Introduction

Although societal roles for women and men have changed in the United States since the second wave of the women’s movement (Twenge 2009), gender stereotypes are still commonly portrayed in U.S. media (Collins 2011; Greenwood and Lippman 2010). A likely reason for the persistence of stereotypes is that the basic structure of society remains patriarchal, with a gendered division of labor and pressures to conform to stereotyped roles (Rudman and Glick 2008). Further, it has been argued that there is a backlash against women’s accomplishments in the workplace, resulting in increased cultural emphasis on gender stereotypes that support patriarchy (Douglas 2010). In the present study, we conducted a content analysis of products in U.S. children’s popular culture that portray male and female characters—including dolls and action figures, Halloween costumes, and Valentine cards—to determine the extent to which the characters depicted had gender-stereotyped characteristics reflecting the power differences between women and men in society.

Although there has been much study of media representations especially television and video games, other cultural products, especially those aimed at children (Rudy et al. 2011), have attracted less attention. The products chosen for our study contain male and female counterparts that we compared to see if there was a simultaneous representation of masculine-stereotyped characteristics for male characters and feminine-stereotyped characteristics for female characters. The products we examined offer children an opportunity to strongly identify with the character represented because they choose a specific character from an assortment available. For example, children interactively play with dolls and action figures, increasing identification (Elias and Berk 2002), and when they wear a Halloween costume, they might “try on” the identity associated with the depicted character.

We synthesized our coding systems from several other studies, creating a comprehensive scheme for measuring traditionally masculine and feminine characteristics relevant to understanding the power differences between women and men in society. We argue that if cues associated with dominance are prescribed for boys but proscribed for girls, and if cues associated with submissiveness are prescribed for girls, but proscribed for boys, a gendered power imbalance is evident. The research cited in our paper is from the United States unless otherwise noted.

Gender Stereotypes and Power

Gender stereotyped traits associated with women encompass the domains of communality and expressiveness, whereas those associated with men represent agency and industriousness. Social role theorists argue that gender-stereotyped traits developed from gender-segregated roles, including women’s care of children and men’s work and protector roles that derived in part from the development of agriculture and industry (Eagly et al. 2000; Wood and Eagly 2013). There are both prescriptive (what people should do) and proscriptive (what people should not do) stereotypes for each gender. Traits strongly encouraged for men include career-orientation, leadership, aggression, assertiveness, and independence, and many of these prescribed traits are associated with higher status persons (Rudman et al. 2012). Prescriptions for women include being emotional, warm, interested in children, sensitive to others, a good listener, friendly, and attentive to appearance. Unlike men’s prescribed traits, women’s are generally status-neutral. Proscribed traits for men include being emotional, naïve, and weak, traits which are associated with low status. Some proscriptions for women include being aggressive, intimidating, dominating, and angry.

Whereas some gender-stereotyped traits (e.g., male industriousness and female nurturance) likely derived from the societal division of labor, others likely arose from the fact that unlike other social groups that differ in status, women and men often form intimate relationships with one another. More specifically, “heterosexual interdependence” coexists with male dominance (Rudman and Glick 2008). It is proposed that women’s sexuality is important in this regard in that it is valued by heterosexual men; thus, it is a potential source of women’s power. Perhaps because of its potential influence, women’s sexuality is given much attention in society, and attempts are made to control it (Valenti 2010). One means of control is to associate sexualized depictions of women with cues of submission or even violence (Stankiewicz and Rosselli 2008). This can occur when women are shown wearing revealing clothing when men are not and when women’s bodies are posed to imply sexual availability (e.g., laying down) whereas men’s bodies are not. Women are frequently depicted as objects of heterosexual men’s desire (Stankiewicz and Rosselli 2008), and sexualized depictions have been found to be increasing for representations of both women (Hatton and Trautner 2011; Reichert and Carpenter 2004) and girls (Graff et al. 2013). The ubiquitous representation of women as sexually submissive likely lowers the status of women (Hatton and Trautner 2011; Jeffrys 2005; Murnen and Smolak 2012). Women portrayed in a sexualized way are seen as less competent than when they are portrayed in athletic or professional clothing (Glick et al. 2005; Gurung and Chrouser 2007). This has also been found true of sexualized portrayals of girls (Graff et al. 2012).

