Sex Roles

, Volume 68, Issue 5, pp 378–389

Children’s Gender Stereotypes Through Drawings of Emotional Faces: Do Boys Draw Angrier Faces than Girls?

Original Article

DOI: 10.1007/s11199-012-0242-3

Cite this article as:
Brechet, C. Sex Roles (2013) 68: 378. doi:10.1007/s11199-012-0242-3

Abstract

The present study was designed to examine the impact of display rules and gender-emotion stereotypes on French children’s depiction of sadness and anger in their drawings of a human face. Participants were 172 school-aged French children (74 boys and 98 girls), who attended state schools in a middle-class district of a southern French city. The exact age range was as followed: 6 years 2 months to 8 years 1 month. They were asked to draw the emotion felt by a character (either male or female) after being told a scenario eliciting sadness and a scenario eliciting anger. By never mentioning the emotion felt by the character, we expected children’s interpretation of these scenarios to be therefore influenced by their own gender and/or by the character’s gender. Results indicate that anger is depicted by more boys than girls in response to the angry scenario, for male as well as for female characters. Furthermore, among the children who did depict anger, the expressive intensity of the drawings was scored lower for children who were presented the feminine character than for children who were presented the masculine character. However, no effect of gender was found on the drawings produced in response to the sad scenario. These results are discussed in terms of the influence of display rules and gender-emotion stereotypes on children. We also suggest some methodological and clinical implications.

Keywords

Display rules Gender stereotypes Emotions Children Drawing 

Introduction

Boys should not cry and girls should not get angry. The existence of such gender-emotion stereotypes is now well recognized. By considering sadness as a feminine emotion and anger as a masculine one, we can refer to gendered emotions (for a detailed discussion, see Fischer 2000). According to McNaughton (2000), children devote a substantial amount of time attempting to behave like a ‘normal girl’ or a ‘normal boy’ and to conform to their own culture’s views of masculinity and femininity. Many studies have demonstrated that children as young as 5 years old conform to such gender-emotion stereotypes, when it comes to judging other people’s emotions or to express emotions themselves. These studies mostly used U.S. samples (e.g., Hubbard 2001; Karbon et al. 1992). Cross-cultural studies indicate that such differences are reported in many western countries, although the size of the difference varies considerably among countries (Fischer 2000). More recently, researchers from many countries (e.g., France, U.K., U.S., Turkey) have also been interested in the way children depict emotions in their drawings of a human figure (i.e., expressive drawings) (e.g., Brechet et al. 2009; Jolley 2009; Missaghi-Lakshman and Whissell 1991; Sayil 2001). In many countries, children’s drawings are considered as a reflection of children’s representations and as a strong tool of non verbal communication, enabling thoughts and emotions to be expressed visually (see Baldy 2002, for France; Brooks 2005, for Australia; Freeman 1980 and Gombrich 1972, for U.K.; Malchiodi 1998, for the U.S.). Surprisingly, little is known about the potential effect of gender on the way children depict gendered emotions.

The present study builds upon these two fields of research by examining the impact of gender-emotion stereotypes on French children’s drawings of emotional faces. First, the present study provides a contribution to research from other authors in many countries and builds on other studies published in Sex Roles and using U.S. samples (e.g., Birnbaum and Chemelski 1984; Brody et al. 1995; Karbon et al. 1992). Indeed, as this study is the first to use a non verbal drawing task to test gender-emotion stereotypes, our aim is to determine whether this could be a valid method of examining French children’s representations of gendered emotions. Secondly, on a more clinical point of view, as the use of expressive drawings is a world-wide phenomenon in children’s psychological evaluations (e.g., see Camara et al. 2000, for the U.S.; Chan and Lee 1995, for Hong Kong; Picard and Baldy 2011, for France), this study could help to understand more fully which mechanisms are involved in boys’ and girls’ production of expressive drawings, in regard to emotion and gender socialization.

Gender Socialization in France

The studies we review in this paper mostly involve English speaking participants from the U.S. and the U.K. As the present study was conducted with a sample of French boys and girls, it seems important to first examine gender socialization and gender stereotypes in a cross-cultural perspective. The Hofstede’s masculinity index reports cultural differences in the strength of stereotyped ideas relative to masculinity and femininity (Hofstede 2003). For instance, Japan is the most “masculine” country, with an index of 95, whereas Sweden is the less “masculine” country, with an index of 5. The U.S. and the U.K. share, in terms of masculinity, similar values (i.e., index of 62 and 66, respectively) and can be considered as more “masculine” countries than France (i.e., index of 43). The question of children’s gender-stereotypes has been a rather neglected area of research so far in France. However, a few studies of interest can be presented, so as to locate the present study within a French context. For instance, Granié (2010) assessed 6-year-old French children’s conformity to gender stereotypes, by using the Pre School Activities Inventory (PSAI, Golombok and Rust 1993), which measures differences between boys and girls in terms of their “masculinity” and “femininity”, based on three scales (i.e., toys, activities and traits). Granié (2010) showed that French girls had a higher femininity score on the PSAI than French boys, and that French boys had a higher masculinity score on the PSAI than French girls. Indeed, various observations indicate that French children conform to gender stereotypes from the age of 3. For instance, Dafflon Novelle (2006) asked 3 to 6-year-old French children to match the picture of a boy and the picture of a girl with pictures of masculine versus feminine objects (e.g., a car versus a doll). Results indicated that boys and girls equally matched the feminine picture with stereotypically feminine objects and the masculine picture with stereotypically masculine objects. Soussignan and Schaal (1996) asked 5 to 12-year-old French children to smell pleasant and unpleasant odors and examined children’s reactions. They found that girls tended to smile more than boys in response to unpleasant odors, when placed in a social condition. These studies indicated that gender has an influence on French children’s attitudes, towards people and objects. Finally, Syssau and Monnier (2009) asked 5, 7 and 9-year-old French boys and girls to rate a corpus of 600 French words, in terms of emotional valence (i.e., positive, neutral and negative). They found little gender differences in the ratings: 6 words out of 600 were rated differently by boys and girls. Interestingly, though, the word “anger” (i.e., “colère” in French) was rated as negative by 7-year-old girls and as positive by 7-year-old boys.

