Picture Images of Girls and Women in Children's Literature

The purpose of the present study was to evaluate the underlying meanings conveyed by media images to preschool age children in three samples of awardwinning picture books . One general hypothesis was that female characters would be underrepresented in illustrations found in children's picture books relative to male characters. A second hypothesis was that female characters would more often be portrayed in subordinate and degrading images than male characters. It was predicted that girls/women would be presented more often than boys/men in traditional and passive roles, shorter, below, behind, in deference (bent over, head tilted), as objects of ownership (arm-lock, shoulder-hold, hand-hold), employing the feminine touch, receiving instruction, smiling, attempting to hide face with hands, sucking/biting fingers , averting head/eyes , and glancing toward an unidentifiable object (mental drifting) . A third hypothesis was that books written during an earlier time period (1967-1976) would contain a greater number of subordinate images of women and dominant images of men (as defined by each of the predictions in hypothesis 2) than books written during a more recent time period (1987-1996). In Part I the sample of books examined were Caldecott Medal and Honor books, Boston Globe Horn Book Award-winning and Honor books, and New York Times Choice of Best Illustrated Children's Books of the Year Award (N = 294), representing two separate time periods, 1967-1976 and 1987-1996. Twenty books were randomly selected from each time period for a total of 40 books. The researcher counted and recorded the total numbers of individual girls/women and boys/men in each of the 40 books. All pictures from each time period were reviewed and given a number. A random sample of forty pictures (twenty from each time period) were selected. The researcher recorded time period, ethnicity of characters, and author for each of the 40 pictures for use in post hoc analyses. The forty pictures were made into slides and presented to a group of 20 raters from the Community College of Rhode Island who rated female and male characters in each picture on function ranking, physical positioning, and facial expression categories. In Part II, 111 participants from Rhode Island College and the Community College of Rhode Island enrolled in a Social Psychology, Human Services, or Marketing course and 19 parents of preschool age children (these persons were asked by the students to participate in the study) were asked to view a series of 36 slides (two pictures from each of the 18 categories listed as predictions in the second hypothesis) and rate the female and male characters on a Semantic Differential scale . Participants were then asked to complete the Modern Sexism scale. There were three main findings from the present study. First , support was provided for the hypothesis that female and male characters would not be represented equally in illustrations. Overall, there were significantly more boys/men presented than girls/women. There was a significant increase in the number of boys/men pictured over time, yet no difference was found for girls/women over time. Second, partial support was found for the hypothesis that girls/women would be presented in subordinate and degrading images more often than men. Females were more likely than males to be presented in passive roles, shorter, in deference (body bent over, head tilted), receiving instruction, and expressing fear. Boys/men were more likely to be shown grasping girls/women in shoulder-holds and hand-holds. Contrary to prediction, males were more often shown below and behind females, and employing touch more often than females. No differences were found between girls/women and boys/men on the remaining seven categories. These findings on images were strengthened through the analysis of visual cues in Part II. Raters interpreted visual cues differently for female and male characters on the factors of activity , potency, and evaluation. Specifically, boys/men were rated as more active and potent, and were evaluated more negatively than girls/women. Third, no support was found for the third hypothesis that pictures from 1967-1976 would contain more subordinate images of girls/women and dominant images of boys/men than those from 1987-1996, with the exception of one analysis. Boys/men from 1967-1976 were more likely to be presented as sucking/biting fingers than boys/men from 1987-1996. Post hoc analyses were conducted to determine whether (a) girls/women of European ethnicity and of African/Asian/Hispanic ethnicity were portrayed in subordinating/degrading images more often than boys/men of the same cultural background and (b) a difference exists between participant scores on the Modern Sexism scale and ratings of pictures on the Semantic Differential. ACKNOWLEDGMENT A specia l message of appreciat ion to those who have supported my efforts in completing this research endeavor . First , I would like to thank Bernice Lott for being such a wonderfu l mentor -provid ing know ledge , guidance , construct ive feedback , and words of encouragement when most needed. I would also like to thank members of my defense committee for their careful review of my manuscript -Karen Schroeder, Larry Grebstein , Kat Quina , and Judith Anderson. Next , I would like to thank Jim & Mary Rossi, and Cathy & Jen Kennedy who freely gave of themselves during my "time of great need" . My parents , Jim & Rose Turner, deserve endless gratitude for all that they have done for me. They deserve special recognition for all the support and love they have provided . could not have accomplished this without them . Last but not least , to my husband, Bob Bowker -a heartfelt "thank you" for so often putting my needs ahead of yours, for your friendship, support, and love.


LIST OF TABLES
101 Age, Ethnicity, and Relationship Status of Participants in Part II Table 2 102 Working Sample of 40 Randomly Selected Pictures Table 3 105 Prevalence of Girls/Women and Boys/Men Characters in Pictures Table 4 106 Chi Square Analyses of Overall Frequencies on Function Ranking, Physical Positioning, and Facial Expression Categories Table 5 108 Chi Square Analyses Comparing Data for Girls/Women Over Time Table 6 109 Chi Square Analyses Comparing Data for Boys/Men Over Time Table 7 110 Post Hoc Chi Square Analyses for Pictures with Characters of European Ethnicity Table 8 111 Post Hoc Chi Square Analyses for Pictures with Characters of African/Asian/Hispanic Ethnicity Table 9 112        Representations are produced in a political socia l structure, in a cultural context, and have implications for development of norms and ways of behaving in our culture and for individuals' constructions of their own reality . The media often present women as desperate , dependent, child-like, weak, and passive. On the other hand, men are presented as independent, decisive, dominant , strong and active. These types of representations affect the attitudes of women and men, setting up gender categories which distinguish certain ways of behaving for those who fall within a particular group. "Cultures distinguish between two or more genders and organize beliefs and activities according to these categories.
Individuals are influenced by the existence of these categories and their perceptions of the world are organized according to them" (Beall, 1993, p.144 ).
In this way, gender is constructed socially and used as means to control women .
For women , the political and social realities of a hegemonic society are efficacious . Misogynist attitudes and stereotypes prime men to take part in negative behaviors against women .  proposes a model of sexist responses to women which include (in order) humor, put-downs, pornography, institutional exclusion, personal distancing, insults and harassment, intimidation, sexual coercion, sexual abuse, physical abuse , and murder . The model is presented as a spiral, with humor and put-downs (as the broader component) at one end and murder (as the narrower component) at the other end in a continuum of hostile behaviors toward women (see Appendix A) .
There exists a relationship between the subtle dimensions of misogyny (humor, pornography) and the more overt forms of violence against women. For example, pornography has "institutionalized a subhuman, victimized , and secondclass status for women" (MacKinnon, 1993, p.279), presenting them as objects , stripping them of their identifications and priming them as recipients of verbal , physical , and sexual abuse. Sut Jhally examined the presentatio~ of women as objects in his video documentaries Dreamworlds I (1993) and Dreamworlds II (1997) . Women of the Dreamworld are presented in degrading images, as desperate and dependent , unable to cope in the absence of men , nymphomaniacs , and child-like . They are open to being watched are passive things --to be used and explored at will. There is a lack of identification for these women as their subjectivity is denied. Camera focus is almost exclusively on body parts, while the whole person (with thoughts, feelings, intelligence) is ignored. As objects, these women are open to the actions of others. They are used for sexual gratification by men, and are available as targets for the frustration and aggression of men.
Studies of pornographic material have found an association between negative images of women and violent behavior against women . Cowan, Lee, Levy, and Snyder (1988) evaluated 45 X-rated videos for themes of dominance and inequality. Of 443 sexually explicit scenes, 54% contained one or both of these themes. Men almost always ejaculated on women rather than in them.
Male characters were presented as wealthier and more powerful than women . Furthermore , women were often presented as child-like --with soft low voices , barrettes in hair, and the absence of pubic hair. Physical aggression appeared in 23% and rape in 51 % of scenes. A more recent study by  measured objectification of women among other scored items in their analysis of X-rated videos. Findings show a large proportion of characters involved in behaviors which present women as objects for use for men's sexual pleasure . Men ejaculating on a woman's face or in her mouth was found to be one of the most frequently occurring behaviors in the videotapes (32%). Anal sex was performed by 52.5% of males . Twenty-nine percent of female characters began sexual activity with fellatio. In addition, 29.8% of women served as objects of physical aggression .
Pornography exists within  model as a means to reinforce and maintain gender inequality. However, this is not the only form of media which serves the political purpose of objectifying and degrading women . Research has revealed systematic subordination of women in other media image presentations .  investigated the relationship between images of women and men in advertisements and cultural behavioral scripts. He suggested that media images establish social order --regardless of the actual experience of its participants. Advertisements do not reflect actual behavior of women and men; instead they reflect how we think women and men behave . Advertisements attempt to convince us that this is how women and men are, or should be, or want to be . The advertisements appear to be "normal" presentations of the behavior of human beings , but what they are actually displaying is the performance of masculinity and femininity prescribed by societal norms (reinforcing the notion of men as naturally dominant and women as naturally subordinate) . "Gender expressions are by way of being a mere show ; but a considerable amount of the substance of society is enrolled in the staging of it." (Goffman, 1979, p. 8) For example ,  examined the association between images of women and the behavior of children --women posing as children, acting like children , looking like children . He suggests they are posed in ways which would save them from seriousness --head tilted to the side, smiling , hands twisted behind the back, the toes of one foot touching the toes of another foot , hands deep in pockets , etc. According to  what categorizes an individual as gender-classed members of society "is their competence and willingness to sustain an appropriate schedule of displays" (p. 8) of certain behaviors . These behaviors are stripped of (historical) context, so that only the content of the displays dist inguishes between the classes . The implications of this are significant as "men and women take their cues about 'gender behavior' from the image of that behav ior that advertising throws back at them, and they contrive to become the 'people' in those ads" (Gornick , 1979, p.viii).