Some stereotypical masculine and feminine roles and traits are positive and helpful to society. It is beneficial to civilization for some people to have caretaking skills and to be industrious, for example. However, if industriousness is prescribed for men but proscribed for women and caretaking is prescribed for women and proscribed for men, patriarchy is not threatened. Further, some individuals adopt exaggerated adherence to extreme gender-role norms in the form of hyper-masculinity and hyper-femininity which help perpetuate the status quo. Hyper-masculine men believe that “violence is manly” and “danger is exciting,” and they often have sexually calloused attitudes towards women (Zaitchik and Mosher 1993). Hyper-masculinity (and a related construct hostile masculinity, Malamuth and Thornhill 1994) are the male gender-role prescriptions most closely associated with self-report of sexual aggression against women (Murnen et al. 2002). Among women, hyper-femininity is associated with “accepting” a sex-object role for women, including the idea that men are dominant (Murnen 1998). A decade of research on objectification theory (Fredrickson and Roberts 1997) has shown that internalizing sexual objectification and treating one’s own body as an object is associated with the development of eating disorders, depression, and sexual dysfunction (Calogero et al. 2011; Moradi and Huang 2008; Tiggemann 2011).

Research on Children

To the extent that there are exaggerated gender-stereotyped traits promoted by the culture, they could potentially affect the gender-role development of children. Children develop stereotypes about gender from an early age (Martin and Ruble 2004). By the age of 2 most children label themselves as either female or male and are aware of gender distinctions. Cognitive social learning theorists (Bussey and Bandura 1999, 2004) maintain that children then start using gender as a categorizing framework and develop stereotyped ideas about characteristics associated with females and males if such stereotypes are available in the culture. Children then might use these stereotypes to guide their own behavior, especially if the stereotyped behavior is portrayed in a positive way and is associated with rewards. Children can learn gender-stereotyped behavior through observing models, by having their own gendered behavior rewarded, and by receiving direct instruction or approval from others in their environment (Bussey and Bandura 1999). Through cultivation processes, gender-stereotyped behaviors can come to be seen as normative if the themes are repeated frequently (Gerbner et al. 2002).

Characters in children’s television have been the subject of many studies. In a study of characters in children’s favorite television programs, Aubrey and Harrison (2004) found that there were more male lead characters than female lead characters as expected, and there was some stereotyped behavior in that male characters were more likely to order others around and demonstrate ingenuity whereas female characters were more attractive and frail. Hoffner (1996) had children ages 7–12 list their favorite television character and rate them on a number of attributes. Almost all the boys listed a male character as their favorite whereas just over half the girls listed a female character. The extent to which either boys or girls identified with the male character was predicted by how intelligent they rated him, but the extent to which girls identified with a female character was solely dependent on her rated attractiveness. Thus these role models were admired for conforming to gender stereotypes.

There are few comprehensive studies of children’s toys, but some researchers have argued that they are very gender stereotyped (Cherney and London 2006). Video games have been the focus of some detailed analysis. In video game magazines, female characters are frequently shown in a sexualized manner (e.g., wearing revealing clothing and posing in a sexually suggestive manner); male characters, in a hyper-masculine manner (e.g., holding weapons and having a hostile facial expression) (Dill and Thill 2007). In the video games themselves, Miller and Summers (2007) found that female characters were more frequently portrayed as attractive, sexy, helpless, and innocent; male characters, as muscular and powerful.

There has been some study of the effects of playing with toys, showing implications for children’s gender-role development (Cherney and Dempsey 2010). For example, playing with stereotypically masculine toys that involve physical manipulation is associated with the development of spatial skills (Blakemore and Centers 2005). There has been concern that fashion dolls like Barbie are damaging to girls’ body esteem in that Barbie is unrealistically thin and sexualized. One study found that playing with Barbie was associated with lower body esteem (Dittmar et al. 2006), although another did not (Anschutz and Engels 2010). Sherman and Zurbriggen (2014) found that playing with Barbie compared to physically neutral Mrs. Potato Head led girls ages 4–7 to imagine a lower number of careers as possible for themselves, but this play had no effect on boys.

In a rare longitudinal study of the effects of exposure to gender-stereotyped models on gender-typed behavior, it was found that boys and girls who were exposed to superheroes on television were more likely to subsequently engage in weapon play (Coyne et al. 2014). Effects were stronger for boys in that they were more likely to engage in a variety of types of male-stereotyped play after exposure. Previous research had found that superheroes are far more likely to be male than female and that they are very likely to exhibit masculine-stereotyped behavior (Baker and Raney 2007).

