Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Sep 1.
Published in final edited form as: Dev Sci. 2024 Jun 5;27(5):e13532. doi: 10.1111/desc.13532

Children’s gender essentialism and prejudice: Testing causal links via an experimental manipulation

E B Gross 1, Rachel D Fine 1, Selin Gülgöz 2, Kristina R Olson 3, Susan A Gelman 1
PMCID: PMC12100744  NIHMSID: NIHMS2078730  PMID: 38837632

Abstract

Despite increases in visibility, gender-nonconforming young people continue to be at risk for bullying and discrimination. Prior work has established that gender essentialism in children correlates with prejudice against people who do not conform to gender norms, but to date no causal link has been established. The present study investigated this link more directly by testing whether children’s gender essentialism and prejudice against gender nonconformity can be reduced by exposure to anti-essentialist messaging. Children ages 6-10 years of age (N=102) in the experimental condition viewed a short video describing similarities between boys and girls and variation within each gender; children in the control condition (N=102) viewed a corresponding video describing similarities between two types of climate and variation within each. Children then received measures of gender essentialism and prejudice against gender nonconformity. Finally, to ask whether manipulating children’s gender essentialism extends to another domain, we included assessments of racial essentialism and prejudice. We found positive correlations between gender essentialism and prejudice against gender nonconformity; both also correlated negatively with participant age. However, we observed no differences between children in the experimental versus control conditions in overall essentialism or prejudice, indicating that our video was largely ineffective in manipulating essentialism. Accordingly, we were unable to provide evidence of a causal relationship between essentialism and prejudice. We did, however, see a difference between conditions on the discreteness measure, which most closely linked to the wording in the video. This finding suggests that specific aspects of essentialism in young children may be modifiable.

Keywords: Gender essentialism, prejudice, gender diversity, children

1. Introduction

In recent years, awareness of gender-nonconforming people has increased dramatically in the United States. However, data from GLSEN’s National School Climate Survey have shown that year after year, gender-nonconforming young people continue to face discrimination and bullying (Kosciw et al., 2020).

One potential contributor to these negative responses is gender essentialism, the belief that gender categories (e.g., boys, girls) have an underlying, non-obvious reality (Gelman, 2003, 2004). Although gender essentialism reflects important insights (that appearances can be deceiving, that nonobvious causal features can give rise to observable features), it is also a cognitive bias that oversimplifies the varied and complex nature of gender categories (see Gelman, 2003; Leslie, 2013 for discussion). This core belief is manifest in several respects. Gender categories and stereotypical features are believed to be: naturally occurring (innate), providing rich information about individuals (informative), distinct and non-overlapping (discrete), and stable over time (immutable) (Gelman, 2003; Rhodes & Mandalaywala, 2017). Essentialist thinking has been documented in a wide variety of categories beyond gender, including not only natural kinds (e.g., tigers, trees, gold; Gelman, 2003) but also other social categories (e.g., race/ethnicity, traits; Haslam et al., 2000, 2004; Hirschfeld, 1995). Among social categories, essentialism is particularly evident in reasoning about gender, and especially so in early childhood (Gelman, 2003, 2004; Rhodes & Mandalaywala, 2017; Taylor et al., 2009). For example, by 4 years of age, children treat gender category membership as more important than appearances when making inferences about a person’s behavior and biological features (Gelman et al., 1986), they treat sex at birth as more important than environment or upbringing in determining a person’s gender-stereotypical properties later in life (Taylor, 1996; Taylor et al., 2009), and they make predictions about a person’s future actions based on gender stereotypes more than past behavior (Berndt & Heller, 1986). Gender essentialism can decrease with age, but does so only in certain social contexts that promote non-essentialist reasoning (Davoodi et al., 2020; Meyer & Gelman, 2016; Taylor et al., 2009).

There are reasons to hypothesize that essentialist beliefs about gender might contribute to negative attitudes toward gender nonconformity. Prior research shows that children tend to treat behavioral regularities (what is) as normatively correct (what should be) (Roberts et al., 2017; Tworek & Cimpian, 2016). Thus, for example, when considering novel social groups, children typically report that it is “not OK” for someone to fail to conform to what is typical for their group (e.g., if Hibbles and Glerks play different games, it is wrong for a Hibble to play a Glerk game; Roberts et al., 2017). Furthermore, the more that a categorical feature is believed to be necessary and inalterable, the stronger the judgment that these descriptive features are normatively correct (Tworek & Cimpian, 2016). Given that gender essentialism entails construing gender-linked properties as innate and immutable, it would then be predicted to heighten the belief that people should conform to gender stereotypes, thus leading to negative evaluations of those who do not conform. Relatedly, because gender nonconformity challenges certain essentialist beliefs (e.g., category boundary discreteness, biological basis, and inductive potential), this may further contribute to negative evaluations among those who hold such beliefs.

Prior work has identified correlational links between social essentialism (on the one hand) and negative evaluation, stereotyping, and bias (on the other hand), especially regarding those perceived as behaving in ways that do not match their social role (Pauker et al., 2016; Rhodes & Mandalaywala, 2017). Most relevant to the present study, Fine et al. (2024) studied gender essentialism and gender-related attitudes in children 6-10 years of age, and found that higher levels of gender essentialism correlated with higher rates of bias against gender-nonconforming children, across the age range. Similarly, Gülgöz et al. (2018) found that children ages 5-10 years preferred cisgender over transgender peers, and that dislike of transgender peers was stronger in children who categorized these peers by natal sex than by gender identity.

An important further consideration is that gender essentialism and negative evaluations of gender nonconformity vary not only across individuals within a given community, but also systematically as a function of the environment in which a child lives (Conry-Murray et al., 2015; Diesendruck et al., 2013; Fine et al., 2024; Shutts et al., 2017; Taylor et al., 2009). For example, a study comparing children in two different communities in the midwestern U.S., one a higher-SES, liberal, and relatively diverse city, the other a lower-SES, conservative, predominantly white, rural town, found that they showed markedly different patterns of gender and race essentialism, with lower levels of essentialism in the liberal city (Rhodes & Gelman, 2009). Such variation suggests that children in different environments may be receiving, and internalizing, different messaging about social categories, including gender, based on the demographic features of their community.