Children’s Gender-Emotion Stereotypes Through Attribution and Expressiveness Studies

Some of the early studies involving children’s use of gender-emotion stereotypes (i.e., attribution studies) consisted in asking children either to determine a character’s gender as a function of the emotion he/she expressed or felt, or to determine the emotion expressed or felt by a character according to its gender. For instance, Birnbaum and Chemelski (1984) told 3 to 5-year-old U.S. children short stories featuring a character in situations eliciting anger and fear, and asked them to determine whether the story’s main character was male or female. Results indicated that children judged the character of the fear story as feminine, whereas they judged the character of the anger story as masculine. Birnbaum et al. (1980) reported similar results, when presenting drawings of emotional faces to U.S. children, asking them to identify the character’s gender according to the emotion expressed. Finally, Haugh et al. (1980) asked 3 to 5-year-old U.S. children to determine the emotion expressed by a baby according to its gender. The same baby was presented to the participants (boys and girls) and the gender of the baby was determined by the experimenter. Results indicated that children attributed feminine emotions (e.g., sadness) to the baby when it was presented as a baby girl and masculine emotions (e.g., anger) when it was presented as a baby boy. Overall, these studies reported an effect of the character’s gender on children’s responses: children avoided associating feminine characters with masculine emotions and masculine characters with feminine emotions. They based their judgments on gender-emotion stereotypes instead.

Some attribution studies also asked children to determine the intensity of a target emotion felt by a character according to its gender. For instance, Karbon et al. (1992) told 5-year-old U.S. children short stories featuring a male or female character in situations eliciting sadness and anger. Then, they asked children to determine the intensity of the emotion of sadness and anger felt by each character. Children attributed a high intensity of sadness and a low intensity of anger to female characters. On the contrary, they attributed a high intensity of anger and a low intensity of sadness to male characters. Identical findings were reported for anger and fear by Brody et al. (1995) who observed 6-year-old U.S. children. To sum up, these studies reported an effect of the character’s gender on children’s responses: children minimized the intensity of feminine emotions for male characters and the intensity of masculine emotions for female characters.

Note that none of these attribution studies indicated any effect of children’s gender on their responses. However, another kind of studies reported an effect of children’s gender on their own expressiveness of sadness and anger. For sadness, Oliver and Green (2001) examined U.S. children’s expressive responses (from 3 to 9 years old) to emotional movie scenes and showed that girls were more likely than boys to express sadness in response to a sad scene. For anger, Hubbard (2001) showed that U.S. boys’ angry facial expressions were more frequent and more intense than those expressed by girls, when responding to emotion-provoking interactions through a competitive game paradigm. Thus these studies indicated that children tended to express emotions, which were stereotypically associated with their own gender.

However, by proposing a limited choice of answers (e.g., sadness or anger, male or female character), experimental designs used in attribution tasks maximize the impact of gender, and have consequently been useful to establish the existence of such stereotypes. Besides, these studies asked children to produce verbal responses. In the area of emotion research, females are known to outperform males in tasks requiring verbal responses. This difference has been reported with U.S. samples (Wester et al. 2002), as well as with French samples (Blanc 2010). Thus, it could be interesting to test those stereotypes by means of a free choice response method, in order to prevent the over-evaluation of the gender stereotypes effect and to cancel out any potential effect of the differences relative to verbal abilities between male and female participants.

Children’s Gender Stereotypes Through Drawings

Another way to study gender stereotypes consists in asking children to freely draw the character. Being considered as a reflection of how the child understands what is around him/her (Baldy 2002; Cox 2005; Freeman and Cox 1985; Golomb 1992; Goodnow 1977), drawing seems an appropriate way to test gender stereotypes. Various studies examined the effect of gender-role stereotypes, through children’s drawing of the human figure. These studies consisted in asking children, from 5 to 16, to freely draw characters presenting a highly stereotyped social role (a masculine one) and then observing the character’s gender drawn by the child. For instance, Colley et al. (2005) asked British children to draw a sportsperson, Chambers (1983) asked Australian children to draw a scientist, Fournier and Wineburg (1997) asked U.S. children to draw a western settler and Brosnan (1999) asked British children to draw a computer-user. In all of these studies, results indicated that boys exclusively draw male characters; whereas girls draw as many male characters as female characters. Thus, the results of these studies show differences when comparing to those obtained with attribution tasks. Indeed, a greater effect of gender stereotyping was found with regard to boys. But it must be noted that the social roles tested were stereotypically male roles (i.e. sportsperson, scientist, western-settler, computer-user) and that the task consisted in letting the child freely draw the character, according to its social role. Yet, it has been shown that children performing a free drawing task strongly tended to draw a character of their own gender. This tendency has been shown with French children (Baldy 2002), British children (Cox 1993) and U.S. children (Wieder and Noller 1950). As this tendency is deeply rooted in their graphic routines (Berti and Freeman 1997), children naturally find it difficult to escape from them. Interestingly, some other studies indicated that children’s graphic routines could be altered, under certain conditions (e.g., the exposure to examples of what is to be drawn) (Burkitt and Barrett 2010; Hollis and Low 2005). From there, if we combine observations, the difference noticed between boys and girls could be simply due to the fact that female drawers followed their graphic routine by drawing female characters (their graphic routine overriding the gender stereotypes effect).