Present Study

In the present study, we examined stereotypes in three different cultural products that have not been examined in a thorough manner and that offer characters with whom children may identify: Halloween costumes, dolls and action figures, and Valentine’s cards. We chose these particular products because they have both female and male characters, allowing for the examination of the portrayal of both feminine- and masculine-stereotyped characteristics in each character. We were interested in determining whether there was a simultaneous representation of feminine-stereotyped characteristics among female characters corresponding to the representation of masculine-stereotyped characteristics among male characters. We tried to choose products based on popularity so that these would be products children actually used. We expected our characters to be stereotyped to a fair degree, consistent with studies of television and video game characters. Television is less gender-stereotyped than video games, and we expected our products to be somewhere between these two domains in terms of level of stereotyping.

One previous study of Halloween costumes (Nelson 2000) found that male costumes were more likely to represent a “warrior” theme and portray villains than female costumes and that female costumes were often based on characters valued for appearance, such as princesses and beauty queens. In our study we examined Halloween costumes with a more comprehensive coding system, and we applied the same system to the other two types of products as well.

In measuring stereotypically feminine characteristics, we distinguished between traditional femininity and hyper-femininity. Characteristics associated with the traditional feminine role involve communion, expressiveness, and a focus on appearance, but did not directly represent sexualization and subordination. An example of such a characteristic is a friendly facial expression. Other characteristics associated with sexualization and subordination were indicative of hyper-femininity. An example of a hyper-feminine cue is when a character is portrayed in revealing clothing.

We also distinguished two types of stereotypically masculine characteristics. Traditional masculine role characteristics involve instrumentality and action, but without explicit violence. An example of a traditional masculine characteristic was portraying the body-in-motion. Hyper-masculine characteristics were those associated with violence such as having armor present or being posed with hands in fists. We expected that traditional masculine characteristics would be more common than hyper-masculine ones among male characters because traditional characteristics are more socially acceptable (Rudman et al. 2012). We thought that hyper-masculine characteristics would be more proscribed for girls than characteristics associated with the merely dominant aspects of masculinity. For example, portraying girls and women engaging in physical activity would be more likely to occur than portraying girls and women with weapons.

Thus our hypotheses focused on the relationship between the gender of portrayed characters and the gender-typing (traditional or hyper) of cues. For gender and feminine-typed cues, we predicted that characteristics representing traditional femininity (communality, expressiveness, appearance focus) would occur with greater frequency for female characters than male characters (Hypothesis 1a). We also hypothesized that feminine-stereotyped characteristics explicitly associated with sexualization and subordination indicative of hyper-femininity would occur more often for female characters than male characters (Hypothesis 1b). Turning to gender and masculine-stereotyped cues, we expected that masculine-stereotyped characteristics associated with traditional masculinity (instrumentality and agency) would occur with greater frequency for male characters than female characters (Hypothesis 2a). We also hypothesized that hyper-masculine characteristics associated with violence would occur with greater frequency for male characters than female characters (Hypothesis 2b). Although there were no specific hypotheses associated with product type, possible differences in gender stereotyped characteristics between product types were examined as well.

Method

Sample of Coded Products

Culture products were examined for which there were comparable items for girls and boys: (a) popular Halloween costumes advertised online (90 female costumes and 90 male costumes), (b) popular dolls (n = 79) and popular action figures (n = 71) advertised online, and (c) Valentines found at two national stores (n = 54 female characters, n = 59 male characters).

Halloween costumes were obtained from three Halloween-specific online websites that were most popular in a Google web search in 2013: Costume Kingdom, Party City, and Spirit Halloween, as well as from three websites of popular national retailers: Kmart, Target, and Walmart. The best-selling 15 costumes for girls and the best-selling 15 costumes for boys on each website were examined. (Walmart was the only site that did not arrange costumes by gender of the child so the best-selling in total were examined and then separated by gender of the child wearing the costume). Pictures of each costume were downloaded to a PowerPoint file for coding approximately 2 weeks before Halloween. There were a few repeat costumes, but if they were not displayed in the same way, they were coded as separate observations (similarly to how we coded Valentines).

We examined dolls and action figures that emerged as most popular from the national retail toy stores of Target, Toys R Us, and Walmart in February 2013. The top 25 from each website were used, searching for ages older than preschool. Pictures were downloaded into a file for coding. Different toys were popular across the three stores, but there was some overlap so that if the toy was depicted in the exact same way, it was only included one time in our sample, and it was randomly assigned to one of the stores from which it derived. In the action figures for boys, there were a few female characters and they were not coded in the group of female characters because they were not necessarily intended for girls.