To this point, all the evidence linking gender essentialism and negative evaluation has been correlational, not allowing for causal inferences. We thus do not yet know if essentialist beliefs contribute to negative attitudes toward gender-nonconforming children, or if instead the correlation may be explained in some other way. For example, the reverse causal pattern could be at play (negative evaluations may lead to higher levels of essentialism), or some other variable(s) (e.g., conservative belief systems) may be responsible for both higher levels of essentialism and negative attitudes.

The current study experimentally tested whether exposure to anti-gender-essentialist messaging leads to reduced gender essentialism in children, and if so, whether such a reduction in essentialism produces a corresponding reduction in bias against gender-nonconforming children. We focused on children 6-10 years of age, as this is when children show high levels of gender essentialism, and this age group has been the focus of recent work finding a correlation between gender essentialism and prejudice against gender-nonconforming children (Fine et al., 2024). Study participants were shown one of two videos: (a) a focal gender video that communicated both between-category similarity (boys and girls are a lot alike) and within-category variability (there is much variation within each of these two genders), or (b) a matched control video that focused on two different climates (deserts and frozen lands) but otherwise was structurally identical to the gender video, communicating that deserts and frozen lands share common characteristics and that there is much variation within each of the two climates.

The anti-essentialist manipulation sought to modify the core gender essentialist belief that boys and girls are distinct kinds, which entails a heightened and overly extreme belief that members of one gender group are uniformly very similar to each other, and highly different from (or “opposite to”) members of another gender group. The current manipulation was designed to reduce this belief by highlighting similarities between boys and girls, and highlighting variation within each category of boys and girls. More specifically, this child-friendly manipulation (described in more detail in the Method section, below) reminded children that boys and girls are alike in many ways, including in their “deep,” internal structure (e.g., blood, heart, brains). The manipulation also reminded children that boys can be highly varied from one another, and that girls can be highly varied from one another. We included measures examining both gender and racial categorization and preferences, in order to evaluate the possibility that our manipulation may affect not only gender essentialism but social essentialism more broadly.

2. Method

2.1. Participants.

The final sample size was 204 participants across the age range of 6-10 years, 102 per condition. We registered that we would include 200 participants, 100 per condition, roughly 20 per year of age across the age range of 6-10 years, but that the number of participants could slightly exceed this number, if more families signed up for the study prior to testing the 200th participant. This sample size was selected because it provides us with an estimated 80% power to detect an effect size of at least Cohen’s d = .40 (alpha = .05) in the main analysis — the t-test comparing our two conditions (Faul et al., 2007). Participants were recruited from rural areas in the midwestern United States, because this was the group in previous research that showed prejudice against gender-nonconforming youth (Fine et al., 2024); children in a more urban, progressive community showed no prejudice and therefore, attempting to change their bias could produce prejudice.

Participants were tested between February 2022 and February 2024. Children who were fluent speakers of English within the predetermined age ranges (see above) with appropriate consent and assent were included (N=204). An additional nine participants were recruited but excluded for the following registered reasons: participation in the prior correlational study (Fine et al., 2024) (n=5); interference by a family member (n=1), cognitive impairment (n=1), and being outside of the registered age range (n=2). No children were excluded for the other registered reasons (technical difficulties, experimenter error, English limitations, special education services, or failing to answer at least half the Prejudice against Gender Nonconformity attention check questions correctly). Additionally, for all measures, we had registered that participants needed to answer at least half of the questions to be included in the analyses relevant to that measure; however, no participants were excluded from analyses for this reason. See Table 1 for full demographic information.

Table 1.

Participant Demographics1

Experimental Control Total
Participants (N) 102 102 204
Age in Years2, M(SD) 8.37 (1.40) 8.34 (1.39) 8.35 (1.39)
Gender N (%)3
 Girl 46 (45.1) 46 (45.1) 92 (45.1)
 Boy 56 (54.9) 55 (53.9) 111 (54.4)
 Nonbinary 0 (0) 1 (1.0) 1 (0.5)
Race/ethnicity N (%)
 White/European American 83 (81.4) 79 (77.5) 162 (79.4)
 Black/African American 2 (2.0) 2 (2.0) 4 (2.0)
 Asian/Asian American 1 (1.0) 1 (1.0) 2 (1.0)
 Hispanic/Latiné 1 (1.0) 3 (2.9) 4 (2.0)
 Native American 0 (0) 2 (2.0) 2 (1.0)
 Bi- or Multiracial/Ethnic 11 (10.9) 10 (9.8) 21 (10.3)
 Unknown 4 (3.9) 5 (4.9) 9 (4.4)
Household Income, N (%)
 Less than $25,000 8 (7.8) 7 (6.9) 15 (7.4)
 $25,001-$50,000 14 (13.7) 17 (16.7) 31 (15.2)
 $50,001-$75,000 18 (17.6) 16 (15.7) 34 (16.7)
 $75,001-$125,000 31 (30.4) 28 (27.5) 59 (28.9)
 Greater than $125,000 25 (24.5) 28 (27.5) 53 (26.0)
 Unknown 6 (5.9) 6 (5.9) 12 (5.9)
Parent political orientation4 M(SD) 3.05 (1.34) 3.09 (1.58) 3.07 (1.46)
Primary Caretaker Education, N (%)
 Some schooling 0 1 (1.0) 1 (0.5)
 High school diploma 5 (4.9) 2 (2.0) 7 (3.4)
 Some college / Associate degree 23 (22.5) 28 (27.5) 51 (25)
 Bachelor’s degree 37 (36.3) 28 (27.5) 65 (30.9)
 Advanced degree 35 (34.3) 38 (37.3) 73 (35.8)
 Unknown 2 (2.0) 5 (4.9) 7 (3.4)
Secondary Caretaker Education, N (%)
 Some schooling 6 (5.9) 1 (1.0) 7 (3.4)
 High school diploma 7 (6.9) 15 (14.7) 22 (10.8)
 Some college / Associate degree 26 (25.5) 22 (21.6) 48 (23.5)
 Bachelor’s degree 28 (27.5) 28 (27.5) 56 (27.5)
 Advanced degree 25 (24.5) 22 (21.6) 47 (23.0)
 Unknown/No 2nd Caregiver 10 (9.8) 14 (13.7) 24 (11.8)
Open in a new tab

Notes:

1

We determined community-level eligibility based on an examination of county-level voting record, population, median income, and education level. All communities were in counties in which the majority voted for the Republican presidential candidate in the 2020 election.