The choice of a drawing task seems an interesting way to examine gender stereotypes. Indeed, contrary to the attribution tasks presented above, drawing tasks do not require a verbal response and do not propose a limited choice of answers. But it may be more appropriate, instead of proposing a free drawing task, to use a drawing task in which the character’s gender would be forced. This experimental design could enable us to ensure the findings were not distorted by children’s graphic routine. Finally, it is worth noting that these studies, using human figure drawings, only focused on gender-role stereotypes. No study had used drawing to examine children’s gender-emotion stereotypes so far.

However, numerous studies have focused on U.S., British, French and Turkish children’s ability to depict emotions in their human figure drawing (Brechet et al. 2009; Brechet et al. 2007; Cox 2005; Golomb 1992; Missaghi-Lakshman and Whissell 1991; Sayil 2001). In general, results indicated that children were able to depict happiness from the age of 5, sadness and anger from the age of 6 and surprise, fear and disgust only from the age of 8, in a completion task. Children usually depict these emotions in a very prototypical way. For instance, sadness is always depicted through tears and/or a downturned mouth; whereas anger is always depicted through a downturned mouth and downturned eyebrows. Brechet et al. (2009) proved that the expressive drawing task (e.g. draw a happy/sad/angry human figure) perfectly reflected French children’s level of emotion conceptualization. Yet, in these studies, the potential effect of the child’s gender on the depiction of an expressive human figure has never been assessed and the potential effect of the character’s gender has been subject to little experimental control (Brechet et al. 2009). However, Picard and Boulhais (2011) found some differences between French boys’ and girls’ expressive drawings of a tree. They explicitly asked children to freely draw a happy and a sad tree. Their results indicated that girls tended to depict these emotions, by using a lot more details than boys (e.g., bright/dark colours, tree surrounded with sun/clouds). But no difference was revealed between boys and girls, in regard to the type and number of facial cues used in the drawings. From then, we can infer that the use of a drawing task focusing on facial cues (i.e., instead of a free drawing task) should prevent gender differences in children’s drawing abilities. Given that the expressive drawing of a human figure is a strong indication of children’s emotion conceptualization, which is influenced by gender-emotion stereotypes, we can assume that there must be a strong effect of the child’s and/or character’s gender in the graphic depiction of emotions in a human figure drawing (especially of gendered emotions).

The Present Study

The present study was designed to examine the impact of gender-emotion stereotypes through drawing of emotional faces. To this purpose, we examined the effect of the child’s and the character’s gender on the depiction of basic emotions in school-aged children’s human figure drawing. Our task consisted in asking children to draw the emotion felt by a character (either male or female) after being told a scenario eliciting sadness and a scenario eliciting anger. By never mentioning the emotion felt by the character, we expected children’s interpretation of these scenarios to be therefore influenced by their own gender and/or by the character’s gender. Because display rules and gender-emotion stereotypes especially involve facial cues of emotion (i.e., rather than content cues for instance) and because there is no gender difference between boys and girls in the ability to depict emotional facial cues (Picard and Boulhais 2011), we chose to use a completion task of emotional faces (i.e., masculine versus feminine faces). Children were asked to complete blank pre-drawn faces with emotional facial cues. In this way, we would be able to make up for the limits of attribution tasks (limited choice of answers) and free drawing tasks (influence of graphic routine). Indeed, in forcing the character’s gender, we could observe not only how children depict the emotions of a same-gender character, but also how they depict the emotions of an opposite-gender character. To do this, we selected two basic emotions children are able to depict from the age of 6: a stereotypically feminine emotion (sadness) and a stereotypically masculine one (anger). We chose to observe school-aged children (6–8 years old), as previous studies indicated that children that age range were able to depict both sadness and anger and were sensitive to gender-emotion stereotypes.

With regard to results reported through attribution studies (e.g., Birnbaum et al. 1980; Karbon et al. 1992), we expected the character’s gender to have an effect on children’s responses, according to gender-emotion stereotypes and display rules. Furthermore, our task consisting in producing graphic emotional cues, we predicted an effect of the child’s gender on responses, in line with results from expressiveness studies (e.g., Hubbard 2001; Oliver and Green 2001). Indeed, as children’s drawings enable thoughts and emotions to be expressed visually, they are considered as a strong tool of non verbal communication (Baldy 2002; Brooks 2005; Gombrich 1972; Malchiodi 1998). Consequently, it is likely that children’s own gender could have an effect on their drawings. By considering sadness to be a stereotypically feminine emotion and anger to be a stereotypically masculine emotion, display rules lead to two types of responses: either a masking of the emotion which does not comply with the gender (i.e., by substituting any other facial behaviour) or a control of its expressive intensity (i.e., by attenuating it) (Saarni 1979). So as to examine these two types of responses, we chose to carry out two types of scoring on the drawings we collected.