Two weeks before Valentine’s Day of 2013, we purchased all the different types of boxes of Valentine’s cards available at Walmart. Only those that had obviously female or male characters were included in the coding. For example, Minnie Mouse cards were included because Minnie is clearly female, but characters from Cars were not because they were gender-indeterminate. Some cards had no human-related gender cues such as cards with guitars on them, so these were not coded. The cards coded in the analysis that contained primarily male characters were Avengers, Batman, Justice League, Power Rangers, Sponge Bob, Star Wars, and Toy Story; the cards that contained primarily female characters were Cinderella, Disney Princess, Dora, Minnie Mouse, and Monster High. In the male-oriented cards there were a few female characters, and they were not coded in the main analysis. In the female-oriented cards there was a male monkey (Dora’s sidekick Boots) that was not coded in the analysis. Each box of cards had 6–10 different cards to code. Additional cards were purchased in 2014 from the grocery store, Kroger, to increase the sample size to be comparable to those of the costumes and the dolls/action figures. Cards that were not represented in the 2013 sample were added, including Teenage Mutant Ninja Turtles with primarily male characters as well as Barbie, Disney Fairies, and The Little Mermaid with primarily female characters. Because a set of cards might have multiple representations of one character, all of the representations were coded because they might present the character in different ways and thus offer different cues to code.

Coding Strategy

Table 1 contains a listing of all of the cues coded that is organized by whether the cues represented the category of traditional femininity, hyper-femininity, traditional masculinity, or hyper-masculinity. We derived the cues from previous research, some of which used the pioneering work of Goffman (1979) who examined gender inequality in advertisements. Goffman proposed that male dominance and female subordination were evident in advertisements through the depiction of differential roles, body positioning, and facial expression. There have been adaptations of Goffman’s codes, and our system was partially based on one used by Mager and Helgeson (2011). We added some additional cues related to hyper-femininity from content analyses that have focused more precisely on the sexual objectification of women, which has been the focus of many recent content analyses (e.g., Collins 2011). We added some cues related to hyper-masculinity from a recent analysis of depictions of men in men’s magazines (Vokey et al. 2013). For each coding category for each character representation, we noted the presence of the cue (coded as 1) or the absence of the cue (coded as 0).

Table 1 Inter-rater reliability of characteristics coded by gender-stereotyped category of characteristic
Full size table

Femininity

Traditional femininity included the two characteristics of friendly facial expression and decorative clothing that are associated with traditionally feminine roles of communality, expressiveness, and appearance-focus, but are not explicitly associated with sexual subordination. Decorative clothing was often represented by wearing a dress for girls and women which does not encourage active motion, but the cue also occurred for male characters in the form of clothing that was embellished in a way that did not add to the clothing’s function such as a shiny gold belt on a male character. Mager and Helgeson (2011) adaptation of Goffman’s (1979) codes involved examining the decorative role associated with femininity versus the functional role associated with masculinity, which is consistent with our coding scheme.

Hyper-femininity is explicitly associated with subordination and/or sexualization. Body positions of “rituals of subordination” (including obvious knee bends, head cants, and curved spine; see Rudman and Verdi 1993) or facial expressions of “drifting” away from the scene (Mager and Helgeson 2011) were included in this categorization. (The Goffman 1979, cue of “laying down” did not occur for any characters so was not included in the coding system.) Sexual objectification included the presence of revealing clothing that provocatively exposed breasts, the midriff, or legs (Downs and Smith 2010; Goodin et al. 2011). Sexually appealing facial expressions (Stankiewicz and Rosselli 2008) were distinguished from simply friendly facial expressions (a traditionally feminine characteristic). A friendly facial expression is one where the mouth does not suggest sexual activity in that the person had closed lips that are smiling (but not in a sultry way) or an open-mouthed smile (Hatton and Trautner 2011).

Masculinity

For masculine-stereotyped characteristics, we distinguished between those associated with the traditional role representing agency and industriousness and those associated explicitly with hyper-masculinity involving violence. We relied on the work of Mager and Helgeson (2011) for operationalizing traditional masculinity, which was represented by functional clothing (something that allows the body to perform an action such as a uniform) and purposeful movement. The facial expressions of being portrayed as stoic or having an unreadable facial expression (perhaps due to wearing a mask) are consistent with the proscription against emotional expressiveness associated with traditional masculinity (Rudman et al. 2012). The cues of weapons present and angry emotional expression were coded as hyper-masculine (adapted from Vokey et al. 2013) in that they were suggestive of violence. Cues that we added that are consistent with hyper-masculinity included the presence of armor and having hands in fists. These additions were not coded in Vokey et al.’s (2013) analysis, but they are consistent with Dill and Thill’s (2007) scheme, which coded for weapon use, attacking postures, hostile facial expressions, armor, and posing with weapons.