2

26 participants provided year-of-age at time of registration only; for these, age was set in the middle of the year (e.g., a 6-year-old was set at 6.5, a 7-year-old-old at 7.5, etc.). All but 4 were tested within 2 weeks of registration. One participant provided birth month and year; age was set at the middle of the birth month.

3

Due to rounding, percentages may not sum to 100.

4

Political orientation was measured on a scale from 1 = very liberal to 7 = very conservative; 17 participants did not provide their political orientation.

2.2. Procedure.

Participants were tested individually by a trained researcher via remote video-conferencing. Following introductions and warm-ups to ensure that the video interface was correctly set up, each participant viewed the manipulation (either gender or climate) and then completed four tasks, in this order: the Gender Essentialism Scale for Children (GES-C), Prejudice against Gender Nonconformity measure, Category Boundary task, and Friendship Preference task (each described below). At the completion of the study, each participant was thanked, and received a $25 gift card as compensation.

2.2.1. Manipulation.

Within gender and year of age (e.g., 6-year-old boys, 6-year-old girls, 7-year-old boys, 7-year-old girls), children were randomly assigned to either the Gender (experimental) video or the Climate (control) video. The scripts for each video appear below; each video was illustrated with animated images corresponding to the information provided, concluding with a set of photographs of either children or climates (depending on condition) to demonstrate within-category variability.

[Gender] “Girls and boys are a lot alike. Sometimes they may look different but really and truly they are the same in a lot of ways. Take bodies, for example. All kids have hearts that pump blood around the body, and brains that let them think. And eyelashes, and elbows, and knees, and lungs. And, girls and boys are the same in other ways. They like to hang out with their friends, ride bikes, do puzzles, read books, go swimming, and help bake (and eat!) cookies. Boys and girls are a lot more alike than they are different. There’s no one way to be a boy or a girl. Take a look at all these different kids! Boys and girls are really the same in lots of ways. What’s really important is that we are all people.”

[Climate] “Deserts and icy, frozen lands are a lot alike. Sometimes they may look different but really and truly they are the same in a lot of ways. Take the outdoors for example. Both deserts and frozen lands have animals and birds. And insects, and plants, and fresh air, and dirt. Deserts and frozen lands are the same in other ways, too. They can get very cold, and windy, and dry, and sunny during the day, and starry at night, and people live in both places. Deserts and frozen lands are a lot more alike than you might think. There’s no one type of desert or frozen land. Take a look at all these different places! Deserts and frozen lands are really the same in lots of ways. What’s really important is that we take care of all the Earth’s lands.”

2.2.2. Gender Essentialism (GES-C).

We employed a 16-item measure, the Gender Essentialism Scale for Children (GES-C) (Fine et al., 2024) (see Appendix A for full scale). This scale assesses four components of gender essentialism: biological basis, discreteness, informativeness, and immutability (4 items per subscale). For each statement, children indicate whether they agree or disagree, and then respond to a follow-up question, “a little or a lot?” These responses are converted to a 4-point scale from 1 = ‘disagree a lot’ to 4 = ‘agree a lot’, with higher numbers indicating more essentialist beliefs after reverse-coded items are recoded. Prior research indicated that the alpha for the full scale was .87 (Fine et al., 2024). Items were presented in randomized order. This task functions as a manipulation check.

2.2.3. Prejudice against Gender Nonconformity.

Children were shown four short cartoon clips, presented in a random order, depicting gender-conforming or gender-nonconforming children, and were asked how much they liked or disliked each child. More specifically, the clips presented a gender-conforming girl, a gender-nonconforming girl, a gender-conforming boy, and a gender-nonconforming boy, along with a voice-over providing the character’s name, gender/sex label, and toy preference. For example, this is the wording for the gender-nonconforming boy item (see Appendix B for full measure):

“One day, a little baby was born. This baby’s name was Jack, because he had a boy body. Jack’s parents bought him toy cars and trucks to play with. Every day, Jack would play outside. But Jack didn’t like to play with his toy cars and trucks. He only wanted to play with his sister’s ballerina costumes and toy unicorns. Jack’s mom was very confused. She asked him, ‘Don’t you want to play with your trucks, like other boys?’ But Jack said, ‘No, mom, I don’t want to play with trucks. I want to play with ballerina costumes and toy unicorns, just like my sister.’ So Jack’s parents decided to buy some ballerina costumes and unicorns for him. This made Jack very happy, and he played with them all the time.”

After each clip, the participant answered two attention check questions, one about the target child’s gender (e.g., “Did this kid have a boy name or a girl name?”), and the other regarding the toy(s) they preferred to play with (e.g., “Did this kid like cars and trucks or ballerina costumes and unicorns?”). If a participant answered either attention check question incorrectly, the experimenter corrected them and made note of the error. If a participant had answered more than half of the attention check questions incorrectly (i.e., 5 or more out of 8), they would have been excluded from the study (though no child was). For each item, following the attention checks, participants were asked, “Do you like this kid, or not like this kid?” which was then followed with three options: for “like”, “Do you kinda like them, like them, or really like them?” and for “not like”, “Do you kinda not like them, not like them, or really not like them?” (with accompanying frowning and smiling faces on the screen)1. Each item received a score from 1 = “really don’t like” to 6 = “really like”. Gender-nonconforming item scores were subtracted from gender-conforming item scores, for a scale from −5 to +5 where higher scores indicated a greater degree of prejudice against gender-nonconforming children.