REVIEW OF RELEVANT LITERATURE
"Texts are important influences that shape us by reflecting the politics and values of our society" (Fox, 1993, p.656) . They are highly interactive; they mold and construct us by presenting images of ourse lves. They define what it means to be female or male in our society . Books provide role models ; from books , children learn what behavior is acceptable for them , for their peers, and for adults around them. They learn what to say and do, they learn what 's expected of them, and they learn right from wrong.
For many years authors of children's literature have portrayed girls and women with narrow characteristics. They are often secondary characters; are regularly found in domestic settings; and are often in need of rescue by male characters . Boys and men are also presented in stereotyped roles , but these roles are more positive and sought-after. For example, boys and men more often serve in central roles (as protagonists); are portrayed as leaders, decisionmakers, and heroes; and are often involved in occupations and roles outside of the home. As  pointed out, female readers of children's literature must identify with the male characters in these stories if they are to gain any sense of achievement from literary role models .
In the early 1970s, studies of gender bias in children's literature emerged as a result of the women's movement. Since then, most research on children's literature has focused on images of women/girls and men/boys in roles, activities, and occupations . Reviewing literature from the 1960s,  found that male characters were more often viewed in dominant, active roles (adventurous, bread-winning) , while females existed in passive, victimized roles. From this data,  concluded "boys do; girls are".  conducted one of the "hallmark" studies in the area of gender bias and children's literature. They evaluated Caldecott Medal-winning and honors picture books, Newbery Award winners and runners-up, Little Golden Books , and others described as "etiquette books" written in the 1940s, 1950s, and 1960s . Their main focus was to determine if gender distinctions existed in the prevalence of characters and the representation of characters in roles . They found that females were greatly underrepresented in titles , central roles, and illustrations, by a 1: 11 ratio. In fact, in approximately onethird of the Caldecott books analyzed, there were no females at all. Another major finding was that female characters were generally "inconspicuous and nameless" (p.1128). They were portrayed in roles which are not valued in the eyes of American society. Girls and women were shown as helpers , caretakers , followers, and servers of others, while boys and men were portrayed as exc~ting and engaged in "heroic activity" (p.1131 ). More specific descriptions of roles held by girls and women included the following: passive , immobile , restricted by clothing, prize for male adventurousness , dependent, pleasers , "saved", static , "pretty dolls --to be admired and to bring pleasure". Typical roles for boys and men included leader , independent , achiever , self-confident, outdoors "in the real world", in constant motion , interacting with the wor ld around them , the "rescuer".
For women , occupations included domest ic worker, garden tender , baker, nurse , child-tender, and launderer. Most often, women were presented as mother, wife , fairy godmother , fairy, witch, or underwater maiden. On the other hand, men were storekeepers, housebuilders , kings , princes , fighters, fishermen, policemen , soldiers , cooks , and bearers of knowledge. Notice the distinctions that can be made between the roles of females and males in terms of power , strength , activity, and richness of character . Furthermore, roles prescr ibed for female characters in the children's books are those that are not as highly valued in our society. This creates an even greater distinct ion in meaning and worth for the two genders.  exam ined pictures from 30 coloring books for children which were prominently displayed in retail market stores in order to determine if differences existed in portrayal of female and male characters.
Among other findings , they determined that there were qualitative distinctions between "boy activities" and "girl activ ities" . Boys were pictured primarily in outdoor and competitive activities , whi le girls were shown in more passive activities inside of the home. The most notable difference was in the portrayal of children imitating , in their play activities, the career roles of women and men. The career roles occupied by women tended to require little skill and preparation , whi le the career roles for men "necessitated some specia l skill train ing, or higher education" (p.533) .  analyzed 100 children's picture books published between 1972 and 197 4 and found that only 68 portrayed women in an actual role/activity . In addition, 68% of those in roles were identified as homemakers or domestics.
During the 1970s, sex stereotyping in children's literature was made salient. Publishers agreed to make changes in the texts, scripts, and pictures of children's books in order to ensure equal treatment of women and men. "By 1978 almost all of the major textbook publishers had issued guidelines to discourage sexist portrayals of women in children's picture books" (Lott, 1994, p.48).
Even after the publishers' guidelines, male characters still appear more frequently in titles, central roles and illustrations than female characters . Although the numbers remain unequal, some researchers have found that in the last few decades, there has been a trend towards greater equality in the literature . Specifically, there has been an increase in the number of female characters in titles, central roles and illustrations . Allen et al. (1993) included a third category, neuter gender, in their analysis of Caldecott Award-winners and honors books for a comparison of two time periods, 1938-1940 and 1986-1988. Characters categorized as of neuter gender were typically animal characters which were not clearly identifiable as either female or male . They found an increase in this category over time and suggested that use of gender neutral characters presents children with an opportunity to model characters that prescribe certain positive behaviors, yet do not exemplify particular gender-roles .
Consistent with past research Rachlin & Vogt, 197 4;, recent findings indicate that female characters are still portrayed more often than male characters in passive, domestic, limited, and devalued roles, while males thrive in active, dominating, valued roles . In a sample of 14 Caldecott Medal-winning children's books,  evaluated the occupations of female and male characters and found gender bias. In particular, her results indicate that men were presented in three times as many different occupations as women.
Research on children's literature has generally not examined the more subtle aspects of the presentation of girls/women and boys/men, although there have been a few exceptions. The present author examined gender stereotyping in children's literature through the use of differential language . Focusing on the adjectives in a sample of Caldecott Award-winning picture books, I found that male characters were described as more potent (powerful), active, masculine, and with more negative evaluation than female characters. Other researchers have investigated the subtleties of images in picture books for preschoolers.  found a difference in the physical presence (power position) of characters . Males were found to be taller, in front of, or leaning over female characters.  examined four children's picture books (Angry Arthur; Now One Foot, Now the Other; Willy the Wimp; and Madeline) to determine how images and text play into young children 's understanding of gender .  suggests that the study of images of a particular historical moment can teach us about the inner lives of people. Spitz ( 1994) followed this suggestion and asked "what we can learn about the psychological development (of children) if we study the visual environment that leaves its mark on them" (pp . 308-309) . She suggested that adults are able to negotiate through from artistic images to reality, yet children are not as able to distinguish the boundaries between art and life.  used the example of the images of girl as nurse and boy as doctor. When children view images like this , what they assimilate is that the images reflect real life roles/occupations. 9 This is problematic as images viewed early in life serve "a significant part of the organizing experience" (p. 311) of the mind. The implications of this are great, since the research literature has shown repeated presentations of women/girls and boys/men in stereotypical roles that do not reflect real life behavior .
According to , "even as greater numbers of women become influential professionals, the old images sustain a haunting power ... (and) are not easily eradicated" (p.328).
Research on visual images of women/girls and boys/men in children's literature is very limited, and thus one must turn to other forms of research to learn more about the variables which may impact the social construction of gender . A number of researchers have investigated the pictorial presentation of women and men in advertising. Studies have shown that women are objectified more often than men (Ferguson, Kershel, & Tinkham , 1990). Specifically, women/girl's body parts are featured more often than their faces in print advertising . In a recent analysis of 59 television beer commercials,  found a significant difference in the presentation of women's and men's bodies. For women there were significantly more camera shots focusing on specific parts of the body (chests, buttocks , legs, crotches) than for men. In addition , men's faces appear twice as often as women's faces . This suggests that women continue to be presented in as "empty vessels" .
Past studies of media presentations have found women to be shown in degrading positions that emphasize body parts (Thomas , 1986;, that place them in submissive positions where they appear to be smaller in stature than men  , displaying more "emotional"  and less serious  facial expressions than men. 10 A more recent study by  examined media portrayal of women in advertisements for exercise equipment, health products, and fitness apparel. They analyzed 191 advertisements from 5 health and fitness magazines for posture, dominant versus passive placement, active versus passive behavior, dismemberment of body parts, and facial emotional display, among other variables. Results indicated that women were more often found in positions with posture inappropriate for exercise (curved and sexually exploitative poses). Also, significantly more women were placed in submissive/passive positions relative to men. In fact, in one of the magazines, more than 80% of women were in passive postures to men . In no instance was a man placed in a submissive position relative to a woman. Women were also shown in inactive poses more often than men. Approximately 40% of all advertisement photos either focused on, or unnaturally emphasized, specific body parts of the models, and approximately 50% of women photographed displayed sexual emotional facial expressions.
Other studies have examined age-ism in image presentations and found that women are regularly presented as younger than men . Furn ham and Bitar ( 1993) examined the portrayal of women and men in a sample of 180 British television commercials . Findings indicated that women were more often described as younger, and men were more often described as middle-aged.
Research in the area of advertising has concentrated on the more subtle aspects of media images of women and men by examining body positioning, "face-isms" (degree to which the camera shot focuses on the face versus the body}, "body-isms" (degree to which the camera shot focuses on the whole body versus body parts}, and "age-ism" (young versus old). Yet no one has taken the study of advertisements to the extreme that  did. Original to his work is the analysis of minute aspects of the images, for their comprehensive meaning. He concentrated on facial expressions, head postures, head-eye aversion, relative size, body positioning and placing, finger biting and sucking.  compared groups of pictures to images which contain the same scene or activity, yet with a switched-gender "character" to allow the reader to see the differences in portrayal and meaning conveyed by the images when women or men are featured in them. Goffman proposed that "(g)esture, expression, posture reveal not only how we feel about ourselves but add up, as well, to an entire arrangement --a scene --that embodies cultural values ... " (Gornick, 1979, p.vii) He found that simple gestural messages in images from magazine advertisements function to construct and maintain gender stereotypes by presenting women in subordinate poses in comparison to men. A more recent study by  replicated the work of Goffman, and found no difference in the way that women are presented in picture images over time.
We communicate both verbally and nonverbally with other organisms in our environment. We make statements about our situation, about what activity is going on, and about what relationship we're in through "glance, posture, and movement" (LaFrance, 1978, p. 3). Our body language is not meaningful out of context, rather it is a function of both the person and environment and is subject to the changeable nature of each .
It has been suggested that gender role expectations for women and men affect nonverbal communication styles, and subsequent attributions about power differentials in relationships Frable, 1987;. Nonverbal behavior indicative of dominance and subordination has been found to be related not only to situations involving gender, but also to more generalized situations regarding status (involving age, socioeconomic status, etc.). For instance, research has demonstrated a high similarity between women's body language and that of lower-status men. Both have been found to have tense posture, smile more, avert their gaze from direct eye contact, and covertly watch others  .
Research has documented that women tend to smile more than men . Women who smile more often have been rated as more interpersonally attractive than those who do not . Reis, Wilson, Monestere, Bernstein, Clark, Seidl, Franco, Gioiso, Freeman, & Radoane (1990) found that female and male college students rated smiling images of women and men as more attractive then nonsmiling images. In addition, stimulus persons who were smiling were rated as more sincere, sociable, and less masculine than those who were not smiling. Similarly,  asked 150 undergraduate students to observe two out of 40 videotaped conversations of a female-male pair who exhibited differing combinations of nonverbal behaviors.
Findings indicated that high maintenance of eye contact, smiling, and close interpersonal distance communicated higher level of composure and less emotional arousal. High maintenance of eye contact and close interpersonal distance alone communicated dominance and control.
Differences in the meaning of physical/body positionings have also been well-documented in the literature.  conducted two studies to determine the influence of a stimulus person's height on impression formation among preschool-age children. Results of both studies supported children's interpretation of taller stimulus persons as more dominant and stronger than shorter stimulus persons . In another study by Schwatrz, Tesser, & Powell ( 1982) 144 undergraduate students were asked to evaluate 32 drawings of women and men in various body positionings. Four positions were varied -lateral opposition Physical/body position can also serve as a function in appeasement. For instance ,  examined the antagonistic behavior of 34 elementary school age boys and found that body signals of submission such as knee ling, bowing, and shoe tying serve to appease escalating antagonistic behavior among the children.
It has also been reported that women tilt or cant their heads more often then men . However, these findings have been inconsistent .
Finally, touch has been identified as an indicator of differing status .  observed 101 instances of touch in various locations in Baltimore, Maryland. She found that men were more likely to initiate touching others, while women were most often the recipients of touch. This finding has been replicated by  who observed people in public settings (i.e., parks, beaches, airports) apd found that men were more likely to touch women than vice versa. However , they did not find this gender difference to exist among children.  also reported that women were less likely to reciprocate a man's touch than for men to reciprocate a woman's touch.
Interestingly, according to , this result has been found to occur outdoors moreso than indoors. Also , findings have suggested that men, older persons , and those of high socioeconomic status are more likely to touch those of lower status (women, younger people , those of low socioeconomic status)  . Other researchers have found that men are more likely to put their arms around women , and women are more likely to link arms with men (armlock) .
Differences also exist in the way that persons interpret touch . Fisher ,  investigated 94 female and male undergraduate student's responses to interpersonal touch when handed a library card . Findings indicate the women generally reported a more positive affect and evaluation after being touched than did men , who were ambivalent to touch. Burgoon (1991) examined the ways in which 622 adolescents and adults interpreted messages associated with touch. Results indicated that people interpreted touch in a number of ways -as a sign of composure, trust, affection, similarity, dominance, and informality. Specifically, face touching and hand-holding was interpreted as indicative of the most composure and informality , while handholding and handshaking expressed the least dominance.
To summarize, a positive relationship has been demonstrated to exist between negative presentation of and hostile behavior toward women. Research on nonverbal behavior suggests that differing meanings of dominance, control, and submissiveness may be derived from images of persons, depending on their gender and status. Although a limited number of researchers have attempted to examine the more subtle aspects of sexism in children's literature, careful examination of picture images has been a neglected area of study. The present investigation will utilize techniques first developed by  to analyze the body language of girls/women and boys/men in award-winning samples of children's picture books.
According to  one way in which social power , authority , and rank is expressed in social situations is through function ranking. Men are more often portrayed in an executive role, as leader, working outside of the home and in active roles . This status position is performed in and outside of occupational role . A second method to distinguish those of differing social power is through body positioning. Here, differences in size and position often correlate with differences in "social weight" (Goffman , 1979, p.28) . Physical positioning can also objectify/define women and children as property. Finally, facial expression of characters relay the subordinate position of women by removing them psychologically (emotionally or intellectually) from the situation at hand. Women are usually presented as smiling [communicates submissiveness and/or facilitating interpersonal relations Konner, 1987)), fearful, hands blocking face as though shy or hiding something or unwilling to express themselves, sucking on finger or biting nail(s) to give the impression of anxiety , as unable to make eye contact (Tseelon , 1991), or glancing away into nothingness .
The purpose of the present study was to evaluate three samples of children's picture books to determine if differences exist in the function ranking, physical positioning, and facial expressions of female and male characters.
Examination of these variab les will allow a comprehensive analysis of the underlying meanings conveyed by media images to preschool-age children. The importance of this research lies in the implications of gender role stereotyping for both girls and boys. Traditional media images teach young children how to behave as a girl/woman , or how to behave as a boy/man. Expectations for behavior are defined through gender roles, and are limiting for persons of either sex. Distinctive patterns are created in terms of acceptable behavior for girls (subordinate, passive, quiet , inactive, etc.) and for boys (dominant, active, "with voice", etc.), rather than on positive human characteristics that may be sought by all. Negative consequences for girls and boys have been documented in activity level, feelings of independence/dependence, aggressive behavior, cognitive performance, school achievement, and vocational aspirations .

