A Media Skills Intervention for Adolescents on Gender Attitudes, Beliefs, and Behaviors

This study examined the effects of a media literacy intervention on the gender attitudes, beliefs, and behaviors of a sample of 9 and 11 grade adolescents. As a time of identity development, adolescence can be especially challenging for girls, as they are faced with traditional cultural expectations about gender. It is proposed that the media, particularly television, reinforce traditional ideas about gender for both sexes, including traditional interactions with other gender peers, and may contribute to low social self-esteem in girls, with negative outcomes. A ?-session media literacy workshop, using a pretest/post-test control group design, was implemented with a group of 69 workshop participants and 67 controls from intact high school health classes. Dependent variables included attitudes toward women, traditional interactions with other-gender peers, social self-esteem, use of stereotyped descriptors, and belief in the reality of television interactions. Independent variables included the workshop , gender, and amount of media exposure. There were positive correlations between several of the dependent variables related to traditional gender ideas and behaviors, between traditional interactions and belief in the reality of television interactions, and correlations between low self-esteem and media use. There were no significant differences in change scores between experimental/control groups at either post-test or follow-up for any dependent variable, except use of stereotyped ideal descriptors, which increased for the experimental group, indicating a reverse effect. A reverse effect was also seen with self-esteem scores, especially for boys. Classroom differences were explored, as well as gender differences, and the features of stereotyped descriptors used. Qualitative results indicated overall satisfaction with the workshop, especially for girls and those in the smaller classes. Results were evaluated in terms of various theoretical interpretations of media effects, and the theoretical approach taken to promote attitude change . It is suggested that an inoculation approach may not be as effective with adolescents as one encouraging independent critical analysis, that identity of this workshop as an outside intervention influenced results, and that inclusion of media literacy in the regular school curriculum is warranted.

beneficial to adolescents, particularly those who rely on the media as a primary socializing agent.
Rarely is popular programming discussed in the classroom, as it is not considered serious material for discussion (Roberts, 1993). Critiques of television and its impact are not usually part of the regular school curriculum. It appears logical that "connecting" with adolescents might be easier if a medium is used, such as television or rock music, with which they are most comfortable. Regarding popular programming as cultural material worthy of use as a vehicle for serious consideration of social issues is necessary, when one considers the extreme pervasiveness of television in our society. As Greeson and Williams (1986) note, even psychologists have overlooked the potential influence of the media on adolescent attitudes, beliefs, and behaviors. Greeson (in Greeson & Williams, 1986) reviewed adolescent psychology textbooks recently published and found that over half of them failed to devote a single page to discussing media effects.
Finally, the increasing problems of teen suicide and eating disorders, as well as early sexual behavior and its association with teen pregnancy and childbearing, future sexual victimization, disease, as well as emotional and social repercussions, all warrant serious attention. Efforts aimed at examining the role of gender stereotyping in the media perhaps can empower both boys and girls by heiping them to understand the social pressures that influence their attitudes, beliefs and behaviors regarding heterosexual interactions.  and an especially vulnerable time for girls. Until this time, boys and girls are basically on par in terms of achievement, involvement in activities, and self-esteem. H~wever, as girls become more aware of the culturally scripted roles for women, their sense of worth, their confidence, and their self-esteem have been found to drop in comparison to boys (Gilligan, 1990;Pipher, 1994). Compared with boys, girls express lower expectations for their abilities and give more internal attributions for failures and more_ external attributions for success (Sadker & Sadker, 1994). It has been widely acknowledged, particularly as it relates to academic and social outcomes, that the drop in self-esteem is due to the cultural assignment of a system of gender stereotypes that define women as less capable and less valuable than men. In addition to diminished self-esteem, achievement, and expectations for success, the adoption of gender stereotypes has been linked to a variety of other negative outcomes for adolescent girls. These include depression, suicide attempts, poor body image, eating disorders, early sexual behavior, sexual victimization, and teen pregnancy. Gender stereotypes encourage boys to be aggressive and competitive, and can lead to constricted emotionality and lack of emotional intimacy with others, often resulting in demeaning attitudes and behaviors toward girls, including harassment and sexual or physical aggression.
