Ferrero and López 2007). For example, girls as young as 11 years of age have endorsed benevolent sexism (Montañés et al. 2012). This focus on adolescence is consistent with theory that proposes that during adolescence, increased other-gender interactions and newfound romantic interests foster benevolent sexism (Glick and Hilt 2000). Mixed-gender interactions among children are described as more combative and relevant to hostile sexism. However, although benevolent sexism might not be present in its entirety, such as the emphasis on heterosexual romantic dependency (Glick and Hilt 2000), perhaps more simplistic paternalistic attitudes toward women and girls begin to percolate in early childhood. These ideas may be expressed in children’s attitudes about who should be heroic and who should receive help and special care.

With the advent of forming a gender identity, young children are often fascinated by and motivated to learn about gender, seeking to understand what is considered appropriate or inappropriate (Martin et al. 2002). Influenced by the Piagetian perspective, cognitive theories of gender development propose that children are like miniature scientists: They formulate basic knowledge representations of gender and frequently test these representations within their surroundings. After testing the accuracy of these gender representations, these makeshift scientists are proposed to reevaluate and revise their schemas (thought systems) as necessary. However, these schemas are prone to errors. As a byproduct of an efficient schema, stereotypes and heuristics may come into play when a child decides what is appropriate to do based on one’s gender. Much important literature has discovered the content of children’s gender stereotypes and children’s application of these stereotypes to themselves (for reviews see Huston 1985; Ruble et al. 2006). Research has yet to directly explore children’s gender stereotypes related to benevolent sexism. However, research assessing stereotypes about the appropriate traits and roles for females and males suggest foundational attitudes of benevolent sexism may be present early in childhood.

Early Gender-Trait and Gender Role Stereotypes

Gender-trait stereotypes describing boys as strong but stoic and girls as nice but weak may contribute to benevolent sexist attitudes that males should be rescuers and females should be placed on a pedestal. Indeed, much work suggests that young children are aware of trait stereotypes of girls as “nice” or “friendly” and of boys as “brave” and “strong” (Albert and Porter 1988; Powlishta 1995; Sani and Bennett 2001; Sani et al. 2003; Trautner et al. 2005; Williams et al. 1999). Young children also often rely on their knowledge of gender stereotypes in their judgments of gender stereotype violations (Sinno and Killen 2009). Beyond just a knowledge of these trait stereotypes, there is evidence that young children endorse stereotypes that girls should be gentle but dependent and boys should be brave and strong (Liben and Bigler 2002); however, much less is known about children’s specific attitudes toward who should be heroes and who needs special care. Our research aims to address this gap by describing children’s attitudes toward these specific gender roles across various ages from early childhood through early adolescence. We were particularly interested in children’s attitudes toward who should engage in these specific gender roles (their prescriptive attitudes), as opposed to only their descriptive knowledge of who usually performs certain roles (Liben and Bigler 2002).

Under this aim, one goal of the current study is to investigate when simplistic benevolent sexist attitudes toward who should engage in heroic and helpless gender roles first develop. Given the availability of information in young children’s environments relevant to heroic and helpless gender roles, it is possible that children learn about and endorse these roles at early ages. For example, Granger et al. (2017) observed that preschool teachers often facilitated masculine-typed play, such as playing pretend “firefighter” or “superhero,” more often with groups of boys than with girls. Additionally, superhero genres are popular among young children, even when the media content is directed toward teens and adults (Brown et al. 2009; Coyne et al. 2014; Do Rozarío 2004). Princess fairytales and superhero stories teach children that males are supposed to be strong, valiant heroes and leaders who protect others and do not rely on anyone else, whereas females are supposed to be loving, nice, and gentle, as well as be receptive to other people’s assistance because they are weak and fearful (Baker and Raney 2007; Brown et al. 2009; Dinella et al. 2017; England et al. 2011; Leaper et al. 2002; Padilla-Walker et al. 2013).