The first scoring was designed to assess children’s tendency to mask an emotion which does not comply with the gender (i.e., by depicting another emotion instead, or no emotion at all). By asking adult judges to determine which emotion was expressed in each drawing, we obtained a number of “correct” drawings (i.e., drawings that depicted the emotion evoked in the corresponding scenario). In H1a, we expected the child’s gender and/or the character’s gender to have an effect on the depicted emotion. More precisely, the sad scenario should lead to a lower number of correct sad drawings for boys and/or for the male character, compared to girls and/or the female character. In addition, the angry scenario should lead to a lower number of correct angry drawings for girls and/or for the female character, compared to boys and/or the male character. Such findings would be consistent with a masking of the emotion which does not comply with the child’s own gender and/or with the character’s gender. Children who did not depict the emotion evoked by the scenario should depict an emotion which is more stereotypical of their own gender and/or of the character’s gender instead. For that reason, we also examined the “incorrect drawings” (i.e., drawings depicting another emotion than the one evoked in the scenario) so as to determine which emotions were depicted to mask sadness and anger. In H1b, we expected the incorrect drawings that were produced by boys and/or that were displaying a male character in response to the sad scenario to depict a masculine emotion (e.g., anger) or even no emotion at all (i.e., neutrality). In addition, we expected the incorrect drawings produced by girls and/or displaying a female character in response to the angry scenario to depict a feminine emotion (e.g., sadness, fear) or even no emotion at all (i.e., neutrality).

The second scoring was designed to assess children’s tendency to attenuate the expressive intensity of an emotion which does not comply with the gender. By asking adult judges to assess the correct drawings on a 3-point Likert scale (1 standing for a “low expressive intensity”, 2 standing for “medium expressive intensity”, and 3 standing for “high expressive intensity”), each drawing was given a score of expressive intensity. In H2, we expected the child’s and/or character’s gender to have an effect on the expressive intensity of correct drawings. More precisely, the correct sad drawings produced by boys and/or displaying a male character should be associated with lower scores of expressive intensity, compared to the correct sad drawings produced by girls and/or displaying a female character. In addition, the correct angry drawings produced by girls and/or displaying a female character should be associated with lower scores of expressive intensity, compared to the correct angry drawings produced by boys and/or displaying a male character. Such findings would be consistent with an attenuation of the emotion which does not comply with the child’s and/or the character’s gender.

Summary of Hypotheses

To sum up, based on the above literature, we made the following hypotheses:
  1. H1a.

    The sad scenario should lead to a lower number of correct sad drawings for boys and/or for the male character, compared to girls and/or the female character. On the contrary, the angry scenario should lead to a lower number of correct angry drawings for girls and/or for the female character, compared to boys and/or the male character.

     
  2. H1b.

    The incorrect drawings produced by boys and/or displaying a male character in response to the sad scenario should depict a masculine emotion (e.g., anger) or even no emotion at all (i.e., neutrality). On the contrary, the incorrect drawings produced by girls and/or displaying a female character in response to the angry scenario should depict a feminine emotion (e.g., sadness, fear) or even no emotion at all (i.e., neutrality).

     
  3. H2.

    The correct sad drawings produced by boys and/or displaying a male character should be associated with lower scores of expressive intensity, compared to the correct sad drawings produced by girls and/or displaying a female character. On the contrary, the correct angry drawings produced by girls and/or displaying a female character should be associated with lower scores of expressive intensity, compared to the correct angry drawings produced by boys and/or displaying a male character.

     

Method

Participants

Participants were 172 French children (74 boys and 98 girls) ranged from 6 to 8 years of age (with a mean age of 7.22 years). The exact age range was as followed: 6 years 2 months to 8 years 1 month. Thirty four boys and 48 girls were told scenarios in which the main character was a male (Pierre). The remaining 40 boys and 50 girls were told scenarios in which the main character was a female (Marie). They were recruited from state schools in a middle-class district of a southern French city, and were in their normal school year. Written parental consent was obtained for each participant. A t-test revealed that male and female participants did not differ in age, t (170) = .23, p = .89.

Material

Children had a pencil, a pencil sharpener and a notebook composed of two sheets with pre-drawn boy’s or girl’s faces. We also used two prototypical emotional scenarios based on stories created by Parmley and Cunningham (2007) and by Brechet et al. (2009). In the present study, the scenarios described a male character (Pierre) or a female character (Marie) and evoked the emotions of sadness and anger without naming them. Thus, the emotion children associated with each scenario was a function of their interpretation of the scenarios. The sad scenario was as followed: “Pierre/Marie has a little cat. Every evening at home Pierre/Marie plays with the cat, stroking it and cuddling it. Pierre/Marie loves this cat. But one evening, Pierre/Marie comes home and the little cat isn’t there. Pierre/Marie looks for the cat everywhere, but can’t find it. The little cat is lost” (“Pierre/Marie a un petit chat. Tous les soirs, Pierre/Marie joue avec le chat, le serre dans ses bras et le câline. Mais un soir, Pierre/Marie rentre à la maison et le petit chat n’est pas là. Pierre/Marie le cherche partout, mais ne le trouve pas. Le petit chat a disparu”). The angry scenario was as followed: “Pierre/Marie is sitting down watching a video with the rest of the class. Another kid keeps talking during the video which makes it hard for Pierre/Marie to hear” (“Pierre/Marie est assis(e) en train de regarder une vidéo avec le reste de la classe. Un autre enfant n’arrête pas de parler pendant la vidéo, ce qui empêche Pierre/Marie de bien entendre”). The studies we took these stories from indicated that at the age of 4, more than 70 % of children were able to label the emotions to be evoked (sadness and anger) through these specific stories (Brechet et al. 2009; Parmley and Cunningham 2007).