HYPOTHESES
One general hypothesis in this study was that female characters would be underrepresented in illustrations found in children's books relative to male characters.
A second hypothesis was that female characters would more often be portrayed in subordinate and degrading images than male characters. The following predictions were made. For each of the predictions presented, characters compared were of roughly similar age. 1. Girls/women would more often be smiling than boys/men.
2. Girls/women would more often be presented with expressions of fear than boys/men .
3. Girls/women would more often be presented as attempting to hide facial expression with hand(s) than boys/men . 4. Girls/women would more often be presented as sucking or biting their finger(s) than boys/men . 5. Girls/women would more often be presented in head/eye aversion than boys/men. [turning head and eyes away from person who is looking directly at you] 6. Girls/women would more often be presented as glancing toward an unidentifiable object than boys/men. [as in a euphoric state, "mental drifting"] A third hypothesis is that books written during an earlier time period (1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976) would contain a greater number of subordinate images of girls/women and dominant images of boys/men (as defined by each of the predictions in hypothesis 2) than books written during a more recent time period (1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996).
Part I Sample METHOD Group 1. The books examined for this portion of the study are Caldecott Medal and Honor books (Association for Library Service to Children, 1996) for the periods 1967-1976 and 1987-1996. The Caldecott Medals are given by the American Library Association to honor the year's most distinguished children's books for preschoolers (ages three to six). The Caldecott award has been presented annually since 1938 for the best in picture books for preschoolers . Runners-up have also been recognized in each category, and are now called Honor books. The popularity of these books, both in libraries and in bookstores, suggests that they are accurate representations of children's actual reading material. Previous investigators have examined these books for gender stereotyping .  1967-1976 and 1987-1996. This award is cosponsored by The Boston Globe and The Horn Book Magazine and has been presented annually since 1967 for the best in text and illustration. More recently the award has been presented to winners in each of three categories -outstanding fiction, outstanding nonfiction , and outstanding illustration. As many as three Honor books per year may be cited (Jones , 1988). For the present study, only those awards presented for illustration in children's picture books (ages three to six) were included in the