Although more recent Disney princesses show more agency than their predecessors, recent research suggests that many girls still expressed that boys must be princes and that it is a prince’s duty to protect princesses (Golden and Jacoby 2017). Further, young girls were highly aware of the classic, more dependent princesses such as Cinderella and Sleeping Beauty (Golden and Jacoby 2017). More general G-rated children’s films also depict male characters as having more strength than female characters and characterize female characters as “good” more than male characters (Smith et al. 2010). Considering the prevalence of these gender-trait stereotypes in the media, children might incorporate these messages into their attitudes toward who should be a hero and who must be helpless.

In contrast to early attitudes toward gender-typed traits, the extent to which children endorse attitudes about the roles of females and males has not been as extensively studied. Much of the work considering children’s attitudes toward appropriate gender roles are in the domain of play and activities (Hilliard and Liben 2010; Liben and Bigler 2002; Martin et al. 1990; Patterson 2012; Weisgram 2016) or occupations (Francis 1998; Liben and Bigler 2002; Meyer and Gelman 2016; Patterson 2012; Teig and Susskind 2008). For example, preschool-aged children have been found to think that boys should fix bikes or be police officers and girls should wash clothes or be baby-sitters (Hilliard and Liben 2010; Meyer and Gelman 2016; Wilbourn and Kee 2010). Further, the English language surrounding certain occupations, such as “policeman,” “fireman,” “businessman,” or “housewife,” are distinctly gendered. Children might internalize these social cues and endorse attitudes that powerful and agentic occupations, such as being on the police or fire force, should be filled by men. Further, among adults across several cultures, expectations that men should fulfill protective roles whereas women should fulfill nurturing, dependent roles are widely endorsed (Glick et al. 2000). These important previous studies suggest that attitudes toward roles regarding being heroes or needing special care might be salient among children; however, to our knowledge, these attitudes have not yet been directly examined.

Gender Stereotypes and Age

Past research on gender stereotypes have made a distinction between children’s prescriptive endorsement of gender stereotypes (e.g., Who should play with dolls?) and their descriptive knowledge of gender stereotypes (e.g., Who plays with dolls?) (Liben and Bigler 2002), which tend to have different developmental trajectories. Consistent with cognitive approaches to gender development, children’s endorsement of gender stereotypes is most rigid during early childhood (approximately ages 3 through 6 years-old), whereas older children in middle childhood (approximately ages 7 through 10 years-old) tend to be more flexible in their attitudes and more accepting of gender norm violations (Conry-Murray et al. 2015; Conry-Murray and Turiel 2012; Halim 2016; Martin and Ruble 2010; Ruble et al. 2007; Sinno and Killen 2009). Relevant to the current study, compared to preschoolers, during middle childhood children have reported stronger attitudes that boys and men should be tough, take care of themselves, and fix things more so than girls and women (Biernat 1991).

In contrast, knowledge of gender stereotypes tends to grow with age (Martin and Ruble 2010; Miller et al. 2006; Powlishta et al. 2001; Ruble et al. 2006; Signorella et al. 1993; Trautner et al. 2005). Children’s knowledge of stereotypes also become more complex as they get older, such as moving beyond simple, binary traits like “nice” versus “bad” to traits describing more abstract concepts like agency or communion (Biernat 1991; Cowan and Hoffman 1986; Martin and Little 1990; Powlishta 1995). In middle childhood, improvements in children’s social cognitive skills might also allow children to recognize social expectations for women and men (Carpendale and Lewis 2015; Halim et al. 2011; Leaper 2015) and, thus, view these roles situated in a larger context of paternalistic benevolent sexism (Conry-Murray et al. 2015). Then, as children approach adolescence, parents and peers may engage in forms of socialization related to appropriate dating roles that are more directly related to benevolent sexism (de Lemus et al. 2010; Glick and Hilt 2000; Madsen 2008; Raffaelli and Ontai 2001).