2.2.4. Category Boundary Task.

This task assessed children’s belief that gender and racial category boundaries are objective (Rhodes & Gelman, 2009). Children were asked to judge a series of statements made by the character Feppy, “a visitor from a place far, far away, where they do lots of things differently than we do” (Kalish, 1998; Rhodes & Gelman, 2009). After a training section in which children assessed whether Feppy is “maybe right” or “wrong” in (a) misstating their name and (b) stating a food preference, the measure included two warm-up trials, one with a pair of cats and one with a pig and a goat; for each, children were asked whether Feppy is “maybe right” in stating that items in each pair were the same kind of animal; they could respond either “yes” or “no.” Children were provided with corrective feedback on the warm-up trials, if needed. Then for the experimental items, children were shown pairs of images of children and told that Feppy thinks they are the same kind of person. For each pair, participants were asked whether Feppy is “maybe right,” again providing “yes” or “no” responses.2 Items included 2 different-gender pairs (Black boy and Black girl; White boy and White girl), 2 different-race pairs (Black boy and White boy; Black girl and White girl), and 2 filler pairs of the same gender and race (Black girl and Black girl; White girl and White girl). Items were presented in a random order, and left-right presentation of images within each pair was counterbalanced. Responses were coded as 0 when a child said that yes, Feppy was “maybe right” that the two images are the same kind of person, and 1 when they said that no, Feppy was not “maybe right.” This yielded two composite category boundary scores, one for gender and one for race, each ranging between 0 (child responded ‘yes’ for both trials) and 2 (child responded ‘no’ for both trials). Higher scores on these measures indicated that children were emphasizing boundaries — a more essentialist tendency. Filler pairs were included for descriptive purposes only, and warm-ups were not coded.

2.2.5. Friendship Preference Task.

As a secondary measure of social preference and prejudice, children were shown six pairs of images of children and were asked, “Who would you like to have as a friend?” Items included 2 pairs of gender-conforming children differing in gender but not race (White girl and boy; Black girl and boy); 2 pairs of children differing in gender expression but not gender or race (White gender-conforming boy and White gender-nonconforming boy; White gender-conforming girl and White gender-nonconforming girl), and 2 gender-conforming children differing in race but not gender (Black boy and White boy; Black girl and White girl). Items were presented in random order, and the left-right presentation of images was counterbalanced. Children were asked to indicate which child in each pair they would like to have as their friend. A friendship preference score was generated as follows: (a) For gender trials, choosing the same-gender picture as the child was scored as 1, choosing the other-gender picture as the child was scored as 0; scores were summed, yielding a composite score ranging from 0-2. (b) For gender-expression trials, choosing the gender-conforming child was scored as 1, choosing the gender-nonconforming child was scored as 0; scores were summed, yielding a composite score ranging from 0-2. The nonbinary participant (n=1) was excluded from this analysis. (c) For race trials, only children whose parents reported their race as either White or Black/African American were scored, and the scoring was based on selecting choices that matched participant race. Therefore, for these children, choosing the same-race as the child was scored as 1, choosing the other-race was scored as 0; scores were summed, yielding a composite score ranging from 0-2.3

3. Analysis Plan

3.1. Scoring.

Each participant received 11 scores: (a) GES-C full-scale score, which was the average of the 16 items, ranging from 1 to 4, with 4 representing the most endorsement of essentialism; (b) Prejudice against Gender Nonconformity (GNC) score, a difference score of the average liking of the gender-conforming targets minus the average liking of the gender-nonconforming targets, ranging from −5 to +5 such that higher scores indicated greater preference for gender-conforming targets; (c)-(f) GES-C subscale scores, one for each of the four components: biological basis, discreteness, informativeness, and immutability; subscale scores each ranged from 1 to 4; in each case higher scores indicated more gender essentialism; (g) and (h) Category Boundary scores, one for gender and one for race, each being a cumulative score from 0-2 such that higher scores indicated stricter category boundaries/more essentialist reasoning; (i)-(k) Friendship Preference scores, one for gender, one for gender-expression, and one for race, each being a cumulative score from 0-2, reflecting children’s ingroup and pro-gender-conformity preferences. Two participants had partial missing data, each on one measure: GES-C and Friendship Preference. For each, scores were adjusted for the number of items completed.

3.2. Descriptive and preliminary data.

We report Ms and SDs for all variables by condition, as well as Cronbach’s alphas for the GES-C full-scale and each component. We also report the Pearson’s r correlations among the following: GES-C full-scale score, Prejudice against Gender Nonconformity score, Category Boundary gender items, Friendship Preference gender items, Friendship Preference gender expression items, and child age (as a continuous variable), collapsing across conditions.

3.3. Primary analyses.

The following are the primary tests we conducted: (a) To test whether we successfully manipulated essentialism with the experimental intervention, we conducted independent-samples t-tests comparing those in the Gender (experimental) condition with those in the Climate (control) condition on: (i) the GES-C full-scale scores and (ii) the Category Boundary gender scores. (b) To test whether the anti-essentialism manipulation resulted in lower levels of prejudice, we conducted independent-samples t-tests comparing those in the Gender (experimental) condition with those in the Climate (control) condition on: (i) the Prejudice against Gender Nonconformity scores, (ii) the Friendship Preference gender scores, and (iii) the Friendship Preference gender-expression scores.

3.4. Extension to race analyses.

In order to determine whether the manipulation affected judgments regarding race, we conducted two additional tests: (a) an independent-samples t-test comparing Category Boundary race scores for those in the Gender (experimental) condition with those in the Climate (control) condition, and (b) an independent-samples t-test comparing Friendship Preference race scores for those in the Gender (experimental) condition with those in the Climate (control) condition.

3.5. Exploratory analyses.

Additionally, as a registered exploratory analysis, we conducted a t-test comparing scores in the Gender (experimental) condition with those in the Climate (control) condition, for each of the four subcomponents of the GES-C scale. We also explored whether age moderated the effects of the intervention on each of the following: the GES-C full scale, the Category Boundary scores for gender, the Prejudice against Gender Nonconformity scores, and the Friendship Preference scores (separately for gender and gender expression), using the PROCESS macro to conduct a bootstrapped regression model for each.

4. Results

4.1. Descriptive and preliminary data.

Means and standard deviations for gender essentialism (GES-C), Prejudice against Gender Nonconformity, Category Boundary Task (separately for gender and race), and Friendship Preference Task (separately for gender, gender-expression, and race) are reported by condition in Table 2. The gender essentialism (GES-C) full scale was reliable, α =.81, as was the Immutability subscale, α =.84. None of the other subscales were reliable (Discreteness: α = .63, Biological Basis: α = .42; Informativity: α = .48).

Table 2.