Procedure
Step 1. The researcher reviewed all books (N = 329) from each time period to determine if any duplicate award-winners were present in the sample.
Thirty-five duplications were found (as indicated with an asterisk in Appendices E, F, & G) . This decreased the overall number of books in the sample to 294.
Step 2. For the time period 1967-1976, all books (N = 145) were given a number. Twenty books were randomly selected from this time period using a table of random numbers . For the time period 1987-1996, all books (N = 149) were given a number. Twenty books were randomly selected from this time period using a table of random numbers. Five of the books chosen randomly were replaced (using random selection) due to the fact that they contained images of animals whose gender was not readily apparent (see Appendix H) The researcher counted and recorded the total numbers of individual girls/women and individual boys/men pictured in each of the 40 books.
Step 3. For the time period 1967-1976, all of the pictures from the randomly selected books were reviewed. Only those pictures including two children (female/male dyad) of roughly similar age or two adults (female/male dyad) of roughly similar age were included in the sample . These pictures were given a number, and a random sample of 20 pictures was selected using a table of random numbers. The same was done for books selected from the 1987-1996 time period. This produced a sample of 40 pictures. The researcher recorded time period , ethnicity of characters, and author for each of the 40 pictures for use in post hoc analyses .
Step 4. The 40 pictures were made into slides and presented to a group of 20 raters . The raters were undergraduate students from the Community College of Rhode Island, participating for an extra credit assignment for their social psychology course. There were two rating sessions with 10 students in each session. Raters were given an informed consent form which was removed from the rating form to ensure anonymity of responses (see Appendix I). A list of raters was provided to the instructor for the provision of credit (amount determined in advance by the instructor). The raters also received a set of standardized instructions (see Appendix I), training by the researcher, and an opportunity to practice rating two pictures (one practice picture was selected from each time period) . The order of stimulus presentation was counterbalanced for the two sessions to contro l for bias . Each rating session lasted approx imate ly 120 minutes (30 minutes for introduction and instructions; 90 minutes for rating task) and took place during one regular class meeting time. A form was designed to facilitate data collection. The raters completed one form per picture. The pictures were identified with a number at the top of each form . Each picture was rated on 18 categories , corresponding directly to the predictions made in hypothesis 2 (see Appendix J) . After completing the ratings participants were debriefed regarding the intent of the study . Students received class credit (or ext ra credit) for participating in this research project. The researcher informed eac h instructor of the student's participation so that credit cou ld be awarded upon completion. The amount of credit provided was determined in advance by each instructor. The remain ing 19 participants were parents of preschool age ch ildren . These individuals were asked by the undergraduate students at the Community College of Rhode Island to participate in the study . The ages, ethnicities , and relationship status of the participants, 94 women and 36 men , are shown in Tab le 1. Overa ll, forty percent of participants were parents (N = 52). Sixty-four percent of parents were undergraduate students and 50% of parents had preschool age children.