It is well recognized that the mass media are purveyors of traditional gender stereotypes and are also a major source of role models for adolescents. According to cultivation theory, heavy media users will come to believe and uphold what they are exposed to as cultural norms (Gerbner, Gross, Morgan, & Signorielli, 1986), and will, according to social learning theory (Bandura, 1977), begin to model those behaviors.
Amount of exposure to television has been positively correlated with the endorsement by both boys and girls of traditional gender beliefs and lower achievement in school.
Exposure to sexually explicit material has been linked to increased sexual activity by both adolescent boys and girls. Characteristics that appear to be linked with later onset of sexual activity are less traditional gender beliefs, high self-esteem, and the greater influence of the family over the media as a primary socializing agent.
However, messages from family and other personal sources are too :frequently overridden by pervasive messages from the media that adopting traditional gender roles, including engaging in sexual activity, is associated with attractiveness, popularity, success, and an exciting and desirable lifestyle.
This study examines the effects of critical viewing skill training on the gender attitudes, beliefs, and behaviors of a sample of 9th and 11th-grade adolescents. A seven-session workshop focused on the general negative effects of gender stereotyping for both boys and girls, on self-esteem, on intergender relationships, and on features of the media that encourage gender stereotyping and misconceptions about heterosexual relationships. Anticipated outcomes included enhanced social self-esteem in girls, less traditional gender attitudes and beliefs, less stereotyped ways of describing ideal males and females, and less traditional girl-boy interactions.

Review of Relevant Literature Adolescent Girls
Gender Identity and Self-Esteem. Identity is the process by which the individual comes to take a place in the social world (Marcia, 1980). As such, identity development represents an interface between the individual and society. As an emergent, exploratory task of adolescence, developing a sense of self in the world requires, in part, deciphering social messages about the choices and values the culture provides. The cultural choices and values about gender are highly salient during this time, and developing a gender identity includes an acceptance of one's gender as a social-psychological construction that parallels acceptance of one's biological sex (Spence, 1984in Bieri & Bingham, 1994. Since young adolescents are typically in the identity statuses of diffusion or foreclosure, having not yet entered a stage of active exploration of values (Marcia, 1980), the cultural messages that contribute to the social-psychological construction of gender are more likely assimilated than critiqued. At this time, as gender identity becomes salient, the flexible behavior that both genders engage in during childhood is typically replaced by the more rigid role behaviors deemed appropriate for each sex.
Many girls recognize that the roles assigned to them are less valued than the roles assigned to boys, affecting their gender self-concept, and leading to significant negative impacts on self-esteem (Bieri & Bingham, 1994;Gilligan, Lyons, & Hammer, 1990;Pipher, 1994;Sadker & Sadker, 1994).
While identity, or self-concept, is concerned with a set of beliefs about the self, self-esteem provides the positive or negative evaluative components of that identity (Coopersmith, 1967). Once considered a global construct, it is now conceptualized as being arena or dimension-specific, and can, for example differ in home, peer, or school contexts (Brehm & Kassin, 1993). Peggy Orenstein (1994), in her research on girls in middle school, used a two-dimensional definition of self-esteem: how a person views her performance in areas in which success is important to her, and how a person believes she is perceived by significant others, such as parents, teachers and peers. A nationwide survey conducted in 1990 by the American Association of University Women (AAUW, 1991) found that only 29% of high school girls felt "happy the way I am," compared to 46% of high school boys, 60% of elementary school girls, and 69% of elementary school boys.
Gender Stereotyping and Self Esteem. The belief that women and men should conform to different cultural standards of appropriate behavior, has shown a high degree of correlation with low self-esteem in girls, though not in boys (AAUW, 1992). Simmons and Rosenberg (1975), for example, found that adolescent girls who were low in self-esteem had significantly less positive attitudes toward women. Galambos, Petersen, Richards, and Gitelson (1985) also found that measures of low self-esteem were significantly correlated with a measure of traditional attitudes toward women.