Reliability of Coding Categories

The first three authors served as the main coders. The coders first worked together to agree on a coding strategy and coded a subset of products to establish consistency in coding. The coders then divided up the remaining items to code. Each coder coded about one-third of the products, ensuring that they coded some with male characters and some with female characters. To check reliability, a fourth coder coded 20 % of the characters (randomly selected). Kappa coefficients were computed to measure inter-rater agreement for each coding category. These values are displayed in Table 1 and indicate acceptable reliability (Neuendorf 2011). When there was disagreement between the two coders, a third coder arbitrated.

When a product was coded, neither the entire group of feminine stereotyped cues nor the group of masculine stereotyped cues were ordered together. Instead, cues were listed to encourage ease of coding and discourage the development of a gender-stereotyped response set by listing all of the cues associated with body position together, as well as those associated with facial expression, etc. Each coder indicated the presence of each cue for each character. Most cues coded were not mutually exclusive categories except those associated with facial expression.

Validity of Coding Categories

A study was conducted to try to obtain information concerning the validity of the coding system. Six female character depictions and six male character depictions were put into a PowerPoint file in random order, and college student participants (N = 139, 71 % women) were shown the PowerPoint and asked to rate each character on 7-point scales in terms of the extent to which they appeared powerful, submissive, friendly, dominant, sexy, and aggressive, among a few other characteristics not relevant to the present study, from 1 (not at all) to 7 (very). Two female characters were chosen to represent the traditional woman stereotype (i.e., friendly facial expression but few sexualization cues; Deaux et al. 1985), two to represent hyper-femininity or the sex object stereotype (i.e., revealing clothing and suggestive body positioning), and two to represent the nontraditional stereotype (i.e., some masculine stereotyped cues present such as physical movement and few feminine stereotyped cues). For the male characters, two were chosen to be traditionally masculine but not explicitly hyper-masculine (i.e., portrayed with a uniform but not with weapons), two were chosen to be hyper-masculine (i.e.,., weapons present), and two to be nontraditional (i.e., smiling in a friendly way and few masculine stereotyped cues). Table 2 lists the specific products tested within each category and presents mean ratings of each product across the dependent measures. The gender of participants was initially included in the analyses, but there was no significant multivariate effects of gender, or interactions of gender with character rated, so we collapsed across participants’ gender in the ratings.

Table 2 Ratings of characters for validity study
Full size table

The ratings made were consistent with the a priori categorization of the character in most cases. For female characters, the traditionally feminine characters were rated the most submissive and the nontraditional, F (11,127) = 37.65, p < .001 ηp 2 = .79. The traditional female characters were rated the friendliest of the female characters, F (11, 128) = 197.20, p < .001, ηp 2 = .95. As expected, the sex-object female characters were rated the most sexy of the female characters, F (11,127) = 36.96, p < .001, ηp 2 = .77. The male hyper-masculine characters (and, unexpectedly, Batman) were rated as the most powerful, F (11,128) = 233.66, p < .001, ηp 2 = .95; dominant, F (11,120) = 136.75, p < .001, ηp 2 = .92; and aggressive, F (11,127) = 331.56, p < .001, ηp 2 = .97 of all of the characters.

Results

Chi-square tests of association were conducted to determine the link between the gender of the character and the presence of gender-stereotyped cues (traditional and hyper-femininity; traditional and hyper-masculinity). As indicated previously, for each coding category within each character representation, we noted the presence of the cue (coded as 1) or the absence of the cue (coded as 0). The chi-square test then compared the presence versus absence of each cue with the gender of the character such that a significant value indicated that the cue was represented more often in one gender compared to the other.

Gender and Feminine-Stereotyped Cues

As predicted, there was a statistically significant gender difference in the frequency of each of the traditionally feminine (Hypothesis 1a) and hyper-feminine (Hypothesis 1b) cues, which were all more commonly associated with female than male characters (see Table 3). The traditional feminine cue that occurred most often for girls and women was decorative clothing, associated with 88 % of the female characters but only 13 % of the male characters. The traditional feminine cue of friendly facial expression occurred for more than half the female characters. In terms of hyper-feminine cues, the most common cue for female characters was revealing clothing, which occurred for more than half the female characters but less than 20 % of male characters. Less frequent for female characters were some of the Goffman (1979) codes that symbolized female sexual subordination including curved spine, bent knee, and head cant; however, all of these cues occurred significantly more often for female than male characters.