Study Overview

To summarize, the goal of the present study was to begin a new strand of inquiry by investigating the development of one facet of early benevolent sexist attitudes, particularly: (a) attitudes that girls require their physical needs to be met over others due to being the more delicate and dependent gender and (b) attitudes that boys are the stronger gender who should protect and rescue others. We aimed to test how these attitudes are associated with age and to ascertain whether these attitudes develop similarly among girls and boys. To address these aims, we interviewed 3- to 11-year-old children and showed them various scenarios involving situations where someone (a girl, boy, or either) would receive certain kinds of preferential treatment when in physical need (e.g., hungry, tired, cold, hurt) or where someone (a girl, boy, or either) was required to come to the rescue. Consistent with past work (Powlishta et al. 2001; Ruble et al. 2006; Signorella et al. 1993; Trautner et al. 2005), we expected that age would be positively associated with greater knowledge of these cultural stereotypes. Because the hero and helpless gender roles in which we were interested were relatively more complex than stereotypes about simple traits (e.g., nice/bad) and activities (e.g., dolls/trucks), we explored whether this knowledge would not be apparent until middle childhood. Similarly, because knowledge of these stereotypic gender roles may not solidify until middle childhood, we expected age to also be positively linked to stronger endorsement that boys should be the protectors and that girls need extra help. The usual greater gender flexibility evinced in middle versus early childhood may be pushed to later in development if knowledge of the hero/helpless roles are just beginning to be consolidated. We adopted an exploratory approach to understand whether these two gender role attitudes would be related to one another at different ages.

Method

Participants and Procedure

The sample consisted of 113 U.S. children ranging from ages 3 to 11 years-old (Mage = 7.13 years, SD = 2.54 years; 58 girls [51.3%]; 46 Latinx [40.7%]; 26 Black/African-American [23.0%]; 26 European-American [23.0%]; 11 Asian-American [9.7%]; 4 Other or unspecified [3.5%]) (see Table 1s in the online supplement for a breakdown of sample sizes by gender and each age). Gender was determined through parents’ report on the informed consent form. Age was determined by subtracting parents’ report of the child’s date of birth from the date the child participated in the study. Age was treated continuously in analyses below, with the exception of examining means from the mid-point of scales and correlations among certain measures by age group (Cohort 1: 44 3- to 6-year-old children [24, 54.5% girls; Mage = 4.34 years, SD = .94]; Cohort 2: 69 7- to 11-year-old children [34, 49.3% girls, Mage = 8.91 years, SD = 1.34]). The university institutional review board approved all procedures. We recruited children from eight daycare centers and after-school programs in Southern California with parent and child consent. Trained research assistants interviewed children one-on-one, showing scenarios to children and asking them questions for approximately 10 minutes. After the interview, children were thanked for their participation and later given a small gift.

Measures

Children were presented with images depicting ten scenarios with a gender-neutral make-believe character named “Fip,” a girl, and a boy. (See the online supplement for all scenarios.) We chose to use a gender-neutral character to decrease bias in children’s responses based on prevalent norms surrounding gender segregation and interaction (girls playing with or being nice to girls, boys playing with or being nice to boys; Martin and Ruble 2010). The gender-neutral character (see Fig. 1) was depicted in a green color, which young children consider to be gender-neutral (Yeung and Wong 2018). (The color version is available in the online supplement.)

Fig. 1
figure 1

Sample pedestal scale scenario of “Fip” having an extra blanket to share. (See the online supplement for a color depiction and all scenarios)

Full size image

Scenarios were inspired by the Protective Paternalism subscale from the well-established and validated Ambivalent Sexism Inventory for adults (e.g., “A good woman should be set on a pedestal by her man,” “In a disaster, women ought not necessarily to be rescued before men [reverse-coded]”; Glick and Fiske 1996). We created scenarios that were accessible to children and extensively pilot-tested the scenarios to ensure young children’s understanding and engagement. The scenarios were designed to be most appropriate for younger children, so older children were told prior to beginning that the questions were created for younger children, but that we were interested in what older children would think about the scenarios as well. Despite being designed for younger children, the older children were engaged with the measure in interview sessions.