These authors note that the combination of low self-image and more traditional attitudes will limit the range of an adolescent girl's pursuit of educational and occupational opportunities . For example, the AAUW (1992) has documented the ways in which sexism in the school system "shortchanges" girls, leading to underachievement in secondary school years. Although girls consistently receive grades that are as high or higher than boys throughout high school, they begin to show significant drops in achievement test scores and loss of interest in pursuing math and science careers.
In addition to school achievement, low-self esteem in girls can have serious personal consequences, such as problems related to poor body image and an overemphasis on sexuality. For example, Holmbeck, Crossman, Mandrel, and Gasiewski (1994) surveyed 300 students aged 14 to 19 years and found a significant positive relationship between knowledge and use of contraceptives and self-esteem in girls.
Although the boys rated higher in self-esteem than the girls, self-esteem was only predictive of contraceptive use in girls.
Gender Stereotyping and Sexuality. Adolescence marks the time when girls and boys begin interacting in sexualized ways. As Lott (1994) notes, enhancing one's attractiveness and searching for a boyfriend appear to be the major developmental tasks prescribed for the adolescent girl in our majority culture, whereas the adolescent boy's primary task is to develop skills to pursue his life goals and to resolve issues of identity. As girls' bodies mature, they become increasingly and uncomfortably aware of how critical their bodies are to their identities (Striegel-Moore, Silberstein, & Rodin, 1986). Lott (1994) notes that the top priorities for girls in our majority culture are being popular (attained by being attractive, fashionable, friendly and available), and being feminine ( attained by behaviors that are predominantly dependent, deferential, flirtatious, and nurturing). These attributes are encouraged by television programs featuring adolescent girls as well as by popular teen magazines, such as

Seventeen.
Included in stereotyped constructs is an emphasis on sexuality and the pressure to engage in early sexual behavior. Adolescents of both genders are engaging in early sexual behavior, defined for the present purposes as coital activity before the age of 16, at an increasing rate, with the years from 12 to 20 as the time when most girls today become sexually active. It is estimated that between 50% to over 70% of African American and European American women will experience coitus by the age of 19 (Wattleton, 1987;Wyatt, 1989). The number of 15-year-olds with sexual experience increased from 17% in 198017% in to 26% in 198817% in (AAUW, 1992. Kalof (1995) surveyed 288 Ohio adolescents between the ages of 13 and 19 and reported that 26% had experienced intercourse, with an average age of first intercourse for boys of 13.7 years, and an average age for girls of 15 .1.
As girls are confronted with a culture that both idealizes and exploits their sexuality, they may be left with the realization that it is crucial to their identity. Kalof (1995) notes that for some women sexual activity might not only be expected behavior but may constitute a form of power since they learn that their self worth is dependent on their sexual attractiveness. Kalof found that participation in unwanted sexual activity was far less for African-American adolescent girls than for boys or for European-American girls, and the most influential factor was more egalitarian gender beliefs among African-American girls. She proposes that possessing other attributes of power, including egalitarian gender attitudes, confidence with members of the other gender, and being popular with peers, might replace sexual activity as the primary source for feeling valued in social relationships.
Several other investigators have found a positive correlation between acceptance of gender stereotyping and early sexual behavior (Gibbs, 1986;Ireson, 1984;Scott-Jones & White, 1990). Foshee and Bauman (1992), in a survey of 1,607 African American and European American 12-to 14-year-olds, found a relationship between stereotyped beliefs about women and sexual behavior on the part of girls, but not boys.
Girls with traditional beliefs were more likely to begin having sexual intercourse earlier than girls with nontraditional beliefs about gender.
Negative Effects of Early Sexual Behavior. The negative consequences for adolescent girls of earlier sexual intercourse are due in large part to their limited knowledge, experieiice, and social and cognitive development. Only one in three adolescents use a regular form of contraception (Wattleton, 1987), which places them at increased risk for sexually transmitted disease, AIDS, and pregnancy. The United States has the highest rate of teenage childbearing in the Western industrialized world, with over half a million babies born to women under 20 years of age (AAUW, 1992).