Means and Standard Deviations of Main Variables

Experimental
Condition		 Control
Condition
Scale Range Mean (SD) Mean (SD)
Gender Essentialism (GES-C) 1 to 4 2.42 (0.54) 2.51 (0.57)
Prejudice against GNC −5 to +5 0.32 (1.41) 0.12 (1.52)
Category Boundary - Gender 0 to 2 1.67 (0.65) 1.59 (0.74)
Category Boundary - Race 0 to 2 0.75 (0.94) 0.81 (0.92)
Friendship Preference - Gender 0 to 2 1.38 (0.56) 1.29 (0.61)
Friendship Preference - Gender Expression 0 to 2 1.17 (0.61) 1.16 (0.71)
Friendship Preference - Race 0 to 2 1.32 (0.68) 1.00 (0.69)
Open in a new tab

Notes: For all measures, higher values indicate: more essentialist beliefs (GES-C), a greater degree of prejudice against GNC targets, stricter category boundaries, or stronger preference for same-gender, gender-typical, or same-race friends.

For all measures; n = 204, n = 102 per condition, except for Friendship Preference - Gender (n = 203, Experimental n = 102, Control n = 101), and Friendship Preference - Race, (n = 165, Experimental n = 85, Control n = 81).

Table 3 displays Pearson’s r correlations among the following: GES-C full-scale score, Prejudice against Gender Nonconformity score, Category Boundary gender items, Friendship Preference gender items, Friendship Preference gender-expression items, and child age (as a continuous variable), collapsing across conditions. Greater gender essentialism, as indicated by the full-scale GES-C, was associated with more prejudice against gender-nonconforming targets, r = .48, p < .0001, and with a stricter assertion of gender category boundaries on the Category Boundary gender items, r = .21, p = .002. Stricter assertion of gender-category boundaries was also associated with increased prejudice against gender-nonconforming children, r = .15, p = .029. With age, children showed less gender essentialism on the GES-C, r = −.32, p < .0001, less prejudice against gender-nonconforming children, r = −.19, p = .006, and less strict gender boundary assertions on the Category Boundary gender items, r = −.19, p =.006.

Table 3.

Zero-order Correlations of Variables

1 2 3 4 5 6
1. GES-C Full -
2. Prejudice against GNC .48 *** -
3. Category Boundary: Gender .21 ** .15* -
4. Friendship Preference: Gender .07 .11 .07 -
5. Friendship Preference: Gender Expression .04 .01 .02 −.05 -
6. Age −.32 *** −.19** −.19** .03 .08 -
Open in a new tab

Notes: Significance: p < .05 *, p < .01 **, p < .001***

For all measures; n = 204, n = 102 per condition, except for Friendship Preference - Gender (n = 203, Experimental n = 102, Control n = 101)

4.2. Primary analyses.

We did not observe significant differences between conditions in gender essentialism on the GES-C full-scale, t(201.5) = −1.21, p = .227, nor in Category Boundary gender scores, t(198.96) = 0.81, p = .421, indicating that the experimental video manipulation did not reduce essentialism overall.

There were no significant differences between conditions on Prejudice against Gender Nonconformity, t(200.92) = 0.98, p = .328, Friendship Preference gender scores, t(199.67) = 1.16, p = .248, or Friendship Preference gender expressions scores, t(197.64) = 0.11, p = .916.

4.3. Extension to race analyses.

To test whether the manipulation affected participants’ judgments regarding race, we conducted a Welch’s independent-samples t-test comparing participants in the Gender (experimental) and Climate (control) condition on Category boundary race scores, as well as a Welch’s independent-samples t-test comparing participants in the two conditions on Friendship Preference race scores. No significant difference was observed between conditions on Category Boundary race scores, t(201.9) = −0.53, p = 0.599. There was a significant difference between conditions on Friendship Preference race scores, t(163.26) = 3.00, p = .003, d = −0.47 such that children in the Climate (control) condition were significantly more likely to choose playmates of a different race than those in the Gender (experimental) condition.

4.4. Exploratory analyses.

As registered exploratory analyses, we conducted t-tests comparing scores in the Gender (experimental) condition with those in the Climate (control) condition, for each of the four subcomponents of the GES-C scale. Means and standard deviations by condition for subscales are reported in Table 4. Participants did not significantly differ by condition on three of the four subscales: Biological Basis, t(201.55) = −0.86 p = .388, Informativeness, t(202) = 1.64, p = 0.103, and Immutability, t(201.36) = −0.64, p = .524. However, participants in the Gender (experimental) condition had lower scores on the Discreteness dimension, t(201.91) = −3.46, p = .001, d = 0.49.

Table 4.

Means and Standard Deviations of GES-C Subscales

Experimental Condition Control Condition
Mean (SD) Mean (SD)
Biological Basis 2.57 (0.67) 2.66 (0.70)
Discreteness 2.29 (0.70) 2.63 (0.71)
Informativeness 1.84 (0.59) 1.71 (0.59)
Immutability 2.98 (0.99) 3.06 (0.93)
Open in a new tab

Notes: Higher values indicate more endorsement of each dimension of essentialism. For all subscales; n = 204, n = 102 per condition. The range for all subscales was 1-4.

We also conducted a bootstrapped moderation analysis examining whether age interacted with the effects of the intervention on the GES-C full scale, the Category Boundary scores for gender, the Prejudice against Gender Nonconformity scores, and the Friendship Preference scores (separately for gender and gender expression), each using 10,000 bootstrapped samples. Main effects of age were observed for gender essentialism (GES-C), Category Boundary gender items, and Prejudice against Gender Nonconformity, consistent with observed zero-order correlations. There were no main effects of age for the Friendship Preferences scores. No significant interactions were observed. Full bootstrapped regression results are presented in Table 5.

Table 5.