In the scenarios, the gender-neutral character either required help (e.g., needing help opening a milk carton) or provided help (e.g., having an extra blanket to share). Consistent with prior work measuring attitudes (Liben and Bigler 2002), children were initially asked to decide which child should either help the gender-neutral character or be helped by the gender-neutral character. To avoid a forced choice situation, children were allowed to choose the girl, the boy, or “either one” (see Fig. 2). If children initially chose “either one,” we then probed to assess stereotype knowledge (Liben and Bigler 2002). (Information and analyses regarding the number of flexible responses chosen can be found in the Table 2s of the online supplement.) To examine separate facets of benevolent sexism, we divided the items into two scales of gender roles based on conceptual constructs: (a) attitudes that females require their physical needs to be met over others due to being the more delicate gender (the pedestal scale) and (b) attitudes that males are the stronger gender who should protect and rescue others (the hero scale).

Fig. 2
figure 2

Sample children used for scale responses of pedestal and rescue scenarios. (See the online supplement for a color depiction)

Full size image

Gender Role: The Pedestal Scale

We included five scenarios in the pedestal scale, which measured attitudes that girls are the more delicate gender and thus require special care for their physical needs and which were presented in the following order: (a) “It’s raining outside and Fip is the only one with an umbrella. Both Bobby and Lisa are getting wet. Who should Fip share the umbrella with?”; (b) “It’s nap time. Both Bobby and Lisa forgot their blanket at home and they are both cold. Fip has an extra blanket. Who should Fip lend the blanket to?”; (c) “Bobby and Lisa are hungry, but they forgot their snacks at home. But look! Fip has brought an extra cookie to school. Who should Fip give the extra cookie to?”; (d) “Both Bobby and Lisa fell down and hurt their knees and feel really bad. Fip has a band-aid, but only has one band-aid. Who should Fip give the band-aid to?”; and (e) “Bobby and Lisa are tired of walking around all day. They both want to sit down because they are so tired, but there is only one chair next to Fip. Who should Fip let sit in the chair?” Each scenario was followed by a probe; for example, for the final scenario, the interviewer added: “Bobby, Lisa, or either one? When I say either one, I mean Fip can let either Bobby or Lisa sit in the chair, it doesn’t matter to you which one.”

Because three responses were allowed (the boy, the girl, or either one), we created two subscales to be able to analyze the results. The first, which we considered to reflect children’s attitudes, was a sum score of all the stereotype-congruent responses (0 = stereotype-incongruent [picking the boy or picking “either one”]; 1 = stereotype-congruent [picking the girl]; possible range 0 to 5; M = 2.15, SD = 1.31). Three children did not provide responses to this measure, opting to skip the questions, and thus there were missing scores.

For children who answered “both” to the test questions, we probed once, asking that if they had to select Bobby or Lisa, which child would they select, similar to past measures used to assess gender-stereotype knowledge (Ruble et al. 2007; Signorella et al. 1993). We then created a second subscale, probed stereotype-congruent responses, which we considered to reflect children’s gender-stereotype knowledge, summed up all of the original stereotype-congruent responses plus all of the probed stereotype-congruent responses (0 = stereotype-incongruent [picking the boy]; 1 = stereotype-congruent [picking the girl]; possible range 0 to 5; M = 2.94, SD = 1.30). This version of the scale was used in subsequent analyses to be able to assess when means were significantly different from chance levels to answer the question of when benevolent sexist knowledge first develops.

Gender Role: The Hero Scale

We included five scenarios in the hero scale, which measured attitudes that males are the strong rescuers who provide help to others and which were presented in the following order: (a) “Fip needs help opening a milk carton. Who should come help Fip?”; (b) “Saz is being mean to Fip, who should help Fip?”; (c) “Fip climbed up a tree but now Fip is stuck. Who should help Fip down?”; (d) “Fip is being attacked by a dragon. Who should help Fip fight the dragon?”; and (e) “Fip needs help opening a door because it’s really heavy. Who should help Fip?” Again, each scenario was followed by a probe: “Bobby, Lisa, or either one? When I say either one, I mean either Bobby or Lisa can help Fip, it doesn’t matter to you which one.”

Similar to the pedestal scale, we created two subscales. One measured stereotype-congruent responses (0 = stereotype-incongruent [picking the girl or picking “either one”]; 1 = stereotype-congruent [picking the boy]; possible range 0 to 5; M = 2.44, SD = 1.30). Identical to the pedestal scale, “both” responses were probed once, and the second subscale measured probed stereotype-congruent responses (0 = stereotype-incongruent [picking the girl]; 1 = stereotype-congruent [picking the boy]; possible range 0 to 5; M = 2.98, SD = 1.28). Three children, separate from the three children missing responses in the pedestal scale, did not provide responses to this measure, opting to skip the questions, and thus were missing scores.