There is also an association between early sexual behavior and the likelihood of future sexual victimization (Gidycz, Hanson, & Layaman, 1995;Himelein, 1995;Koss, 1985). Himelein (1995) suggests that the tendency for an adolescent girl with low selfesteem to internalize blame will hamper her ability to be assertive during a dating situation in which there is pressure to engage in sex. If she is surrounded by cultural messages suggesting that she be sexually attractive and available in order to establish or maintain intimacy, the probability of sexual victimization (whether she labels it so or not) increases. Sexual victimization may, in turn, lead to a downward spiral of increased feelings of unworthiness, and repeated involvement in abusive relationships.
As Koss and Burkhart (1989) note, the adolescent girl is particularly sensitive to social mores, and may be most susceptible to rape mythology that lays the blame for rape on the victim. Thus, cultural expectations, many of which are portrayed in the media and are reflected back to the adolescent, are part of the process linking low self-esteem and sexual behavior in adolescent girls.
Low self-esteem has been identified as a factor in teenage pregnancy (Patton, 1981 ). A connection between gender stereotyping and teenage pregnancy has also been found, with those holding more traditional views less likely to use regular contraception than those with more liberal gender views (Philpps-Y onas, 1980). In 1988, 34% of all women under the age of20 who had babies were married , compared to 52% in 198052% in (AAUW, 1992. Plotnick and Butler (1991) studied the relationships among self-esteem, attitudes towards women, and nonmarital childbearing by surveying 1,184 girls at 14-15 years, and again at 19 years. By age 19, 16.9% had a nonmarital child, and these were girls who, 4 years earlier, had lower self-esteem, less positive attitudes toward school, and more traditional beliefs about family and gender roles. The AAUW (1992) notes that the pervasiveness of sexism and gender stereotyping in our culture has verbally honored the role of mother and may offer an escape route for some teens who are living under social or economic deprivation.
These young mothers are beginning their adult lives prematurely, often burdened not only with a child and the hardships that entails, but with low self-esteem and a traditional and restricted view of what roles and opportunities are available to them in their future.

The Media
Gender Stereotyping in the Media . A primary source for the reinforcement of gender stereotyping is the mass media. Television, especially, has taken over many of the socializing roles previously held by the family, the school, and the community, and the average American adolescent will spend one and a halftimes as many hours watching television as hours spent in school (Liebert, Sprafkin, & Davidson, 1982).
Commonly referred to as the "window to the world," the television set is turned on for over 7 hours each day in the average American home (Signorielli, 1993) , and, as Kubey (1998) notes, television will consume about one-fifth of an average American's waking life. Although television viewing decreases somewhat in later adolescence (Lloyd-Kolkin, Wheeler, & Strand, 1980), it is still the primary source of media exposure. Consumption ofradio (Larson, Kubey, & Colletti, 1989;Lont, 1990), music television (Sun & Lull, 1986), and popular magazines (Peirce, 1990) also remains high, and the latter two, especially, have been shown to represent concentrated amounts of stereotyped portrayals of women and men.
There is an extensive literature on gender stereotyping in the media. Many studies have shown that role models in the media support dominant cultural ideals of feminine beauty and traditional gender expectations. These investigations cover the print media, especially popular magazines (Goffinan, 1976;Kang, 1997), film (Walkerdine, 1997), radio (Hurtz & Durkin, 1997), and, more recently, interactive video (Gailey, 1996), but television has received the most intense scrutiny (Bretl & Cantor, 1988;Downs & Harrison, 1985;Iijima Hall & Crum, 1994;Herrett-Skjellum & Allen, 1996;Kolbe & Langefeld, 1993;Lovdal, 1989). Widespread interest in this issue began in the 1970s and, as Browne (1998a) notes, much of the evidence is now dated. However, recent analyses have yielded similar results to earlier ones. In spite of an increase of women in more contemporary, independent roles, the overall portrayal of gender stereotypes has changed little in 20 years (Browne, 1998a;Davis, 1990;Kang, 1997;Signorielli, 1993).