Bootstrapped Moderation Analyses

Variable Coefficient Mean SE 95% CI
Lower Upper
Gender Essentialism (GES-C)
  Constant 2.47 2.47 0.04 2.40 2.54
  Condition 0.09 0.09 0.07 −0.06 0.23
  Age −0.13 −0.13 0.03 −0.18 −0.08
  Condition * Age −0.02 −0.02 0.05 −0.12 0.09
Category Boundary - Gender Items
  Constant 1.63 1.63 0.05 1.53 1.72
  Condition −0.08 −0.08 0.10 −0.27 0.11
  Age −0.10 −0.10 0.04 −0.16 −0.03
  Condition * Age 0.01 0.01 0.07 −0.13 0.15
Prejudice Against GNC
  Constant 0.22 0.22 0.10 0.02 0.42
  Condition −0.21 −0.21 0.20 −0.60 0.18
  Age −0.20 −0.20 0.08 −0.35 −0.06
  Condition * Age 0.08 0.08 0.15 −0.22 0.38
Friendship Preference - Gender Items
  Constant 1.34 1.33 0.04 1.26 1.41
  Condition −0.09 −0.10 0.08 −0.26 0.06
  Age 0.01 0.01 0.03 −0.05 0.07
  Condition * Age 0.02 0.02 0.06 −0.10 0.14
Friendship Preference - Gender Expr. Items
  Constant 1.16 1.16 0.05 1.07 1.25
  Condition −0.01 −0.01 0.09 −0.19 0.17
  Age 0.04 0.04 0.03 −0.02 0.10
  Condition * Age −0.02 −0.02 0.07 −0.15 0.11
Open in a new tab

5. Discussion

We tested whether being exposed to gender-based anti-essentialist messages would reduce gender essentialism and prejudice against gender-nonconforming peers in 6- to 10-year-old children in non-metropolitan Midwestern communities. We also tested whether these messages would have an impact on essentialism and prejudice regarding race. This anti-essentialist messaging did not shift either (overall) essentialism or prejudice, for either gender or race — as compared to an unrelated control condition containing messaging about different climates. It is perhaps not surprising that the anti-essentialist messaging did not shift prejudice, given that it did not shift essentialism overall. In brief, our data did not provide support for a causal relationship in which gender essentialism affects prejudice against gender nonconformity.

It is worth noting, however, that we did obtain one condition difference in a registered exploratory analysis. Namely, on judgments of gender discreteness, those in the Gender (experimental) condition endorsed less gender discreteness than those in the Climate (control) condition. We suspect this result may reflect that the intervention emphasized the similarities between boys and girls explicitly (e.g., “Girls and boys are a lot alike”), which directly counters discreteness, as expressed on the Discreteness subscale items (e.g., “Boys and girls are opposites.”). The video was designed to also provide anti-essentialist messaging about the biological basis of gender (by noting that boys and girls are alike in their internal and external anatomy) as well as the informativity of gender (by noting that boys and girls are alike in their interests and activities). However, the clarity of the discreteness messaging may have come through most strongly, as it was most consistent throughout. Regardless, caution should be taken in interpreting this finding unless it is replicated, given that it was the only dimension revealing a condition effect, and the subscale did not reach a conventional level of reliability.

Why might we have seen only a limited difference in essentialism across the two conditions? There are several possibilities, any or all of which may have been operating. One possibility is that the messaging may have emphasized 'boys' and 'girls' as identities, to the exclusion of other ways of grouping people, and thus reminded children of their existing gender concepts and stereotypes (Bigler, 1995; Spinner et al., 2021; Zosuls et al., 2009). A second possibility is that, although the messaging about boys and girls was inclusive, the use of generalizing language about these categories (e.g., “Girls and boys are a lot alike”, also known as generics) in the narration of the video may have inadvertently increased gender essentialism (see Rhodes et al., 2012). A third possibility is that the manipulation was not effective because it did not provide an alternative (non-essentialist) account for why boys and girls often do tend to differ in current society. Perhaps children need a structural explanatory account to replace an essentialist framework (Vasilyeva et al., 2018; Yang et al., 2022). Finally, the manipulation may have simply been too brief. We had designed the video to be brief, as our goal in this study was to determine whether a short manipulation within a single session could be effective. However, this leaves open the possibility that a more in-depth intervention would be more effective. It is possible that a few minutes of a single intervention is not enough to override years of explicit and implicit teaching about gender (e.g., Gelman et al., 2004). Perhaps longer-term interventions would be more suitable to shifting children’s essentialist beliefs and prejudice (see Spinner et al., 2021, for a recent example of a successful long-term intervention regarding gender stereotypes).

Although we did not observe an effect of our experimental manipulation on essentialism or prejudice and thus cannot establish a causal link between the two constructs, we did find an association between them, such that children exhibiting greater gender essentialism also exhibited more prejudice against gender-nonconforming individuals, consistent with the correlation observed by Fine et al. (2024). Other studies have similarly observed a significant association between essentialism and prejudice while failing to find a causal link between the two (e.g., Glazier et al., 2021 with an adult sample). It is possible that a third variable (e.g., gender roles and related social pressures) might be driving the essentialism and prejudice scores, meaning they are not causally related themselves.

We also found that older children essentialized gender less, and demonstrated less prejudice against gender-nonconforming targets, than younger children. This is consistent with prior work on age-related reductions in gender essentialism (Fine et al., 2024) and non-gender prejudice over middle childhood (Raabe & Beelmann, 2011); prejudice in particular has been observed to peak at 5-7 years of age, then decrease through 8-10. Notably, however, in certain contexts, essentialism and many types of prejudice can increase into adolescence and adulthood, especially in contexts that support such beliefs or attitudes (Raabe & Beelmann, 2011; Rhodes & Gelman, 2009). It is thus important to consider age effects within their cultural context, rather than as invariant maturational effects.

One unexpected finding was that children in the Climate (control) condition were more likely to select other-race friends than those in the Gender (experimental) condition. However, given that the finding was unexpected, that the main manipulation check was not successful, and that a relatively large number of tests were run, we also cannot rule out the possibility that this result was significant by chance. Unless and until this finding is replicated, we would urge caution before interpreting it.

In addition to the limitations discussed above, the concurrent political and social environment likely affected the composition of our final sample. At the time of testing, issues involving gender (especially gender nonconformity in children) were highly visible and polarized (Associated Press, 2023). Recruiting participants for a study on children’s beliefs about gendered behaviors with our target communities (more rural, conservative areas of the Midwestern U.S.) proved challenging. We received emails and comments on our social media advertisements expressing concerns about the content of the study. Although participation in our study posed no significant risk, these parents’ concern for their children was real. As seen in our demographics table, our sample skewed toward those who were relatively well-off, educated, and politically liberal, even though participants lived in areas that were generally less well-off, had fewer adults with college degrees, and tended to be politically conservative. As a result, the participating families may have been more accepting of gender diversity than other families in their areas, thus limiting the generalizability of our results to their larger communities. The reluctance of other families in these communities to participate is a limitation. It is possible that children in more conservative families would have found the messages in our intervention more novel and thus potentially more effective in introducing new ways of thinking.