Results

To determine age and gender patterns in the endorsement of helpless and heroic gender role attitudes, we first conducted two multiple regressions. The outcomes for the separate regressions were the stereotype-congruent responses of the pedestal scale and the hero scale. For all regressions, predictors included participant gender (dummy coded with boys as the reference group), age (mean-centered), and the interaction between gender and age (see Table 1 for regression coefficients). (Distributions of all responses by each age and gender can be found in Table 3s of the online supplement.)

Table 1 Multiple regression coefficients for stereotype-congruent responses across both scales
Full size table

Next, to answer the question of when knowledge of these gender-typed roles first develop, we tested whether means on the probed stereotype-congruent scales by gender and cohort significantly differed from chance levels by conducting one-sample t-tests from the midpoint of the scale (2.50). Here, we used cohort groups instead of age continuously because the cells by each age were too small to powerfully test each age group on its own. Finally, to test whether female-helpless and male-heroic attitudes develop in concert or uniquely, we examined the correlation between the pedestal scale and the hero scale for each cohort.

Pedestal Scale: Age and Gender Patterns

Stereotype-Congruent Responses

The regression revealed significant main effects of participant gender and age for the endorsement of the pedestal scale such that girls showed stronger endorsement of the pedestal scale compared to boys, and age was positively associated with stronger endorsement. However, these main effects were qualified by a participant gender by age interaction. Simple effects indicated contrasting age trends for boys and girls. For boys, endorsement of the pedestal scale (i.e., agreement that girls should receive special assistance and benefits) was positively associated with age, β = .35, t(109) = 3.01, p = .003. For girls, endorsement of the pedestal scale was negatively associated with age, β = −.47, t(109) = 3.85, p < .001 (see Table 1a and Fig. 3).

Fig. 3
figure 3

Predicted values of the pedestal scale by age and gender. Higher numbers indicate more stereotype-congruent responses (selecting girls). ***p < .001

Full size image

Probed Stereotype-Congruent Responses

Boys in Cohort 1 (ages 3–6) on average selected boys to be placed on a pedestal (M = 1.85, SD = 1.27), t(19) = 2.29, p = .033, d = .51, but boys in Cohort 2 (ages 7–11) responded at chance levels (M = 2.74, SD = 1.24), t(34) = 1.15, p = .256. Girls from both Cohorts 1 and 2 believed that girls should be placed on a pedestal at greater than chance levels, although the effect size was larger among the younger girls (M = 3.79, SD = 1.06), t(23) = 5.96, p < .001, d = 1.22, than the older girls (M = 3.18, SD = 1.03), t(33) = 3.83, p = .001, d = .66. These results suggest that the belief that girls should be placed on a pedestal is present as young as in early childhood among girls, but is not yet present among boys across early to middle childhood.

Hero Scale: Age and Gender Patterns

Stereotype-Congruent Responses

The regression revealed a significant gender main effect in the endorsement of the hero scale. Specifically, boys were more likely to choose the boy to come to the rescue than were girls. No significant effects were found for age or for the interaction between participant gender and age (see Table 1b).

Probed Stereotype-Congruent Responses

Boys from both Cohorts 1 (ages 3–6) and 2 (ages 7–11) believed that boys should be the heroes at greater than chance levels, although effect sizes were larger among the older boys (M = 3.51, SD = .98), t(34) = 6.12, p < .001, d = 1.03, than among the younger boys (M = 3.20, SD = 1.28), t(19) = 2.44, p = .025, d = .55. Girls in Cohort 1 did not differ from chance levels in beliefs about who should be the heroes (M = 2.04, SD = 1.43), t(23) = 1.57, p = .130, but girls in Cohort 2 on average believed that boys should be the heroes at greater than chance levels (M = 2.97, SD = 1.11), t(33) = 2.46, p = .019, d = .42. These results suggest that the belief that boys should be heroes is present as young as early childhood for boys, but not until middle childhood for girls.