Prime-time TV. Davis (1990) reviewed 50 hours of prime-time network programming during 1987 and found that the portrayal of the young, sexy, attractive woman who is more ornamental than functional is still a prevalent representation.
Men characters appear 2 or 3 times more frequently than women, are older, more active problem-solvers, more powerfui higher achievers, and are portrayed as more easily combining work and family (Signorielli, 1989). Women are less likely to be seen working outside the home, unless they are divorced or not married, or are seen in stereotyped occupations. Most are younger than the men and are more attractive and more nurturing (Signorielli, 1989;. A review of 1997 Saturday morning cartoon offerings across four major television channels revealed not a single lead female character (Silver, unpublished observation).
As a cultural barometer, television both reflects and directs mainstream cultural norms. It has been shown repeatedly that the more one is exposed to gender stereotyping in the media, the more one will endorse stereotyped views (Durkin, 1985; Gerbner, Gross, Morgan, & Signorielli, 1986;Morgan, 1982). In a meta-analysis of major studies from 1974, Herret-Skjellum and Allen (1996 found that as the amount of television exposure increased, so did the acceptance of stereotypical images of women. The survey data indicated that traditional occupational choice was the most significant indicator of stereotyping related to media exposure. Signorielli and Lears (1992) found that among a large, ethnically diverse sample of 4th and 5th graders, the amount of television watched was positively correlated with stereotyped beliefs about the different chores men and women should do. McGhee and Frueh (1980) divided 64 children from grades 1,3,5, and 7 into heavy TV viewers (25 or more hours a week) and light TV viewers (10 hours or less). Heavy viewers were found to have more stereotyped views than light viewers. Also, light viewers showed a downward progression of stereotyping with age, while heavy viewers showed a steady increase.
The connection between exposure to media stereotypes and aggression against women has the most serious negative consequences for women. Research in this area has focused on film (Malamuth & Donnerstein, 1983) and music television, but Brinson (1992) reviewed prime-time television dramas for the presence o±: and opposition to, rape myths, which are misconceptions about the motivation and culpability for rape (Burt, 1980). These myths include "she asked for it," "she wanted it," "she lied about it," and "she wasn't really hurt by it." Brinson found that rape myths were pervasive in television, reflecting a cultural attitude of acceptance.
Twenty-six storylines about sexual assault were collected in a purposive sample from prime-time dramatic programs; content analyses revealed that, overall, rape myths were used an average of 5.08 times per storyline and were opposed by a TV character only 3.27 times per storyline. Brinson argues that one effect of watching this type of programming is that it reinforces the cultural belief that women are somehow responsible for rape. These beliefs include a set of behavioral expectations for women that include ways to dress, look, and behave. A failure to conform to these expectations may result in the 'just" punishment of rape, shifting responsibility for the attack from the rapist to the victim.
Commercial TV Advertising. Like prime-time programming, commercials are also sex-typed and, in spite of superficial changes, have changed little in their basic messages about men's and women's roles (Hurtz & Durkin, 1997;Lovdal, 1989).
Although women and men are now seen equally often in commercials, women still tend to represent home products and promote attractiveness and beauty. Over 90% of the narrators, the "voices of authority," are men (Bretl & Cantor, 1988;Downs & Harrison, 1985). Brown (1998a) reviewed American and Australian advertisements shown during Saturday morning children's programming, and found that commercial advertising for children also portrayed stereotyped sex roles. Boys were portrayed as more knowledgeable, active, aggressive, and instrumental than were girls. Also, nonverbal behaviors involving dominance and control were associated with more boys than girls.
According to Iijima Hall and Crum (1994) , women are more likely to be dehumanized and portrayed for their physical appearance alone. These researchers (1994) found a preponderance of "body-isms" or the portrayal of parts of women's bodies during beer commercials, which are shown excessively during sporting events, suggesting an association between male aggression/competition, alcohol, and the sexualization of women.