Instruments 22
A form was prepared for use in data collection. One form was used for each picture rated . Each picture was identified with a number at the top of each form. There were two pictures from each of the 18 categories listed as predictions related to the second hypothesis , for a total of 36 (see Appendixes B, C, and D). The form included 11 Semantic Differential rating scales (see Semantic Differential scale . The Semantic Differential scales, originally developed by Osgood, Suci, and Tannenbaum (1957), employs bipolar adjectives and seven-point rating scales. The respondent is asked to rate some idea, concept, or issue by checking off one of the seven spaces between the bipolar adjectives . Test-retest reliability of the scale was determined by Osgood et al. to be .85. Reliability coefficients for individual items were not computed as the Semantic Differential scores are too consistent. On many items, there is such close agreement on scale position that variance approaches zero. Factor-score analysis produced three factors: activity, evaluation, and potency (power).
Average error of measurement (expected to be smaller when the instrument is more reliable) was much smaller in the evaluative scales than in either the activity or potency scales . Criterion validity of this measure could not be assessed by Osgood et al. , as no quantitative criterion of the measure of meaning was available to compare to (for correlating scores). Face validity , the extent to which the measure's "distinctions ... correspond with those which would be made by most observers without the aid of the instrument" (p. 141 ), was determined by asking participants whether discriminations made by the instrument correspond with her/his own judgments. Data collection on participants rating 1 O concepts using the Semantic Differential scales produced three clusters, including words of similar meaning in each of the three clusters. Osgood et al. suggest that most people would cluster the concepts in the same way based upon their meanings, without the use of the Semantic Differential scale. In other words, the rating of factors on the scales often reflect what is expected through common sense .

23
Three factors measured by the Semantic Differential were included in the present study: activity, evaluation , and potency (power) . Activity was measured by three subscales (warm-cold, loud-quiet, moving-still) . Evaluation was measured by five subscales (good-bad, pretty-ugly, friendly-unfriendly, healthysick, happy-sad). And potency was measured by three subscales (strong-weak, big-small, heavy-light) .
The 11 Semantic Differential scales were completed by participants for the girl/woman presented in each picture and the boy/man presented in each picture. Therefore , each form contained two scales with the following instructions, "Look at the girl/woman in the picture and rate her on the following scale", and "Look at the boy/man in the picture and rate him on the following scale". There were 36 pictures , with a female/male dyad of similar age in each . Thus , each participant was asked to complete a total of 72 scales (two scales per form for a total of 36 forms). The scales were counterbalanced and the order of bipolar adjectives was varied to avoid practice effects and rating error. Participant ratings on the Semantic Differential tested the meaning of visual cues in the pictures. This allowed the researcher to determine if a quantitative difference exists between the images of girls/women and the images of boys/men on the factors of activity, evaluation, and potency.
Modern Sexism scale . The Modern Sexism scale is an eight-item inventory designed to measure covert or subtle sexism. It measures "whether respondents tend to (a) deny the existence of discrimination against women, (b) · resent complaints about discrim ination, and (c) resent special 'favors' for women" (Swim & Cohen , 1997, p.105 Modern Sexism scores were sign ificantly higher than scores for women (M = 2.14), t (628) = 8.55, p<.001 . This find ing was replicated in a second study by the authors  . To determine construct validity of the measure , Swim et al. (1995) examined the relationship between scores on the Modern Sexism scale and attributions for job segregation among women and men.
Findings indicate that scores on the Modern Sexism scale are predictive of individual attributions for job segregation. Specifically, those who earned high scores were more likely to attribute sex segregation to biological determinants ; while those with low scores were more likely to indicate that socialization, prejudice, and discrimination were causes. Modeling past research on the Modern Racism scale, Swim et al. (1995) examined the relationship between Modern Sexism scores and voting preferences to determine construct validity.
They predicted that Modern Sexism would be a better predictor than Old-Fashioned Sexism for a female or male candidate . Controlling for liberalism and party affiliation,  found that respondents with lower Modern Sexism scores were more likely to prefer to vote for a woman candidate (Modern Sexism was determined to be a better predictor of voting behavior than Old-Fashioned Sexism measures) . More recent work by  compared the Modern Sexism scale and the Attitudes Toward Women Scale (AWS)  . Their results lend support for convergent and discriminant evidence of construct validity. Specifically, the AWS and the Modern Sexism scales were found to tap different, but related constructs.
Both are related to negative attitudes about women , yet the AWS seems to measure overt or blatant sexism, while the Modern Sexism scale seems to measure more covert or subtle forms of sexism.
For this measure, participants were asked to rate their agreement with individual items on a 7-point Likert-type scale (from strong ly disagree to strong ly agree) . Low scores (0-3.99) indicated less sexist attitudes, while high scores (4-7) indicated more sexist attitudes .

Procedure
Participants were asked to complete an informed consent form (student or parent version , as appropriate) and demographic survey (see Appendix L). The informed consent forms were removed from the answer form to ensure anonymity of responses. The names of student participants were provided to the instructor for the provision of credit (amount to be determined in advance by the instructor).
All participants were then provided with a set of standardized instructions (see Appendix M). Pictures (transformed into slides) were presented one at a time to each group of participants. They were asked to review each slide and complete two survey forms (one for the girl/woman ; one for the boy/man) for each slide presented . Next, participants completed the Modern Sexism scale  (see Appendix N) and were debriefed as to the intent of the study.