Association Between Pedestal and Hero Scales by Cohort

Among Cohort 1 (ages 3–6 years-old), the pedestal scale and the hero scale (stereotype-congruent responses) were negatively correlated, r(44) = −.46, p = .003. This means that young children who selected girls to receive special care for their physical needs tended to also select girls to be the heroes (and vice versa: young children who selected boys to receive special care for their physical needs tended to also select boys to be the heroes). In contrast, among Cohort 2 (ages 7–11 years-old), the pedestal scale and the hero scales were positively correlated, r(66) = .34, p = .005. These correlations did not significantly vary by participant gender. This means that older children who selected girls to receive special care for their physical needs tended to also select boys to be the heroes (which was the case for 42 [60.9%] of the older children who responded above chance levels with the two scales). Only a small minority of older children who responded above chance levels (8 [11.6%]) selected boys to receive special care and also selected girls to be the heroes.

Discussion

The aims of the present study were to understand children’s attitudes toward stereotypical gender roles that reflect basic benevolent sexist attitudes, test the association of these attitudes with age, and investigate whether girls and boys showed differential patterns. To our knowledge, our study was the first to test foundational benevolent sexist attitudes in children younger than adolescents. We provided evidence that children endorse and show knowledge of heroic and helpless gender roles that reflect dimensions of early benevolent sexist ideas.

Interestingly, two facets of the data suggest that female-helpless and male-heroic gender role attitudes have distinct developmental patterns and do not necessarily develop in concert with each other. First, we found differential associations of helpless and hero gender roles depending on age. Second, we observed differential age patterns in the endorsement of the individual constructs. Overall, these findings suggest that learning the gender roles that encompass one component of benevolent sexism does not necessarily go hand-in-hand with learning another. Children may learn these gender-typed roles at different times.

Regarding the associations of these gender roles, attitudes that females should be placed on a pedestal and males should be the heroes were negatively related among the younger cohort (ages 3–6 years-old). This negative correlation may reflect the strong ingroup bias typically seen during this developmental period (Halim et al. 2017; Martin and Ruble 2010), with girls favoring and selecting girls for most items across the two components of benevolent sexism and boys favoring and selecting boys. By middle childhood, however, the data suggest that these gender roles reflecting benevolent sexism become more related, but are still distinct. Specifically, we found a positive but modest correlation between attitudes that females should be placed on a pedestal and attitudes that males should be the heroes.

When does knowledge of heroic and helpless gender role stereotypes emerge? Examinations of probed responses suggest that children might have some knowledge about the roles stereotypically associated with their own gender from as young as age 3. Boys in both cohorts displayed knowledge that boys are the heroes at greater than chance levels. Similarly, girls in both cohorts displayed knowledge that girls are to be placed on a pedestal at greater than chance levels. These data are consistent with some previous work that suggests children learn what is appropriate for their own gender first (Albert and Porter 1988; Biernat 1991). Also, these findings may not be surprising given that media describing which gender should be placed on a pedestal and which gender should be the agentic hero is pervasive among preschool-aged children (Coyne et al. 2016; Orenstein 2006; Wohlwend 2009). Children’s early knowledge about the roles stereotypically associated with their own gender might further reflect ingroup bias, which is common during this developmental period (Martin and Ruble 2010). Future work is needed to ascertain how much children’s responses reflect knowledge and/or ingroup favoritism.

Knowledge about stereotypical roles for the other gender occurred later. Boys in the younger cohort thought that boys were to be placed on a pedestal, and boys in the older cohort responded at chance levels. Older (ages 7–11), but not younger (ages 3–6), girls showed knowledge that boys are the heroes. These developmental patterns might reflect the advancement of children’s social cognitive skills, which help them tune into societal expectations for women and men (Carpendale and Lewis 2015; Halim et al. 2011; Leaper 2015).