Music Television. Music television, which has become a major cultural phenomenon over the past two decades, is watched by 75% to 90% of adolescents (Greeson & Williams, 1986;Sun & Lull, 1986), and has been found by many researchers to be highly gender stereotyped, very often in combination with images of violence (Jhally, 1993;Sherman & Dominick, 1986). As Sut Jhally (1995) notes in his video expose on the harmful combination of sex and violence on music television, these subtle and pervasive messages serve to reify patriarchal ideas about men as dominant and aggressive, and women as submissive and sexual. Men are generally depicted as powerful, in control, hostile, and sexually aggressive, whereas women are portrayed as passive, submissive, and highly sexualized (Brown & Campbell, 1986).
For example, a song from the recording artist Prince contains the lyrics, "I knew a girl named Nikki. I guess you could say she was a sex fiend. I met her in a hotel lobby masturbating with a magazine" (cf. Greeson & Williams, 1986).
In an effort to provide a gender role analysis of music television, Sommers-Flanagan, Sommers-Flanagan, and Davis (1993) analyzed 40 music videos and found the following themes: men appeared nearly twice as often as women; men were more dominant and aggressive than women; women engaged in more implicitly sexual and subservient behavior; and women were more often the object of both explicit and implicit sexual advances. The authors conclude that the most salient message on music television for both genders is to act sexual.
Music videos have been found to promote gender stereotypes , aggressive attitudes towards women, and sexual behavior. Hansen and Hansen (1988), for example, investigated the priming effect of music videos on the activation of gender schemas. College participants who were exposed to a sample of stereotypic music videos later interpreted a male-female interaction scenario more stereotypically than those not exposed to the music videos. Greeson and Williams (1986) studied a sample of thirty-four 10 th graders and thirty 7 th graders and found that adolescents who were exposed to less than an hour of music videos at home were more likely to approve of premarital sex than those not exposed to the videos. Barongan and Hall (1995) exposed 27 men to misogynous rap music and exposed 27 men to neutral rap music. Afterward, men in the first group were more likely to show an assaultive film vignette to a female confederate in a laboratory situation, and were more likely to report the confederate as more upset and uncomfortable than men previously exposed to neutral music. The authors concluded that exposure to misogynous themes may promote sexual aggression against women.
They suggest that cognitive distortions about the social acceptability of sexual aggression may occur in men who are exposed to a high level of misogynous themes.
Women with high levels of media exposure to sexualized violence may also be more tolerant of aggressive sexual advances, either due to cultivation effects ( discussed below) or the low self-image that misogynous rap music might promote.
Misogynist themes are also present in heavy metal music, which involves harddriving, frenetic , and intense sensations, and is popular with many adolescents. Arnett (1991) surveyed 15-to 19-year-old adolescents to explore a connection between exposure to heavy metal music and reckless behavior. For girls, those who liked heavy metal music and had lower general self-esteem were significantly more likely to engage in sex without contraception than those who did not. For boys, those who liked heavy metal music, regardless of self-esteem levels, were more likely to engage in casual sex than those who did not. Both boys and girls who liked heavy metal music were more likely to express confidence about dating and sexuality. The author suggests that more research be done to determine the direction of causality among low self-esteem in girls, sexual behavior, and exposure to the sexist themes present in heavy metal music.
Television Role Models and Self-Esteem. Considering the co-occurrence of sexrole stereotyping and low general self-esteem in girls and the prevalence of gender stereotyping in the media, it is not surprising to find a link between low self-esteem and amount of television watched. Martin and Gentry (1997) found that when adolescent and preadolescent girls compared themselves with models in advertisements, it affected their self-perceptions of attractiveness and sometimes led to lower scores on measures of self-esteem. Duck (1990) found that low self-esteem children identified more with TV characters as ideals and were more likely to perceive them as real than children with high self-esteem. Duck (1990) reports on an early study by Campbell (1962) that investigated the source of ideal role models in adolescents. This study surveyed over 600 adolescents before the introduction of TV and then 3 years later and found that parents and parent surrogates dropped in prominence as role models (from 25% to 18%), and glamorous adults rose in prominence from 7% to 29%. By 1988, up to 75% of313 children surveyed in the 5th through the 9th grade chose media ideals over people they personally knew (Duck, 1990 Broen, Childers, and Waszak (1990) reviewed the literature on sexuality and television and concluded that the frequency of sexual references has increased in the last decade and are increasingly explicit. They indicate that adolescents who depend on television as a primary source of information about sexuality will place more importance on female beauty and will be more likely to endorse premarital intercourse with multiple partners. Brown and Newcomer (1991) reviewed research by Greenberg and colleagues that shows an increase of 103% between 1980 and 1985 in references to, and depictions of: sexual activity on soap operas popular with adolescents.