Part I
For each time period, the prevalence of girls/women and boys/men in pictures was assessed by the researcher who counted and recorded numbers of characters of each gender in the books. The numbers of female and male characters in each book was counted twice by the researcher . From 1967From -1976 there are 808 girls/women and 1234 boys/men presented in pictures . From 1987From -1996 there are 734 girls/women and 1589 boys/men pictured. These data are shown in Figure 1. Table 2 presents data on time period, ethnicity of characters, and author for the sample of 40 randomly selected pictures .
Differences between visua l presentation of girl/women and boy/men characters were determined using 1 x k (Goodness of Fit) chi square analyses .
For the overall data set, a significant difference was found between the prevalence of girls/women (N=1542) and boys/men (N=2823) [x,.(1) = 375.94, p < .05]. No significant difference in prevalence was found for girls/women over time [x 1 1) = 3.55, p > .05]. A significant difference in prevalence was found to exist for boys/men over time [x11) = 44.64 , p < .05]. Table 3 presents the relevant data. It can be seen that, overall, there were significantly more boys/men pictured than girls/women, and that pictures of boys/men increased significantly from the earlier to the later time period .
For each of the 40 randomly selected pictures, participant rated girls/women and boys/men on each of the 18 categories corresponding directly to the predictions made in hypothesis 2. Ratings were tabulated by the reseacher.
Chi square analyses were conducted to determine if there was a significant difference in frequencies of function ranking (traditional versus nontraditional versus not sure, active versus passive); physical positioning [shorter, below , behind , in deference (bent over , head tilted) , as objects of ownership (arm-lock , shoulder-hold, hand-hold) , employing the "feminine touch"; receiving instruction] ; and fac ial expressions (smiling , showing fear, hiding face with hands, sucking/biting fingers , head/eye aversion, mental drifting) between girls/women and boys/men. Specifically, a 3 x 2 chi square analysis was used to evaluate the "traditional versus nontraditional versus not sure" category, and a 2 x 2 chi square was used to analyze the "active versus passive" category. The 16 remaining categories were evaluated using 1 x k chi square analyses .
For the overall data set , significant differences were found between girls/women and boys/men on the following categories : function ranking (active versus passive); physical positioning (shorter , below, behind , bent over, head tilted, shoulder-hold, hand-hold , feminine touch, receiving instruction); and facial expressions (showing fear) . Girls/women were more likely than boys/men to be presented in passive roles [x' \1 ) = 66.02 , p < .05] . More often , females were presented as shorter [x -\ 1) = 28.88 , p < .05] than males. In addition , girls/women were more likely to be presented in positions of deference to boys/men, bent over presented employing the "feminine touch" [x 1 1) = 9.13 , p < .05]. Table 4 presents these data.
Chi square analyses comparing earlier data for boys/men (1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976) to more recent data for boys/men revealed a significant difference only on the sucking/biting fingers category . Boys/men were more likely to be shown sucking/biting fingers in pictures from the time period 1967-1976 than in those from 1987-1996 [x~1) = 7.00, p < .05]. Analyses conducted on the 17 remaining categories for boys/men reveal no significant difference over time. Table 6 presents these data. Chi square analyses were conducted to determine if significant differences exist in the function ranking (traditional versus nontraditional versus not sure, active versus passive); physical positioning [shorter, below, behind, in deference (bent over, head titled), as objects of ownership (arm-lock, shoulder-hold, handhold), employing the "feminine touch", receiving instruction]; and facial expressions (smiling, showing fear, hiding face with hands, sucking/biting fingers, head/eye aversion, mental drifting) between girls/women and boys/men. Specifically, a 3 x 2 chi square analysis was used to evaluate the "traditional versus nontraditional versus not sure" category, and a 2 x 2 chi square was used to analyze the "active versus passive" category . The 16 remaining categories were evaluated using 1 x k chi square analyses. Chi square analyses were calculated twice, as follows: ( 1) using the data set for characters of European ethnicity, and (2) using the data set for characters of African, Hispanic, and Asian ethnicity .
Chi square analyses for the European American character data set revealed significant differences between girls/women and boys/men on the following categories: function ranking (active versus passive), physical positioning (shorter, below, behind , bent over, head tilted, shoulder-hold, feminine touch, receiving instruction), and facial expression (fear , head/eye aversion) . European American girls/women were more likely than boys/men to be presented in passive . These are the same results that were obtained overall. Table 7 presents these data.
Chi square analyses for the African/Hispanic/Asian character data set showed significant differences between girls/women and boys/men on the women students versus men studen ts) and score (low versus high) ; and then compared to factor mean ratings for girl/woman pictures and boy/man pictures using independent t-tests .
From each participants ' response to items on the Modern Sexism scale , mean scores were calculated . Two independent t-tests for samples of unequa l size were conducted to determine if the re was a significant difference between mothers and fathers, and female and male students.
Independent t-tests were conducted to determine if differences exist between participant scores on the Modern Sexism sca le and ratings of pictures on the Semant ic Differential. Specifica lly, mean scores for mothers and fathers , female and male students , and low (0-3.99) and high (4-7) scorers were compared to factor mean ratings for girl/women pictures and boy/men pictures (see Table 9). were once aga in rated more pos itive ly than boys/men (M = 4 .24) , t (3094) = 5.39 , p < .05 (see Figure 10).

DISCUSSION
Resu lts of this study reflect the genera l pattern of gender bias found in other studies of children's picture books . There are three main findings from the present study. First, female and male characters are not represented equa lly in illustrations . Overall, there were significantly more boys/men presented than girls/women in award-winn ing chi ldren's picture books . There was a significant increase in the numbers of boys/men pictured over time , yet no difference was found for girls/women over time . Partia l support was provided for the prediction that girls/women would be presented in subordinate and degrading images more often than boys/men. It was found that females were more likely than males to be presented in pass ive roles ; as shorter than males ; in deference to males (with body bent over and head tilted) ; receiving instruction from males ; and expressing fear. In addit ion, boys/men were more often shown grasping girls/women with hand-ho lds and shou lder-ho lds. These findings take on added sign ificance from the fact that raters in Part II of th is study interpreted visua l cues differently for fema le and male characters on the factors of activity, potency , and evaluation .
The boys/men in the illustrations were rated as more act ive and potent (powerful) , and were evaluated more negative ly than the girls/women in the illustrations.
Contrary to prediction , males were more often shown below and behind fema les than vice versa, and as emp loying the "feminine touch" more often than fema les . Post hoc analyses revealed that regardless of ethnicity, females tended to be presented in subordinate/degrading postures more often than males.
Significant differences between girls/women and boys/men were observed on 11 of the 18 visual categories for pictures with European American characters, and on 1 O of the 18 categories for pictures with persons of color. European female characters were more likely to be presented as passive and shorter than males, in deference to males, grasped by males using a shoulder-hold , receiving instruction from males, expressing fear, and averting head/eyes from males, while boys/men were more likely to be presented behind and below girls/women, and employing the feminine touch more often than girls/women . Female characters of color were more likely than males to be shown as passive, shorter, in deference (with body bent over and head tilted), grasped by shoulder-and hand-holds, receiving instructions, and expressing fear, while boys/men were more often shown behind and below girls/women.
Post hoc analyses on the Modern Sexism scale found that fathers scored as more sexist than mothers, but there was no significant difference between female and male students. No differences were found to exist between participant scores on the Modern Sexism scale and ratings of pictures on the Semantic Differential. All parents (mothers and fathers), students (female and male), and scorers (low and high) rated girls/women and boys/men in the same manner on the Semantic Differential. Male characters were rated as more active and potent than female characters, while, for evaluation, female characters were rated more positively than male characters .