Procedure

The study was conducted individually, in a quiet room in the school, and lasted an average of 20 min per child. Each child was presented with the two scenarios (sad and angry) involving either a masculine character (Pierre) or a feminine character (Marie). More precisely, 34 boys and 48 girls were presented with Pierre; whereas 40 boys and 50 girls were presented with Marie. After having been told the story, each child was invited to complete a pre-drawn face, following this instruction: “Complete Pierre/Marie’s face so that we can see how he/she feels.” Scenarios evoking sadness and anger were presented in a counterbalanced order. While children were taking part in the completion task, no comment was made about their drawings. Then, children were thanked for their participation. This task allowed us to collect 344 drawings in total (i.e., each of the 172 children completed a drawing in response to the sad scenario and a drawing in response to the angry scenario).

Scoring

Correct Versus Incorrect Drawings

We presented each drawing to three judges. These judges were female psychology students (21 years of age). The choice of the judges’ gender was based on studies indicating than women tended to have a better understanding of non-verbal emotional cues than men (Collignon et al. 2010; Fischer 2000; Hall 1978; McClure 2000). So as to reach inter-rater agreement, the judges were first shown a sample of drawings of emotional faces, which were taken from previous studies in which 6 to 8 years old children were explicitly asked to depict sadness and anger. The judges knew neither about the children’s gender nor about the character’s gender (i.e. the characteristics relative to gender of the pre-drawn face lines were hidden for this assessment). For each drawing, they had to decide which of six basic emotions (happiness, sadness, anger, surprise, fear or disgust) was depicted in the drawing. They were also given a neutral category for drawings depicting no emotion at all. Given that previous studies on expressive drawings indicated that such a forced-choice labeling method was as relevant as a free labeling method, when using a large enough number of possible responses (i.e., between 6 and 7), we chose to delimit the number of possible responses to the six basic emotions and a neutral category (Brechet et al. 2007). We considered a drawing as expressive when at least two of three judges agreed about the emotion depicted in the drawing. When judges did not agree about the emotion depicted, the drawing was considered as non-expressive. We considered an expressive drawing as correct when it corresponded to the target emotion (sadness or anger). Lastly, an expressive drawing was considered as incorrect when it corresponded to an emotion other than the one evoked in the scenario (sadness or anger).

Number of correct drawings. First, we noted the number of correct drawings corresponding to sadness and anger. The inter-rater agreement was 98 % (k coefficient for inter-rater reliability was .98, p < .01).

Distribution of incorrect drawings. Secondly, we focused only on the incorrect drawings. We noted the number of incorrect drawings depicting neutrality, happiness, anger, surprise, fear and disgust, in response to the sad scenario. We also noted the number of incorrect drawings depicting neutrality, happiness, sadness, surprise, fear and disgust, in response to the angry scenario. The inter-rater agreement was 98 % (k coefficient for inter-rater reliability was .98, p < .01). Our aim was to determine which emotion was depicted instead of the one evoked in the scenario. According to gender-emotion stereotypes, happiness, sadness, surprise and fear were considered as stereotypically feminine emotions; whereas anger and disgust were considered as stereotypically masculine ones. As display rules dictate to mask an emotion which does not comply with the gender, by expressing another emotion instead (e.g., fear instead of anger for girls, anger instead of sadness for boys) or by not expressing any emotion (i.e., neutral face), we expected incorrect drawings to be distributed in accordance with gender-emotion stereotypes.

Note that this binary type of scoring, in terms of correct/incorrect drawings, is classically used in research studies about expressive drawing (Brechet et al. 2009; Brechet et al. 2007; Cox 2005; Ives 1984; Missaghi-Lakshman and Whissell 1991; Sayil 2001).

Expressive Intensity of Correct Drawings

We then chose to perform another analysis, only on drawings beforehand considered to be correct. Thus, each correct drawing was shown to three new judges, who were asked to assess its overall expressive intensity on a 3-point Likert scale: 1 for “low expressive intensity”, 2 for “medium expressive intensity”, and 3 for “high expressive intensity”. These judges were female psychology students (21 years of age). For this rating, the judges were not given examples of drawings of the different intensity levels. However, so as to reach inter-rater agreement, the three judges were first shown all of the correct drawings that were to be rated. This way, they had a rather satisfying overview of the different levels of expressivity. The judges knew neither about the children’s gender nor about the character’s gender (i.e. the characteristics relative to gender of the pre-drawn face lines were hidden for this assessment). For this rating, the inter-rater reliability was very good (k = .86, with an inter-rater agreement of 90 %). We obtained a score of expressive intensity for each drawing, from 1 to 3. Out of the 156 drawings, the three judges reached identical scores of expressive intensity for 141 drawings (90 %), and scores differing by 1 point for 15 drawings (10 %). For these 15 drawings, we computed a mean score of expressive intensity, from 1 to 3, which corresponded to the average score that the three judges gave to the drawing. Some drawings are shown in Fig. 1, for the two extreme points of the Likert scale.
Fig. 1

Examples of correct drawings produced in response to the sad and to the angry scenario, for the two extreme points of the Likert scale

Results

Number of Correct Drawings

According to H1a, the sad scenario should lead to a lower number of correct sad drawings for boys and/or for the male character, compared to girls and/or the female character; whereas the angry scenario should lead to a lower number of correct angry drawings for girls and/or for the female character, compared to boys and/or the male character. Table 1 presents the number of correct drawings as a function of the character’s and child’s gender for sadness and anger. Table 1 suggests that the number of correct sad drawings does not differ between boys and girls or between the male and the female character. However, there seems to be differences between boys and girls in the depiction of anger, for the male character as well as for the female character.
Table 1