There is significant evidence that exposure to sexually explicit material, especially when associated with violence, can influence judgments of sexual scenarios as well as sexual behavior. Witnessing frequent associations of sex and violence, which occurs in increasing proportions in prime-time programming and music television, has been found to increase tolerance for, and engagement in, sexual aggression (e.g., Mullin & Linz, 1995). Cook (1995) found that although a sample of 546 college students generally did not accept aggressive sexual behavior in a dating situation, many women expected it. The author interprets these findings as suggesting that expectations may make women less effective or less willing to ward off sexually aggressive advances. If adolescent girls are cultivated by television and other media sources into expecting a high level of sexual activity, pressure, or aggression in dating relationships, they may be more likely to engage in early sexual behavior.
The media tend to portray sexual activity as risk-free (Brown & Newcomer, 1991). Unmarried heterosexual couples on television engage in sexual intercourse as much as 8 times more frequently than married couples. Also, contraceptives, pregnancy, and sexually transmitted diseases are rarely mentioned. As Strouse and Fabes (1986) note, sexuality on television is shown predominantly as a carefree, exploitative, recreation-oriented activity reflecting uncontrollable attraction between non-married, highly attractive persons. They note the increase in popularity of daytime soap opera for adolescents, a medium that is highly sexualized and presents marital infidelity as a common family problem. Brown and Newcomer (1991) conducted a correlational study in an attempt to link exposure to TV with sexual behavior. Independent raters judged the sexual content of popular shows watched by adolescents. A group of391 adolescents were surveyed for their TV use, and the proportion of "sexy" programming exposure relative to total TV exposure was determined. The authors found that both boys and girls who watched more "sexy" TV programming were more likely than those who chose less "sexy" content to have had sexual intercourse. The adolescents were interviewed in 3 waves, the fall of 1978, the fall of 1979, and the spring of 1981. By the third interview, over a third of the European American girls and over 40% of the European American boys and African American girls had had sexual intercourse.
There was no relationship between total amount of viewing or total amount of sexy program exposure to sexual behavior, but there was a significant relationship between the proportion of sexy programming watched and such behavior. This relationship existed regardless of social class, pubertal development, or friends' influences.
Theoretical Frameworks. Brown and Newcomer's (1991) study lends some evidence to a causal pathway that goes from media exposure to sexual behavior, and thus fits well into a cultivation analysis of media effects. According to cultivation theory, television cultivates common predispositions and preferences that used to be acquired from primary sources, such as the family {Gerbner, Gross, Morgan, & Signorielli, 1986 (Rubin, 1986). Assuming an active viewer stance, the needs and gratifications approach, taking into account individual differences and selective exposure, suggests that programs are chosen which will produce desirable, comforting results to each individual viewer.

Media Literacy
Theoretical Grounding and Current Status. Media literacy has been defined as the process of accessing, critically analyzing, and evaluating media messages, as well as communicating messages using media tools (Hobbs, 1998). Conceptually, media literacy has a broad definition and a wide range of applications, although most programs are in a school-based context. Although initiatives to provide media literacy education have been ongoing in Europe for the past 30 years, the United States is only recently focusing on organized training efforts, and these have been limited to a few statewide initiatives (Kubey, 1998). According to Kubey, the inability to organize a cohesive, national action plan can be attributed to several factors. First, media literacy is relevant to many disciplines, and thus is conceptualized in a wide variety of ways with a wide variety of goals and objectives. Second, unlike other countries, the United States exports more media than it imports, and sees little need to examine its cultural products. Third, school curricula are already overflowing, and both parents and teachers are more interested in other goals for their children, such as computer literacy.