Table 3 Number and percentage of feminine-stereotyped characteristics by gender of character
Full size table

There were a total of eight feminine stereotyped characteristics. The presence of each cue was added across the eight cues for each product (i.e., character representation) creating a total score that could range between 0 and 8. The total number of feminine-stereotyped cues for the female characters (M = 3.56, SD = 1.5) was significantly greater than it was for male characters (M = .43, SD = .65), t (302.21) = 28.57, p < .001, ηp 2 = .73 (using separate variance estimates due to violating the homogeneity of variance assumption of the t-test).

Gender and Masculine-Stereotyped Cues

As hypothesized, there were statistically significant gender differences in the portrayal of both traditional (Hypothesis 2a) and hyper (Hypothesis 2b) masculine-stereotyped cues, which were more commonly associated with male characters than female characters (Table 4). The most frequent cue associated with traditional masculinity was functional clothing, which was associated with 78 % of male characters but only 14 % of female characters. The next most common traditional masculinity characteristic was the body-in-motion. Thus these cues associated with traditional masculinity, but not explicitly hyper-masculinity, occurred with great frequency for male characters. Although cues indicative of hyper-masculinity were less common than those for traditional masculinity, hyper-masculine cues were again more frequent for male characters than for female. The most common hyper-masculine characteristic involved having hands in fists (associated with 51 % of characters) and then wearing armor (associated with 41 % of characters).

Table 4 Number and percentage of masculine-stereotyped characteristics by gender of character
Full size table

There were a total of eight masculine stereotyped cues, and a sum of these was created for each product (i.e., character representation), which could range from 0 to 8. The total of masculine stereotyped cues varied by gender of character, such that it was lower for female characters (M = .53, SD = 1.04) compared to male characters (M = 3.43, SD = 1.58), t (377.2) = 22.73, p < .001, ηp 2 = .58 (using a t-test with separate variance estimates due to violating the homogeneity of variance assumption).

Differences Among Product Types

Although there were no hypotheses associated with product type, it was of interest to determine if the various cues were presented differently by product type: costume, doll/action figure, and Valentine card. For each gender separately, chi-square tests of association were conducted to examine whether the frequency of each characteristic varied by product type. Again, only those tests that were significant at p < .001 were considered statistically reliable, controlling for Type I error.

For the depictions of female characters, there were significant differences in product types in the representation of head cant, χ 2 (2) = 17.81, p < .001, and appealing facial expression, χ 2 (2) = 97.59, p < .001 (see Table 3). Valentine cards had the highest frequency of both head cant and appealing facial expression compared to costumes and dolls. For the male characters, there were significant differences between product types in terms of an unknown facial expression, χ 2 (2) = 14.51, p = .001, and a stoic facial expression, χ 2 (2) = 15.34, p < .001 (see Table 4). Costumes had the highest portrayal of unknown facial expression (most likely due to the presence of masks), and both action figures and Valentines had higher proportions of a stoic facial expression compared to costumes.

Discussion

As expected, we found that gender-stereotyped characteristics were frequently portrayed in children’s popular culture products. Across popular Halloween costumes, dolls and action figures, and Valentines, female characters were more likely than male characters to be depicted with feminine-stereotyped characteristics, especially with the traditional feminine characteristics of decorative clothing and a friendly facial expression. In contrast, male characters were more likely to be portrayed with masculine-stereotyped characteristics, especially the traditional masculine characteristics of functional clothing and the body-in-motion. A basic distinction between decorative clothing versus functional clothing strongly differentiated characters by gender such that 88 % of female characters were portrayed with decorative clothing; 78 % of male characters, with functional clothing.

More specifically, Hypotheses 1a and 1b, which compared the presence of feminine-stereotyped characteristics in female characters versus male characters, were supported. Traditional feminine-stereotyped characteristics not explicitly associated with sexualization were very common for female characters (greater than 50 %) but uncommon for male characters (less than 25 %). Sexualized hyper-feminine characteristics were less common for female characters, but still occurred more often for females than males. The most common hyper-feminine cue of revealing clothing occurred for slightly more than half the female characters but only for 7 % of male characters. Other sexualizing cues were less common, but almost all occurred for at least one-quarter of female characters.

Some feminine-stereotyped characteristics are not problematic, such as friendliness which promotes positive social relations. However, a sexualized portrayal of female characters, likely disempowering, was not uncommon. Revealing clothing was associated with about half the female characters; an openly sexualized facial expression, with about one-quarter. Along with decorative clothing and subordinating body positions, the sex-object stereotype is clearly associated with many female characters, resulting in a fairly frequent overall portrayal of hyper-femininity (Murnen 1998). This pattern is consistent with other research concerning the increased sexualization of girls’ culture (American Psychological Association 2010).