Our data also suggest that girls’ and boys’ development of attitudes toward different roles associated with benevolent sexism varied. Partially supporting our hypotheses, for boys, age was associated with being less likely to say that boys (and more likely to say that girls) should receive certain kinds of preferential treatment (e.g., when hungry, wet, cold, or tired). This finding is even more remarkable when recognizing that boys’ selection of the other gender was pitted against tendencies to favor boys, which typically remains strong even in middle childhood (Martin and Ruble 2010). Further, this age pattern is consistent with the idea that with age, children may be more directly taught gendered scripts of benevolent sexism, as they prepare to enter adolescence and entertain the idea of romantic relationships (de Lemus et al. 2010; Glick and Hilt 2000; Raffaelli and Ontai 2004).

The pattern we found for girls’ selection of girls being placed on a pedestal was particularly interesting. Unexpectedly, age was negatively associated with girls selecting females to be placed on a pedestal. Given that adolescent girls and many women have been found to endorse benevolent sexism (Glick et al. 2000; Moya et al. 2007; Silván-Ferrero and López 2007), we speculate that the developmental trajectory of these beliefs could be U-shaped. A U-shaped curve would be consistent with other studies that have found that middle childhood is sometimes a unique period when children, and especially girls, show a decreased adherence to gender-typed expectations and increased ability to recognize, and even reject, gender inequality (Bigler et al. 2008; Brown et al. 2011; Conry-Murray et al. 2015; Conry-Murray and Turiel 2012; Katz and Ksansnak 1994; Liben et al. 2001; Signorella et al. 1993; Sinno and Killen 2009), as they gain more sophisticated social cognitive skills (Halim et al. 2011). Indeed, older girls’ (aged 7–11 years-old) probed responses indicated knowledge of the cultural stereotype that girls should receive special help, yet they did not endorse this stereotype on average. Perhaps girls, once strongly in favor of receiving preferential treatment, later recognize in middle childhood that both girls and boys should be treated the same.

Regarding children’s attitudes about who should be the hero, young boys, as young as 3 years of age, thought that boys should be the heroes, and this remained steady without significant change across age. Girls were less likely than were boys to say that boys should be the heroes across age. As previously reviewed, girls may recognize an inequality in expectations that only boys should be heroes and reject these ideals (Bigler et al. 2008; Conry-Murray et al. 2015; Conry-Murray and Turiel 2012; Katz and Ksansnak 1994; Liben et al. 2001; Sinno and Killen 2009). Girls might also be sensitive to the popularity of recently released media depicting female superheroes, like Wonder Woman (Women’s Media Center and BBC America 2018).

Limitations and Future Research Directions

Our study provides a valuable start in examining the development of benevolent sexism. However, there are limitations of our study, which future research should address. First, as we noted earlier, young children often display in-group favoritism (Cameron et al. 2001; Halim et al. 2017; Martin and Ruble 2010), and because of our limited data, we were not able to directly tease apart whether young children’s endorsement of certain roles were a reflection of broader cultural stereotypes or in-group favoritism, although we speculate a combination of the two is most likely. Future work can include a measure of in-group bias to more clearly make this distinction.

We additionally provided evidence to answer questions about when helpless and hero gender role attitudes develop; however, we do not know yet what contributes to this development. Future studies could test whether exposure to certain kinds of media—like princess fairytales and superhero adventures, conversations with adults (particularly with parents) about gender roles, or observations of certain dynamics between heterosexual parents or other romantic couples—might lead children to learn stereotypes that may plant the seeds of benevolent sexism. Additionally, future work could test whether increased interaction with other-gender peers predicts greater endorsement of benevolent sexist gender roles (Glick and Hilt 2000).

Future work could also test similar questions among adolescents for several reasons. First, we observed that the two constructs of helpless and hero gender role attitudes were negatively associated in early childhood, but positively associated in middle childhood. It would be interesting to test whether these associations become even more strongly linked during adolescence. Second, boys in elementary school were still at chance levels in showing cultural knowledge that girls are to be placed on a pedestal. We speculate that it might not be until adolescence that boys show this knowledge. Third, girls’ attitudes toward being placed on a pedestal were negatively associated with age. We speculate that as girls enter adolescence, endorsement might increase because romantic relationships become more salient.