Prevalence
During the 1970s, publishers set new standards for equity in children's picture books as a result of the women's movement. For more than 20 years, researchers have examined whether actual progress has been made over time.
Although equitable representation has not yet been reached , recent studies have demonstrated an increasing trend for female characters to be depicted in central roles . However, boys and men continue to be presented more frequently in book titles and illustrations .
The focus of the present study was on presentation of girls/women and boys/men in illustrations. Results indicate that girls/women continue to be underrepresented in comparison to boys/men in picture book illustrations. In addition, although there has been no change in prevalence rates over time for women, prevalence rates for boys/men in illustrations have increased over time .
The disproportionate numbers are reflective neither of the gender distribution in the United States, nor of the numbers of girls and boys who use these books.

Subordinating/Degrading Images
Another main goal of this study was to determine whether physical differences exist in the ways female and male characters are presented in book illustrations . Rather than focus on the characters' activities, occupations, etc., this research focused on the appearance of characters -the ir roles, posture , body position, and facial expressions.
Three areas of concern were function ranking (traditional roles versus nontraditiona l roles , active versus passive roles); physical positioning (shorter, below , behind , bent over , head tilted , arm-lock, shoulder-hold, feminine touch, receiving instruction); and fac ial express ions (smiling , fear , hiding face with hands, sucking/b iting fingers, head/eye aversion, mental drifting).
Contrary to pred iction, most of the ana lyses conducted did not show a difference between girls/women and boys/men in traditional versus nontraditional roles, but it is important to note that most of the characters of both genders were shown in traditional roles . In the present study, as in previous ones, girls/women were more often presented in domestic roles, inside of the home , serving others , in need of help or rescue, etc., wh ile boys/men were more often portrayed as leaders, working outside of the home , dec ision-maker, and hero McDona ld, 1989;Rach lin & Vogt , 1974 ;. These roles are soc ial positions regulated by norms which define "proper" and acceptable behavior for women and men , boys and girls.
Overall , girls/women were shown in passive roles significantly more often than boys/men in both of the time periods examined (1967-1976 and 1987-1996).

Female characters of both European and Afr ican/Asian/Hispanic ethnic ity were
presented as pass ive more often than male characters . These results demonstrate that females cont inue to be presented in ineffective roles. Past research from children's literature  and advertising literature  has also shown that girls/women are presented as weak , passive , immobile, vict imized, dependent , objectified , inconspicuous , while boys/men are portrayed as active , independent, adventurous , dynamic.

38
Overall, girls/women were more likely to be presented as shorter than boys/men . Female characters, both of European and African/Asian/Hispanic ethnicity , were systematically shown as shorter than male characters. These findings are also consistent with other data from children's literature  and advertising  . The implications of these findings are made salient through  studies on impression formation among preschool age children, which demonstrated that children interpret taller stimulus persons as more dominant and stronger than shorter stimulus persons.
In the present study, males were more often positioned below and behind females. This differs from the report by  who found male characters in children's picture books to be positioned in front of and leaning over females, assuming a domineering or power position. Present findings also contradict  who found that a figure preceding (one positioned in front of the other) another was more readily chosen as the dominant figure in a series of drawings presented to undergraduate students. One possible explanation for the present surprising result may, in fact, be the gender stereotypes that exist for boys/men. In American society, social interactions often require "gentlemanly" behavior on the part of boys/men that may include positioning themselves behind and/or below girls/women (i.e., men are often expected to hold a door ajar so a girl/woman may be the first to pass through).
As expected, girls/women were more likely to be placed in positions of deference to boys/men than vice versa . The overall analysis of data revealed that female characters were more likely to be physically positioned with their bodies bent over and heads tilted toward male characters . These results are consistent with past research in children's literature ) and advertising literature , and indicate status differentia ls through subordination of self and appeasement of others. No significant difference was found for girls/women or boys/men over time on these categories .

39
Results which examined female characters as objects of ownership by male characters were mixed . No difference was found for the arm-lock category on any of the analyses conducted. However , a demonstrated difference was found to exist between girls/women and boys/men on the categories of shou lderhold and hand-hold . Consistent with past research , male characters were more often shown grasping female characters with an arm over the shoulder and with a forward-facing hand-hold.
Likewise, present results support findings that girls/women are more often the recipients of controlling touch by boys/men than vice versa . No differences were found for girls/women and boys/men over time on each of the three categories -arm-lock, shoulder-hold , and hand-hold.
For the overall data set , a significant difference was found to exist between girls/women and boys/men on the category of feminine touch, yet not in the predicted direction. Male characters were more likely to be shown employing the feminine touch than female characters . This was quite surprising, as past research has demonstrated difference on this factor favoring women . For this study, feminine touch was defined as "lightly touching and/or caressing" as opposed to grasping , manipulating, and/or shaping. Instructions provided to participants may not have clearly distinguished between touch and feminine touch . If this were the case , it would explain the present results , supporting past research which also indicates that boys/men are more likely to touch girls/women Major, 1991;. The present findings support the work of Fisher at al. ( 1976) and Burgoon ( 1991) who suggest that touch may be interpreted in many different ways, and have a variety of meanings, depending upon context.
The overall data set reflects a significant difference between girls/women and boys/men on the category of receiving instruction. As expected, female characters were more likely to be shown receiving instruction from male characters than vice versa. No differences on this factor were observed for female or male characters over time. Results support past research .
Contrary to prediction, most of the analyses conducted showed no significant differences between girls/women and boys/men on the smiling category . Past studies in advertising media have demonstrated a clear distinction between females and males on smiling behavior. Girls/women have been consistently shown smiling (with a less serious expression) more often than boys/men (Briton & Hall, 1995;.
Overall, pictures of girls/women were more likely than boys/men to show facial expressions of fear. This was true for illustrations of both European and African/Asian/Hispanic characters. No differences on the overall data set were found girls/women over time or boys/men over time. The present findings support past research on emotional displays of fear .
Contrary to prediction, no significant differences were generally found to exist between female and male characters on the categories of hiding face with hands, sucking/biting fingers, head/eye aversion, and mental drifting . However, a difference was found to exist for boys/men over time on the sucking/biting fingers category. Males from the earlier time period (1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976) were more likely than those from the more recent time period (1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996) to be shown sucking/biting fingers. Likewise, a difference was also found to exist between European American females and males on the category of head/eye aversion . Here, girls/women were more likely to engage in head/eye aversion than boys/men.
Generally, the present results do not support past findings in the advertising literature (Dividio et al., 1988;.