On the other hand, there was a feminine-stereotyped character (Minnie Mouse) who did not have many sexually submissive characteristics but rather who had a friendly facial expression and decorative clothing. Thus girls do not necessarily have to choose a sex-object character among those available in popular culture. However, Starr and Ferguson (2012) found that girls ages 6–9 were more likely to choose a sexualized paper doll for their “ideal self” than one that was not sexualized, and they also thought the sexualized doll would be a more popular choice for other girls. Girls might be attracted to sexualized products that make them seem older, and they might not realize that sexualization can be associated with negative perceptions (Murnen and Smolak 2013; Smolak and Murnen 2011).

Support was also found for Hypotheses 2a and 2b concerning the presence of masculine-stereotyped characteristics. Among the male characters, characteristics associated with traditional masculinity, such as functional clothing and the body-in-motion, were common. Again, these characteristics are not necessarily negative in that they are associated with industriousness. Some other characteristics are more hyper-masculine in nature: a weapon was present in half the male depictions, armor in 41 %, hands in fists in 36 %, and an angry emotional expression in 21 %. Thus the idea that anger and aggression might be used to solve problems is implied by a fair number of male characters that we coded.

Boys could choose between an industrious, active character consistent with traditional masculinity or a more hyper-masculine one. It would be interesting to know if the choice of characters varies by the socio-economic status of boys. Vokey et al. (2013) found that magazines directed at men of lower socio-economic status (SES) had more hyper-masculine cues than magazines aimed at higher SES men. Those who lack a dominant position in society might be attracted to products that promise power through physical means.

There were few female characters who had traditional masculine characteristics. The most frequent masculine cue associated with female characters was to portray the body-in-motion, which was associated with fewer than one-quarter of female characters. Because our coding involved examining static products (i.e., downloaded images or Valentine cards), it might have seemed difficult to portray motion, but it still was associated with 61 % of male characters. Physical control of the body is important for empowerment. The heavy focus on appearance versus action for female characters mimics societal expectations that girls and women should be objects rather than actors. Sexual objectification in society encourages girls to engage in self-objectification which is associated with the development of depression and eating disorders (Calogero et al. 2011; Moradi and Huang 2008; Tiggemann 2011).

Looking across the group of characteristics and products, there was much evidence of gender-stereotyping. Additionally, there were a few significant differences between product types in the representation of characteristics. For the representations of female characters, Valentines were the most likely to portray a head cant and an appealing facial expression. This is most likely because Valentine characters sometimes represented adult women such as Disney princesses who are sexually objectified characters. Even though some of the costumes represented adult characters, it was girls who were shown in the costumes and a too-sexualized portrayal might not be socially acceptable. For the male characters, boys in costume were the most likely to be depicted with an unknown expression (probably because they were often wearing masks), and both action figures and Valentines had the highest proportions of showing a stoic facial expression.

In sum, stereotypic feminine characteristics were common for female characters but not male characters; and conversely, stereotypic masculine characteristics were common for male characters but not female characters. This stereotypic pattern likely reflects and reinforces unequal societal roles for women and men. Further, some of the cues associated with femininity emphasize sexual objectification, reinforcing hyper-femininity (Murnen 1998); and some of the cues associated with masculinity emphasize the use of aggression, reinforcing hyper-masculinity (Zaitchik and Mosher 1993). Hyper-femininity and hyper-masculinity are exaggerations of gender-role beliefs that are associated with negative outcomes for individuals and perpetuate patriarchy.

Comparing the characters in the present study to those that Dill and Thill (2007) analyzed from video-game magazines, there is a comparable amount of stereotypes portrayed. Dill and Thill (2007) found that almost 40 % of their female characters were “scantily clad”; 53 % of the female characters in the present study wore revealing clothing. Similarly, 42 % of Dill and Thill’s (2007) male characters were posed with weapons, and that was true of 51 % of male characters in the present study.

Practice Implications

It is likely that using such gender-stereotyped products could increase stereotyped behavior, explained by both cultivation theory (Gerbner et al. 2002) and cognitive social learning theory (Bussey and Bandura 1999, 2004). If a girl dresses like a princess versus a zombie for Halloween, she will likely engage in different behaviors to play out the role associated with her costume. Similarly, a boy wearing armor and carrying a weapon is going to be more likely to engage in active and perhaps aggressive play than a boy dressed as a friendly dinosaur. If a girl chooses a Dora doll to play with rather than a Monster High doll, she is associating herself more with action and adventure versus appearance and fashion. Similarly, a boy who gives out Avengers Valentines is likely choosing to communicate something different about himself than the boy (or girl) who chooses SpongeBob.