An additional limitation of our study considers children’s perceptions of the target used in the illustrations. Despite pilot testing that suggested children viewed the target used as gender-neutral, most children in our study reported viewing the target as a boy (n = 69, 61.1%), a quarter reported not knowing the gender (n = 28, 24.8%), and a few thought the character was a girl (n = 10 [8.8%]) or neither a girl nor boy (n = 5, 4.4%). Although examination of these perceptions did not reveal differences in response patterns, more agreement across the sample about the target’s gender would further strengthen conclusions of our findings.

Relatedly, another valuable future direction would be to test children’s beliefs about who should assist whom or who should receive special treatment from whom when the target is female or male. We consider it to be a strength of our measure that we found our effects using a target with no specific gender because it dampens potential noise caused by ingroup favoritism (Rhodes 2012), especially when testing very young children. However, future studies could additionally test benevolent sexism by varying the gender of the target to understand the specificity of children’s beliefs (e.g., whether males should be heroes to everyone or especially when damsels are in distress). This could be particularly beneficial when examining pre-adolescent children’s (ages 11–13) responses.

Further, it is possible that certain scenarios of our measure for rescuing roles (e.g., holding a door open, opening a milk carton) might have been interpreted by some children as representing more feminine communal helping behaviors rather than as masculine chivalry (Eagly 2009). Exploratory analyses showed that these individual items showed no significant differences in age and gender patterns compared to the other items. However, future research could explore a broader array of heroic scenarios to determine which scenarios children view as paternalistic versus communal and whether these views remain consistent with age.

Practice Implications

Attitudes about the appropriate roles of women and men, and of girls and boys, might provide a foundation upon which benevolent sexist beliefs are built. Social contexts that children have to navigate provide messages about who engages in paternalistic roles and who is patronized (Brown et al. 2009; Coyne et al. 2014; Do Rozarío 2004). This might be concerning because young children are actively constructing their gender identities while encountering these messages. In line with cognitive theories of gender development, children are highly aware of these cues and might internalize these gender roles as a part of their gender identity (Martin et al. 2002). As children get older and romantic attractions facilitate more conversations about the interdependence of women and men, gender role expectations of who are the helpless or the heroes might already be set in place.

Endorsement of these restrictive gender roles might also be detrimental to children. Boys' belief that they should come to the rescue might put them in difficult situations and cause them to refuse to seek help for themselves (e.g., mental health, instrumental or social support) because of expectations that the rescuers should not need rescuing from anyone else (Burns and Rapee 2006; Mahalik et al. 2003; Way et al. 2014). The over-idealization of being a hero could also translate to an unbalanced pressure for boys to join the military or other physically dangerous professions when it might not specifically fit their personal goals or talents. For women, benevolent sexism has been associated with decreased cognitive performance and self-esteem (Dardenne et al. 2007; Dumont et al. 2010) and greater acceptance of gender inequality (Barreto and Ellemers 2005; Becker and Wright 2011). If girls begin adopting benevolent sexist attitudes at young ages, these consequences seen in adulthood might emerge even earlier to hinder girls’ experiences.

A way to combat these attitudes in children might be to provide alternative messages about the appropriate roles of girls and boys. Parents and teachers of young children can have girls and boys rotate “rescuing” roles between one another in play to emphasize that no matter what their gender is, anyone can be a hero. They can also consider actively mediating or discussing the media children are watching to help children understand that what they watch is only one set of roles for girls and boys (Nathanson et al. 2002).

Conclusion

Our findings on children’s early gender role attitudes concerning who is helpless and who is heroic shed light on the development of benevolent sexist ideology. Our study is the first known to empirically test foundational benevolent sexist gender role attitudes and challenge its assumptions about childhood. More specifically, our findings suggest that there may be more nuance involved, especially among older children (ages 7–11). Components of benevolent sexism emphasizing paternalism might be present earlier than expected. Our findings indeed suggest that even before adolescence, the seeds of paternalistic attitudes toward who should be a hero and who needs special care may begin to germinate. If children learn that both girls and boys can be bold and that both boys and girls can be wonderful, steps toward greater gender equality can be taken.