Number (and percentage) of correct drawings as a function of the character’s gender (male character/female character) and the child’s gender (boys/girls), for sadness and anger

 

Male character

Female character

Boys

Girls

p

Boys

Girls

p

N = 34

N = 48

N = 40

N = 50

Sadness

23 (68 %)

27 (56 %)

ns

25 (63 %)

34 (68 %)

ns

Anger

14 (41 %)

10 (21 %)

p < .05

14 (35 %)

9 (18 %)

p = .06

For each emotion, differences due to gender were examined using Chi-square tests (2 × 2). For sadness, there were no significant differences between the number of correct drawings produced by boys and girls, neither for the depiction of the male character, χ2 (2) = 1.09, p = .30, nor for the depiction of the female character, χ2 (2) = .30, p = .59. For anger, the number of correct drawings produced by boys differed from the number of correct drawings produced by girls. This difference was significant for the depiction of the male character, χ2 (2) = 3.98, p < .05, and marginal for the depiction of the female character, χ2 (2) = 3.38, p = .06. To sum up, in line with H1a, results indicate that boys produced a higher number of correct drawings than girls in response to the angry scenario. However, no significant differences appeared between boys and girls in response to the sad scenario.

Distribution of Incorrect Drawings

Then, we chose to focus on the 188 incorrect drawings, which did not depict the emotion evoked in the scenario. We wondered if in this case, mistakes were distributed by chance among the various emotions which could be depicted or if these mistakes targeted a particular emotion. According to H1b, incorrect drawings should be distributed so that the girls who did not depict anger tended to depict a stereotypically feminine emotion (e.g. sadness) or even no emotion at all (i.e., neutrality) and that boys who did not depict sadness tended to depict more a stereotypically masculine emotion (e.g. anger) or even no emotion at all (i.e., neutrality). Table 2 presents the numerical distribution of incorrect drawings. Table 2 shows that incorrect drawings are mainly distributed amongst the categories of neutrality, happiness, sadness and anger. More precisely, boys who did not depict the target emotion seem to have depicted anger instead of sadness and happiness or sadness instead of anger. Girls who did not depict the target emotion seem to have depicted happiness instead of sadness and sadness instead of anger.
Table 2

Number (and percentage) of incorrect drawings as a function of the emotion depicted and of the child’s gender (boys/girls)

Emotion depicteda

Emotion evoked in the scenario

Sadness

Anger

Boysb

Girlsc

p

Boysd

Girlse

p

Neutrality

6 (23 %)

9 (25 %)

ns

5 (11 %)

18 (23 %)

p = .07

Happiness

5 (19 %)

13 (36 %)

ns

12 (26 %)

14 (18 %)

ns

Sadness

 

14 (30 %)

37 (47 %)

p < .01

Anger

12 (46 %)

7 (19 %)

p < .05

 

Surprise

1 (4 %)

2 (5 %)

ns

3 (7 %)

3 (4 %)

ns

Fear

1 (4 %)

2 (5 %)

ns

2 (4 %)

3 (4 %)

ns

Disgust

1 (4 %)

2 (5 %)

ns

3 (7 %)

2 (2 %)

ns

Non expressivef

0 (0 %)

2 (5 %)

ns

7 (15 %)

2 (2 %)

ns

Total incorrect

26 (100 %)

37 (100 %)

46 (100 %)

79 (100 %)

aAccording to gender-emotion stereotypes, happiness, sadness, surprise and fear were considered as feminine emotions; whereas anger and disgust were considered as masculine emotions

bBoys who were told the sad scenario produced 65 % of correct drawings (48/74) and 35 % of incorrect drawings (26/74)

cGirls who were told the sad scenario produced 62 % of correct drawings (61/98) and 38 % of incorrect drawings (37/98)

dBoys who were told the angry scenario produced 38 % of correct drawings (28/74) and 62 % of incorrect drawings (46/74)

eGirls who were told the angry scenario produced 19 % of correct drawings (19/98) and 81 % of incorrect drawings (79/98)

fA drawing was considered as non expressive when judges did not agree about the emotion depicted

We first performed Pearson Chi-square tests of independence to determine whether the distributions presented in columns differed from distributions that could be expected by chance. The Chi-square tests indicate that these distributions do differ significantly from distributions that could be expected by chance, for boys and for girls, in response to the sad scenario: respectively, χ2 (6, N = 26) = 30.03, p < .01 and χ2 (6, N = 37) = 22.58, p < .01. The same result was revealed for boys and girls in response to the angry scenario: respectively, χ2 (6, N = 46) = 20.35, p < .01 and χ2 (6, N = 79) = 90.65, p < .01. Then we examined each of these distributions, in order to check whether the most frequently depicted emotions differed as a function of the child’s gender. To do so, Fisher exact probability tests were used. Among children who did not depict sadness in response to the corresponding scenario, it appears that a higher number of boys depicted anger (12/26, 46 %) compared to girls (7/37, 19 %) (p < .05). Among children who did not depict anger in response to the corresponding scenario, a higher number of girls depicted sadness (37/79, 47 %) compared to boys (14/46, 30 %) (p < .01) and a higher number of girls depicted neutrality (18/79, 23 %) compared to boys (5/46, 11 %), but this last difference was only marginal (p = .07). No other significant differences were found between boys and girls. To sum up, in line with H1b, results indicate that boys tended to depict anger more often than girls, in response to the sad scenario; whereas girls tended to depict sadness or even neutrality more often than boys, in response to the angry scenario.