41
Taken together, the findings from this study suggest a new form of sexist discrimination, a "modern" discrimination, where overt negative behaviors toward girls/women decrease, yet subtle covert negative behaviors remain. Modern discrimination is often viewed as more insidious, a "wolf in a sheep's clothing" .
Here, feelings of prejudice and negative behaviors continue to exist. Yet, they are expressed only in certain situations when a person may feel safe/comfortable or in covert, socially acceptable ways that often elude notice (Gaertner & Dividio, 1986  visual images of girls/women and hostile behaviors toward girls/women Jhally, 1993;MacKinnon, 1993.
Past studies have examined the role of characters in children's picture books for evidence of gender stereotyping. Focus was on the type of roles, occupations, and activities of the characters. The present study differs from past research in that its focus is not on the overt content of children's literature, but on the hidden messages conveyed through the physical presentation of characters in picture images. This study looked at gender stereotyping from a different perspective , examining more subtle aspects of sexist bias. Overt signs of sexism may be on the decline, yet many of these have been replaced by modern forms of gender discrimination. First glance at a picture book may suggest that characters and topics are presented in a gender-fair manner. Yet, closer examination may reveal this to be only an illusion. Modern forms of sexism may be more insidious than blatant discrimination, because they are much more difficult to identify and address .
The present study also differs from past studies in terms of methodology .
Prior studies of children's literature have been mainly qualitative, conducted through content analyses. Following an earlier study , the present study combines both qualitative and quantitative methodologies, allowing for a more complex analysis of the data.
New questions have been generated by this study . First, the evaluation of characters needs to be examined in greater depth. Are female characters evaluated more positively because they fulfill roles deemed appropriate for their gender? Are females liked more when they take on less competent roles?
Second, research must be conducted to determine the direct effects of sexist books on preschool age children. What are the behavioral consequences after consequences after exposure to sexist forms of media? For instance, is there a direct relationship between negative presentation of girls/women in children's picture books and hostile behavior toward girls on the school playground? These questions should be examined empirically, in order to determine the direct effects of exposure to sexist media images on behavior among children. For instance , one might expose a group of preschoolers to sexist books, and another group to egalitarian books , and measure the children's play behavior (e.g., roles, activity level, physical posturing, level of aggression, etc.) immediately following exposure. Third, methods used in the present study may be applied to investigate other forms of visual media which influence preschool age children . Is there a differential impact of books versus television (e.g., cartoons) on gender role socialization? Next, what is the impact of the presence of a mediator in the acquisition of gender stereotypes? What effect, for example, does the reader have on the younger child's learning of gender, as opposed to older children who are able to read and interpret books/illustrations on their own? Furthermore, who are the judges of "the best" in children's literature? What is the selection criteria used in order to make these determinations of award-winning status? Finally, many studies, including the present one, have examined award-winning books for sexist content. In choosing this type of sample, an assumption is made that the award-winning books represent those which are typically read to children (based on anecdotal accounts of parents, librarians, preschool teachers, and bookstore employees) . In order to ensure generalizability of findings, future research should address whether these books are accurate representations of books that are popular among children.
According to    I have been asked to take part in the research project described below . The researcher will explain the project to me in detail. I should feel free to ask questions . If I have more questions later , Diane Turner-Bowker , the said person responsible for this study [(401)874-2193)] , will discuss them with me.
I have been asked to take part in a study which will investigate how women and men are presented in pictures taken from children's literature . Specifically , I will evaluate the physical positioning of persons in pictures to determine similarities/ differences in the placement of women and men. I will be asked to view some pictures and answer a series of questions; it will take approximately 80 minutes to complete . The research will take place during one regular class meeting time.
Remaining class time will be dedicated to a debriefing session , allowing for questions and answers .
I will receive class credit for participating in this research project. The researcher will inform my instructor of my participation and the amount of credit awarded will be determined by my instructor.
My part in this study is strictly confidential. None of the information will identify me by name . The decision whether or not to take part in this study is up to me. do not have to participate. If I decide to take part in the study, I may quit at any time . The purpose of this study is to examine the placement of persons in relation to one another in a series of pictures. In taking this test, you will need to answer a series of questions for each picture you see.
Each picture will be represented by a number , as indicated on the top of each survey form . You will need to answer a total of 18 questions for each picture. Some of the questions have more than one section for you to respond to.
Overall, there are two types of questions that you may encounter : The first (#1) will ask you to rate each person in the picture as traditional versus nontraditional , and as active versus passive.
If the girl/woman is presented inside of the home, in the kitchen, cleaning, cooking, caring for children, playing with dolls , or otherwise engaged in domestic work or play, you should check that she is presented in a traditional role. If the girl/woman is presented outside of the home, in a business, working on the car , playing with trucks , engaged in nondomestic activities, etc. you should check that she is presented in a nontraditional role.
If the boy/man is presented outside of the home, in a business , working in the car, playing with trucks, engaged in nondomestic activities, etc . you should check that he is presented in a traditional role. If the boy/man is presented inside of the home, in the kitchen , cooking , caring for children, playing with dolls , or otherwise engaged in domestic work or play, you should check that he is presented in a nontraditional role.
If you cannot tell the role of the girl/woman or boy/man from looking at the picture , check the box labeled not sure.
If the girl/woman appears to be in motion or involved in a task, you should check that she is presented in an active role. If the girl/woman appears to be motionless , inert, or the recipient of the actions of others , you should check that she is presented in a passive role.
If the boy/man appears to be in motion or involved in a task, you should check that he is presented in an active role. If the boy/man appears to be motionless, inert, or the recipient of the actions of others , you should check that he is presented in a passive role.

Set II
The next series of questions (#2-18) will ask you to respond "yes" or "no" to each statement. The questions may be very specific about the body placement of each person in the picture. Some of these questions will be one-part, for example: Is the girl/woman's body bent over?
yes no

□ □
Here, you will answer the one-part question by simply placing a check mark in the the box labeled "yes" or the box labeled "no".
Other questions may have more than one part or section to complete , for example: Is the boy/man using a(n) ______ on the girl/woman?

□ □ □
Here, you must choose "yes" or "no" for each section of the question individually. Therefore , you must place a check mark in the box labeled "yes" or the box labeled "no" for arm-lock . You must place a check mark in the box labeled "yes" or the box labeled "no" for shoulder-hold. And , finally , you must place a check mark in the box labeled "yes" or the box labeled "no" for hand-hold .
You will receive training from the researcher on how to identify each body position by viewing a sample of pictures which clearly identify each category .
Please work at a steady pace through the test, as you will be viewing and rating 40 pictures.
Picture# ------ I have been asked to take part in a study which will investigate how women and men are presented in pictures taken from children's literature. Specifically , I will evaluate the physical positioning of persons in pictures to determine similarities/ differences in the placement of women and men. I will be asked to view some pictures and answer a series of questions ; it will take approximately 30 minutes to complete . Afterwards , a debriefing session will occur , with time for questions and answers regarding the study .
Although there is no direct benefit to me for tak ing part in this study, the researcher may learn more about the placement of persons in pictures .
My part in this study is strictly confidential. None of the information will identify me by name. I have been asked to take part in the research project described below . The researcher will explain the project to me in detail. I should feel free to ask questions . If I have more questions later, Diane Turner-Bowker , the said person responsible for this study [(401)874-2193)], will discuss them with me.
I have been asked to take part in a study which will investigate how women and men are presented in pictures taken from children's literature. Specifically, I will evaluate the physical positioning of persons in pictures to determine similarities/ differences in the placement of women and men. I will be asked to view some pictures and answer a series of questions ; it will take approximately 30 minutes to complete. The research will take place during one regular class meeting time .
Remaining class time will be dedicated to a debriefing session, allowing for questions and answers. I will receive class credit for participating in this research project. The researcher will inform my instructor of my participation and the amount of credit awarded will be determined by my instructor.
My part in this study is strictly confidential. None of the information will identify me by name. The decision whether or not to take part in this study is up to me. do not have to participate. If I decide to take part in the study , I may quit at any time.

97
The purpose of this study is to measure the meaning of pictures by having people judge them against a series of descriptive scales . In taking this test, please make your judgments on the basis of what you see in each picture.
Each picture you observe will contain a girl/woman and a boy/man . You will be asked to rate the same picture twice --once for the girl/woman and once for the boy/man. Therefore, you will complete two forms per picture.