6. Conclusions

Although we were not successful in reducing children’s overall gender essentialism in the present study, we found that, within the age range of 6-10 years, an explicitly worded anti-gender-essentialist video manipulation was effective in conveying to children that boys and girls are alike in many ways. This itself was not enough to reduce prejudice against gender nonconformity — perhaps in part because the intervention was designed to be brief (to keep momentum and hold participants’ attention through a series of tasks), and in part because we were unable to target all areas of essentialism equally in such a short video. We encourage future work making use of more intensive interventions, as well as assessments beyond the period directly following viewing, and beyond the lab. We end by acknowledging that gender as a construct is highly complex and varied. As such, there is much room to explore ways to support children’s understanding and acceptance of gender expressions and identities beyond stereotypical portrayals of “boy” and “girl” alone.

Supplementary Material

Supplementary Material

Research Highlights.

  • Consistent with prior research, we found that greater gender essentialism was associated with greater prejudice against gender-nonconforming children; both decreased with age.

  • We randomly assigned children to view either an anti-essentialist video manipulation or a control video to test if this relation was causal in nature.

  • The anti-essentialist video did not reduce overall essentialism as compared to the control, so we did not find support for a causal link.

  • We observed a reduction in the dimension of essentialism most closely linked to the anti-essentialist video language, suggesting the potential utility of anti-essentialist messaging.

Acknowledgements.

The project was approved by the University of Michigan Institutional Review Board: “Children’s Gender Understanding Project (CGUP)” (HUM00154316). Written informed consent was obtained from the parent/guardian of all participants, and child assent was obtained orally. Research reported in this publication was supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health under Award Number R01HD092347. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. We would like to thank Athena Bryer, Katie Chang, Esther Kim, Kalina McNeil, Sydney Boland, Sara Morehouse, and Jade Crosby, for their excellent assistance in study organization, recruitment, and data collection; without their efforts this work would not have been possible. We would also like to thank Natalie M. Gallagher and Jackson H. Loper for technical assistance in reviewing our analyses for publication.

Glossary

Sex

The category assigned to a person at or before birth, based on their genitals or other biological markers. The most common sexes are male, female, and intersex.

Gender Identity

A person's gender self-identification. The most common gender identities are girl/woman, boy/man, and nonbinary.

Gender Expression

The way in which a person expresses aspects of their gender, typically through their appearance, dress, and behavior. Gender expression is often thought to run along a masculine and/or feminine spectrum.

Gender Conforming

People whose gender identity and gender expression align in ways consistent with cultural expectations for their sex assigned at birth.

Gender Nonconforming (GNC)

People whose gender identity and/or gender expression do not align in ways consistent with cultural expectations for their sex assigned at birth. Examples include assigned females who live as boys, boys who primarily wear dresses, and children who identify as neither boys nor girls.

Footnotes

We have no conflicts of interest to disclose.

The data that support the findings of this study are openly available in Open Science Framework (OSF) at: https://osf.io/z8p9c/.

1

The registered wording of this measure was, “How much do you like this kid?” After running the first six participants, we modified the presentation into a two-step question (as written above), as younger participants were struggling to use the 6-point scale.

2

The actual wording (above) differed slightly from the registered wording, which stated that children would be asked whether Feppy was “maybe right” or “wrong” in stating that both warm-up trial images (cat/cat; pig/goat) and trial image pairs were the “same kind” of animal/person. In the measure as run, children provided “yes” or “no” responses to the questions asking whether Feppy was “maybe right.”

3

We had registered that if more than 25% of the sample had been children whose parents gave their race as other than White or Black/African American, scoring would be based on preference for White children. Because this did not occur, we used the scoring described in the main text.