Expressive Intensity of Correct Drawings

For further analysis, we focused on the 156 correct drawings, considered as expressing the target emotion in the first rating (109 sad drawings and 47 angry drawings, see Table 1 for the detailed number of correct drawings in each condition). According to H2, the correct sad drawings that were produced by boys and/or that were displaying a male character should be associated with lower scores of expressive intensity, compared to the correct sad drawings produced by girls and/or displaying a female character; whereas the correct angry drawings that were produced by girls and/or that were displaying a female character should be associated with lower scores of expressive intensity, compared to the correct angry drawings produced by boys and/or displaying a male character.

Table 3 shows the mean score of expressive intensity as a function of the child’s gender and the character’s gender for sadness and anger. Table 3 suggests that the mean scores of expressive intensity associated with sad drawings do not differ between boys and girls or between the male and the female character. However, there seems to be a difference for anger between the male character and the female character, for boys as well as for girls.
Table 3

Mean score of expressive intensity (and standard deviations) as a function of the child’s gender (boys/girls) and the character’s gender (male character/female character), for sadness and anger

 

Boys

Girls

All childrena

 

Male character

Female character

Male character

Female character

Male character

Female character

p

Sadness

1.78 (.15)

1.60 (.14)

1.64 (.15)

1.91 (.13)

1.69 (.10)

1.78 (.10)

ns

Anger

2.73 (.13)

1.31 (.13)

2.60 (.15)

1.11 (.16)

2.67 (.10)

1.21 (.11)

p < .01

Scores of expressive intensity ranged from 1 to 3 (1 = “low expressive intensity”; 2 = “medium expressive intensity”; 3 = “high expressive intensity”)

a“All children”, regardless of their gender

Data was subject to an analysis of variance with child’s gender (2) and character’s gender (2) as between-subject factors. For sadness, results did not indicate any significant main effect of the child’s gender, F (1, 105) = .11, MSE = .06, p = .74, η2p = .01, P = .12, or the character’s gender F (1, 105) = .33, MSE = .18, p = .57, η2p = .01, P = .13 on the mean scores of expressive intensity. And there was no significant interaction between these two factors, F (1, 105) = 2.48, MSE = 1.35, p = .12, η2p = .02, P = .20. However, for anger, results showed a significant main effect of the character’s gender, F (1, 43) = 102.00, MSE = 23.96, p < .01, η2p = .36, P = .99, indicating a significant difference between the mean scores of expressive intensity associated with drawings of a masculine face (M = 2.67) and of a feminine face (M = 1.21). There was no significant main effect of the child’s gender, F (1, 43) = 1.30, MSE = .31, p = .26, η2p = .03, P = .14, and no significant interaction between the child’s and the character’s gender, F (1, 43) = .10, MSE = .01, p = .82, η2p = .01, P = .08. To sum up, in line with H2, the expressive intensity of correct angry drawings was scored lower for children who completed the feminine face (Marie) than for children who completed the masculine face (Pierre), regardless of the child’s gender. However, the expressive intensity of correct sad drawings did not vary as a function of either the child’s gender, or the character’s gender.

Discussion

The present study was designed to examine the impact of gender-emotion stereotypes on French school-aged children’s depiction of sadness and anger in human faces. In accordance with our expectations, we observed an effect of the child’s and character’s gender on the depiction of anger.

First, our results are in line with expressiveness studies showing that gender-emotion stereotypes and display rules influence the way U.S. or French children express and feel gendered emotions such as anger (e.g., Hubbard 2001; Soussignan and Schaal 1996). Indeed, display rules tolerate or even exhort boys to express and feel anger; whereas they dictate girls not to express or feel anger. Consequently, boys produced a higher number of angry drawings than girls in our study. Instead of anger, girls tended to depict sadness or even neutrality more often than boys.

Interestingly, no effect of the child’s gender has ever been reported through attribution studies, consisting in asking children to judge other people’s emotional state. By underlining an effect of the child’s gender, our results suggest that, contrary to attribution tasks, a drawing task may make children more sensitive to their own gender. For that matter, U.S. self-report studies showing that girls report anger less often compared to boys also indicate that they justify their responses by mentioning their fear of adults’ disapproval (Fuchs and Thelen 1988; Meerum Terwogt and Olthof 1989). In line with these studies, Birnbaum and Croll (1984) also showed that parents reported greater acceptance of anger in boys than in girls. Thus, in our study of children, it could be almost as if depicting anger, even in the face of a distant character, would be equivalent to expressing anger themselves. Given that this emotion is socially less accepted from girls, they drew angry faces less often than did boys (whatever the character’s gender). Girls’ self-restraint with respect to anger would then be observed even through their drawings of emotional faces.

Nevertheless, our results are also in line with attribution studies, which show that gender-emotion stereotypes and display rules influence the way U.S. children judge the intensity of emotions expressed or felt by other people (e.g., Karbon et al. 1992). Indeed, our results indicate that the expressive intensity of correct angry drawings was scored lower for children who completed the face of the feminine character (Marie) than for children who completed the face of the masculine character (Pierre). This significant effect of the character’s gender was found for boys, as well as for girls. Thus, boys and girls were equally sensitive to the character’s gender, in that they attenuated the expressive intensity of anger, when completing the feminine face.