More research is needed to determine the extent to which these gender-stereotyped models influence behavior. Social cognitive theory suggests that models are important, along with children trying out behavior and observing the consequences, and children receiving direct instruction about gender-related behavior from parents and peers (Bussey and Bandura 1999). If we wanted to encourage boys to adopt socially desirable feminine-stereotyped characteristics from cultural models, and discourage hyper-masculine characteristics, the models should be male because boys will likely choose male models (Hoffner 1996). If more male characters in boys’ culture had a smile on their face rather than an expression of anger that would be helpful. Some girls appear to identify with male characters, and some with female, and it would be interesting to know more about that choice. Girls who watched action figures in cartoons were subsequently more likely to play with a weapon, but not to engage in other masculine-stereotyped behavior (like the boys did), so behavior is not just a matter of modeling (Coyne et al. 2014). Parents could play a role in trying to encourage socially desirable masculine- and feminine-stereotyped behavior in their children regardless of gender, perhaps by encouraging play with a variety of types of toys. Parents likely have to be fairly explicit if they are encouraging counter-stereotypic toy play, though, because Freeman (2007) found that most children thought their parents would only approve of gender-stereotyped toy choices (which was not true, according to the parents’ self-report).

Limitations and Future Directions

One major limitation of our study is that the products we chose likely exaggerated gender stereotypes. Choosing products for which there are separate male and female characters likely cast them as more opposite than might be the case for more gender-neutral items. Thus, studying more gender-neutral toys would likely have led to different results. It should be pointed out that there were very few gender-neutral popular Halloween costumes, and no popular dolls were male. There were a few action figures that were female that we did not code. Among the least feminine-stereotyped of female characters was the female power ranger Valentine. Among the Valentines, there were many that did not have obviously gendered characters so they were not included in our analysis. However, many of these were still gendered in their subject matter such as a set of baseball-themed cards (stereotyped male) and My Little Pony cards which have pastel colors (stereotyped female).

The coding system of our study is another major limitation. The authors of the study were the coders of the products, which could lead the validity of the method to be suspect. Our validity study suggested that our categorization of the characters was consistent with how they were viewed, and we did achieve fairly high consistency across the various coding cues, which argues against an invalid system. It would be helpful if other research used the coding system and also achieved consistency in coding. We compiled the cues we used from several different studies, but more information could be gathered to test their validity. Although we tried to match up the cues we coded with gender prescribed and proscribed characteristics garnered from other research (Rudman et al. 2012), there are likely many gender-related codes that could be relevant. For example, some research (e.g., Auster and Mansbach 2012) has investigated gender stereotyping in the color of toys which we did not do in the present study. Including color as a way to potentially stereotype a toy might moderate the stereotyped depiction.

It is difficult to derive samples of commercial products. We tried to select the most popular products but were not able to do that for the Valentine cards. We might have a different dataset if we had chosen other stores. Popularity might not be measured the same way by different stores and their websites, so it is not a clearly operationalized variable and it may change over time. Nevertheless, it seemed important to examine products that were popular to try to examine products that children actually use. The fact that the costumes and the dolls and action figures were the most popular means that less popular items were not coded, and perhaps these products would have been less stereotyped. Among the Valentines, there were gender-neutral options available that were not coded.

We also did not code for racial/ethnic diversity in the characters because there was very little racial/ethnic diversity in the portrayals. In the Valentine cards, for example, there was only one male character (Avengers) and two female characters (Dora and Jasmine) who did not appear Caucasian. Most disturbingly, although the African American Disney Princess Tiana was portrayed along with other Disney princesses on the outer box of Valentines, Tiana was not represented on any of the Valentines cards themselves. This lack of diversity in products available from popular vendors needs to be explored in future research.

Conclusion

The products we examined were found to be very gender-stereotyped in a way that could perpetuate patriarchy in that male characters were more often associated with traditionally masculine characteristics associated with high status and dominance whereas female characters were more often associated with cues of sexual submissiveness. These patterns of dominance cues being prescribed for males but proscribed for females and sexual submissiveness prescribed for females are problematic. The fact that these are products with which children might identify could lead them to be strongly associated with the development of gender-stereotyped behavior among children. Future research should examine how children play with gendered cultural products and whether their use is associated with the adoption of gender-stereotyped behavior. Although women’s roles in U.S. society have changed and women have increasingly endorsed instrumental traits, men have not increasingly endorsed communal traits (Twenge 2009). This might be due in part to very inflexible cultural models for masculinity. The gendered nature of the portrayals we found is consistent with the general cultural portrayal of women and men noted more than four decades ago that “males act and females appear” (Berger 1972, p. 45).