References

  1. The Associated Press. (2023, April 6). The governors from Indiana and Idaho sign bans on gender-affirming care. NPR. https://www.npr.org/2023/04/06/1168323019/the-governors-from-indiana-and-idaho-sign-bans-on-gender-affirming-care [Google Scholar]
  2. Berndt TJ, & Heller KA (1986). Gender stereotypes and social inferences: A developmental study. Journal of Personality and Social Psychology, 50(5), 889–898. 10.1037/0022-3514.50.5.889 [DOI] [Google Scholar]
  3. Bigler RS (1995). The role of classification skill in moderating environmental influences on children’s gender stereotyping: A study of the functional use of gender in the classroom. Child Development, 66(4), 1072–1087. [Google Scholar]
  4. Conry-Murray C, Kim JM, & Turiel E (2015). Judgments of gender norm violations in children from the United States and Korea. Cognitive Development, 35, 122–136. 10.1016/j.cogdev.2015.04.002 [DOI] [Google Scholar]
  5. Davoodi T, Soley G, Harris PL, & Blake PR (2020). Essentialization of social categories across development in two cultures. Child Development, 91(1), 289–306. 10.1111/cdev.13209 [DOI] [PubMed] [Google Scholar]
  6. Diesendruck G, Goldfein-Elbaz R, Rhodes M, Gelman SA, & Neumark N (2013). Cross-cultural differences in children’s beliefs about the objectivity of social categories. Child Development, 84(6), 1906–1917. 10.1111/cdev.12108 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Faul F, Erdfelder E, Lang A-G, & Buchner A (2007). G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behavior Research Methods, 39(2), 175–191. 10.3758/BF03193146 [DOI] [PubMed] [Google Scholar]
  8. Fine R, Olson KR, Gülgöz S, Horton R, & Gelman SA (2024). Gender essentialism predicts prejudice against nonconformity in two cultural contexts. Social Development, 33(1), e12720. 10.1111/sode.12720 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Gelman SA (2003). The essential child: Origins of essentialism in everyday thought. Oxford University Press. [Google Scholar]
  10. Gelman SA (2004). Psychological essentialism in children. Trends in Cognitive Sciences, 8(9), 404–409. 10.1016/j.tics.2004.07.001 [DOI] [PubMed] [Google Scholar]
  11. Gelman SA, Taylor MG, & Nguyen S (2004). Mother-child conversations about gender: Understanding the acquisition of essentialist beliefs. Monographs of the Society for Research in Child Development. Volume 69, No. 1. doi: 10.1111/j.1540-5834.2004.06901002.x [DOI] [Google Scholar]
  12. Gelman SA, Collman P, & Maccoby EE (1986). Inferring properties from categories versus inferring categories from properties: The case of gender. Child Development, 57(2), 396–404. 10.2307/1130595 [DOI] [Google Scholar]
  13. Glazier JJ, Gomez EM, & Olson KR (2021). The association between prejudice toward and essentialist beliefs about transgender people. Collabra: Psychology, 7(1), 25528. 10.1525/collabra.25528 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gülgöz S, Gomez EM, DeMeules MR, & Olson KR (2018). Children’s evaluation and categorization of transgender children. Journal of Cognition and Development, 19(4), 325–344. 10.1080/15248372.2018.1498338 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Haslam N, Bastian B, & Bissett M (2004). Essentialist beliefs about personality and their implications. Personality and Social Psychology Bulletin, 30(12), 1661–1673. 10.1177/0146167204271182 [DOI] [PubMed] [Google Scholar]
  16. Haslam N, Rothschild L, & Ernst D (2000). Essentialist beliefs about social categories. British Journal of Social Psychology, 39(1), 113–127. 10.1348/014466600164363 [DOI] [PubMed] [Google Scholar]
  17. Hirschfeld LA (1995). Do children have a theory of race? Cognition, 54(2), 209–252. [DOI] [PubMed] [Google Scholar]
  18. Kalish C. (1998). Natural and artifactual kinds: Are children realists or relativists about categories? Developmental Psychology, 34(2), 376. 10.1037/0012-1649.34.2.376 [DOI] [PubMed] [Google Scholar]
  19. Kosciw JG, Clark CM, Truong NL, & Zongrone AD (2020). The 2019 National School Climate Survey: The experiences of lesbian, gay, bisexual, transgender and queer youth in our nation’s schools. New York: Gay, Lesbian and Straight Education Network (GLSEN). https://www.glsen.org/research/2019-national-school-climate-survey [Google Scholar]
  20. Leslie S-J (2013). Essence and natural kinds: When science meets preschooler intuition. In Gendler TS & Hawthorne J (Eds.), Oxford Studies in Epistemology (Vol. 4, pp. 108–165). Oxford University Press. [Google Scholar]
  21. Meyer M, & Gelman SA (2016). Gender essentialism in children and parents: Implications for the development of gender stereotyping and gender-typed preferences. Sex Roles, 75(9), 409–421. 10.1007/s11199-016-0646-6 [DOI] [Google Scholar]
  22. Pauker K, Williams A, & Steele JR (2016). Children’s racial categorization in context. Child Development Perspectives, 10(1), 33–38. 10.1111/cdep.12155 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Raabe T, & Beelmann A (2011). Development of ethnic, racial, and national prejudice in childhood and adolescence: A multinational meta-analysis of age differences. Child Development, 82(6), 1715–1737. 10.1111/j.1467-8624.2011.01668.x [DOI] [PubMed] [Google Scholar]
  24. Rhodes M, & Gelman SA (2009). A developmental examination of the conceptual structure of animal, artifact, and human social categories across two cultural contexts. Cognitive Psychology, 59(3), 244–274. 10.1016/j.cogpsych.2009.05.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Rhodes M, Leslie S-J, & Tworek CM (2012). Cultural transmission of social essentialism. Proceedings of the National Academy of Sciences, 109(34), 13526–13531. 10.1073/pnas.1208951109 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Rhodes M, & Mandalaywala TM (2017). The development and developmental consequences of social essentialism. WIREs Cognitive Science, 8(4), e1437. 10.1002/wcs.1437 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Roberts SO, Gelman SA, & Ho AK (2017). So it is, so it shall be: Group regularities license children’s prescriptive judgments. Cognitive Science, 41(S3), 576–600. 10.1111/cogs.12443 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Shutts K, Kenward B, Falk H, Ivegran A, & Fawcett C (2017). Early preschool environments and gender: Effects of gender pedagogy in Sweden. Journal of Experimental Child Psychology, 162, 1–17. 10.1016/j.jecp.2017.04.014 [DOI] [PubMed] [Google Scholar]
  29. Spinner L, Tenenbaum HR, Cameron L, & Wallinheimo A-S (2021). A school-based intervention to reduce gender-stereotyping. School Psychology International, 42(4), 422–449. 10.1177/01430343211009944 [DOI] [Google Scholar]
  30. Taylor MG (1996). The development of children’s beliefs about social and biological aspects of gender differences. Child Development, 67(4), 1555–1571. 10.1111/j.1467-8624.1996.tb01814.x [DOI] [PubMed] [Google Scholar]
  31. Taylor MG, Rhodes M, & Gelman SA (2009). Boys will be boys; cows will be cows: Children’s essentialist reasoning about gender categories and animal species. Child Development, 80(2), 461–481. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Tworek CM, & Cimpian A (2016). Why do people tend to infer “ought” from “is”? The role of biases in explanation. Psychological Science, 27(8), 1109–1122. 10.1177/0956797616650875 [DOI] [PubMed] [Google Scholar]
  33. Vasilyeva N, Gopnik A, & Lombrozo T (2018). The development of structural thinking about social categories. Developmental Psychology, 54(9), 1735–1744. 10.1037/dev0000555 [DOI] [PubMed] [Google Scholar]
  34. Yang X, Naas R, & Dunham Y (2022). Testing the limits of structural thinking about gender. Developmental Science, 25(2), e13169. 10.1111/desc.13169 [DOI] [PubMed] [Google Scholar]
  35. Zosuls KM, Ruble DN, Tamis-LeMonda CS, Shrout PE, Bornstein MH, & Greulich FK (2009). The acquisition of gender labels in infancy: Implications for sex-typed play. Developmental Psychology, 45(3), 688–701. 10.1037/a0014053 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
NIHMS2078730-supplement-Supplementary_Material.docx (37.8KB, docx)

RESOURCES