Therefore, the present study demonstrates the relevance of the use of a drawing task in order to study the impact of gender-emotion stereotypes in school-aged children. This drawing task is not based on a limited number of responses and does not therefore maximize the impact of gender, contrary to attribution tasks (e.g., Birnbaum and Chemelski 1984). In addition, this drawing task enabled us to test the impact of gender-emotion stereotypes in a non-verbal way, which prevented our results from being influenced by females’ superiority in terms of verbal abilities (Blanc 2010; Wester et al. 2002). Finally, the effect of the child’s gender and the effect of the character’s gender on children’s responses, as a consequence of gender-emotion stereotypes and display rules, are usually assessed and brought into light through distinct types of studies (expressiveness and attribution studies, respectively). Interestingly, our study captured an effect of the child’s gender as well as an effect of the character’s gender on children’s responses relative to anger: children followed gender-emotion stereotypes and display rules in their drawings, but they applied these rules either to themselves or to the character. Hence, the drawing task we used in the present study seems to have a particular status, half way between attribution studies and expressiveness studies and therefore contributes to our knowledge of gender issues.

However, we did not notice any effect of the child’s or character’s gender on depiction of sadness, in response to the corresponding scenario. The only observation in conformity with gender-emotion stereotypes we made was through the analysis of incorrect drawings: boys who did not depict sadness tended more than girls to depict anger instead. According to the Parallel-Constraint-Satisfaction Theory of Kunda and Thagard (1996), participants are more likely to use gender stereotypes when the character is in an ambiguous context (e.g. a situation that could elicit sadness or anger) than in an unambiguous context. Parmley and Cunningham (2007) tested this theory by proposing an attribution task to 5-year-old U.S. children. Their results indicated that, in the case of sadness, the use of gender-emotion stereotypes was observed only when the situation was ambiguous. Yet, for anger, the use of gender-emotion stereotypes appeared whatever the situation was, being stronger in an ambiguous situation. In the present study, the scenarios unambiguously elicited emotions of sadness and anger. Our results are in conformity with the ones of Parmley and Cunningham’s (2007), explaining the fact that we did not observe any effect of the gender for sadness. Therefore, we can imagine that the use of ambiguous scenarios could have driven us to observe an effect of the gender for sadness and could have reinforced the effect of gender we observed for anger.

Although we validated many of our hypotheses, there were some limitations in the present study. First, even if the scenarios we presented to children were used and validated in past research, only one scenario was used for each emotion in the present study. Future research might use multiple scenarios for each emotion, so as to ensure that children’s responses are not only linked to the scenario presented to them. Secondly, we expected boys’ number of correct angry drawings to be higher than what was observed. And we did not expect children (boys and girls) to depict happiness in response to the sad or to the angry scenario. These observations could be explained in reference to grapho-motor difficulties that can affect children’s ability to depict emotions. Indeed, children’s grapho-motor abilities are subjected to a strong inter individual variability. Consequently, some children in our study may not have been able yet to depict the emotions evoked by the scenarios. This variability could then explain the fact that some boys did not depict anger and that some children (boys and girls) depicted happiness instead of sadness or anger, happiness being the easiest emotion to depict and corresponding to children’s graphic routine when depicting a human figure. Let us remember anyway that boys significantly depicted anger more often than girls and that we found no significant difference between the number of boys and girls who depicted happiness in our study. However, it would have been interesting to ask children to label the emotion felt by the character after the drawing task, so as to ensure that the emotion they depicted was the emotion they intended to depict.

Until now, the present study is the only one having tested the potential effect of the child’s and character’s gender on the graphic depiction of emotion in drawings. These two factors are usually never studied or even considered in studies about children’s ability to depict emotions in a human figure drawing (Brechet et al. 2009, 2007; Cox 2005; Golomb 1992; Missaghi-Lakshman and Whissell 1991; Sayil 2001). Our results underline a real impact of the child’s and character’s gender on graphic depiction of anger, according to gender-emotion stereotypes and display rules. Therefore, we strongly recommend, for further studies, that these two factors be taken into account. Furthermore, these data suggest that clinicians should exercise caution when interpreting the meaning of children’s expressive drawings. The evaluations using projective drawing are amongst the ten types of evaluations, which are the most commonly used with children in many countries: US (Camara et al. 2000; Watkins et al. 1995), UK (Bekhit et al. 2005), France (Picard and Baldy 2011), Hong Kong (Chan and Lee 1995), Australia (Sharpley and Pain 1988). The use of projective interpretations of drawings consists in viewing the drawing’s content as a reflection of the child’s emotional state, personality, or previous experiences. From our study, it seems crucial to take into account the fact that boys tend to draw angry characters more often than girls, and that children (i.e., boys as well as girls) when drawing an angry character tend to depict this emotion more intensely on a male character than on a female character. As France has been proved to be a less “masculine” country than the US and the UK for instance (Hofstede 2003), this impact of gender-emotion stereotypes on French children’s drawings is likely to be found with U.S. or British samples as well (maybe even to a larger extent). This assumption could be tested by replicating the present study with children from other countries. Until then, we think that readers from other countries should be warned that such an influence can be found, as the same kind of projective drawing tasks are used in many countries, for instance with the Machover’s Draw-a-Person Test (1949), The Koppitz Draw-A-Person (1968) test, or the Buck’s House-Tree-Person Technique (1948).

Copyright information

© Springer Science+Business Media New York 2012

Authors and Affiliations

  1. 1.Department of Psychology, Laboratory Epsylon, E.A. 4556-Dynamics of Human Abilities & Health BehaviorsUniversity of Montpellier IIIMontpellierFrance
  2. 2.Department of PsychologyUniversity of Montpellier IIIMontpellierFrance

Personalised recommendations