Finally, it is difficult in the culturally and geographically diverse United States to gain consensus on national issues dealing with the socialization and education of children. For example, Kubey (1998) cites successful political efforts in the late 1970s to squash several pilot "critical viewing" educational programs funded by the U.S. government at the preschool through college level. He notes that the split in theoretical approaches to media literacy has prevented any cohesive programmatic development. The cultural studies approach has been taken by European educators since the 1960s, but is barely recognized in this country as an educational alternative.
This approach involves analyzing and appreciating media as a vehicle and representation of popular culture. Instead, the United States has adopted a protectionist, inoculatory approach t4.at focuses on avoiding exposure or rejecting harmful media messages. Based on social learning and cultivation theories, harmful media messages can be internalized by the viewer and modeled, and thus are viewed as risk factors. Learning how to resist those messages through a skill-based media literacy program is seen as a protective process. As yet, however, because social science research has not demonstrated the external validity of inoculatory approaches outside the laboratory, there has been resistance from administrators, teachers, and parents. This has contributed to the slow development of a curriculum that deals with anything more than technical aspects of media production (Kubey, 1998).
Progress has been made in the last 5 years in incorporating media literacy programs into school curricula, though not on a national level. At the 1993 Media Literacy National Conference (Aufderheide, 1997), the following key ideas were agreed upon: 1. media messages are constructed; 2. media messages are produced within economic, social, political, historical, and aesthetic contexts; 3. the interpretative meaning-making processes involved in message reception consist of an interaction between the reader, the text and the culture; 4. media representations play a role in people's understanding of social reality.
Media Literacy Education in the School Setting. Before 1991, at least 32 major critical viewing skills projects had been developed in the United States (Brown, 1991), and the 1990s has seen increased growth in this area. Most programs are exploratory or entrepreneurial in nature, and only a few have reported measured effects. These projects are large and broad based, usually implemented in the classroom, and they cover a wide array of topics. Most focus on the negative cognitive and social effects of television exposure, and are inoculatory in approach, as students are taught to evaluate and reject negative messages present in the media (Masterman, 1997).
The diversity of approaches to media literacy has thwarted widespread implementation of programs. Hobbs (1998) Linz, Wilson, and Donnerstein (1992) suggest that inoculation messages are most effective when they offer warnings about decreased sensitivity to the material and the adoption of negative attitudes. These authors, concerned about the effects ofR-rated movies and televised sexualized violence on adolescents, have recommended strategies for the development of media literacy programs. These include considering the unique developmental characteristics of adolescents, methodological issues, and the need to evaluate the long-term effects of the intervention.
From a content perspective, violence on television is more often a focus than are interpersonal interactions, including sexual activity. Gender stereotyping has typically been embedded in a larger agenda rather than as a distinct topic. Common goals of these programs are reflected in the goals of a program developed for teenagers by Lloyd-Kolkin and colleagues (1980) in response to a request from the U.S. Department of Education in 1978. These goals were: 1) to be able to evaluate and manage one's own television viewing behavior; 2) to question the reality of television programs; 3) to recognize the persuasive arguments and messages on TV and to be able to counterargue; and 4) to recognize the effects of television on one's own life. Johnston and Ettema (1982) developed a television program, Freestyle, specifically designed to confront gender stereotypes, particularly as they affect occupational choices. The stated goal of the project was to" expand the career awareness of 4th to 6th graders by making sex and ethnicity less significant predictors of pre-occupational knowledge, interest and preferences." The format consisted of a series of 30-minute dramas that utilized young adolescents in situations that highlighted role choices and challenges. Geared to a younger audience, there was no direct focus on heterosexual interactions. Funded by the National Institute of Education, Freestyle was shown in classrooms to over 6000 students across the United States. The results of the intervention were positive for the most part, and reflect the benefits of using television pro-socially as a learning tool.
Rocket and McMinn (1990) describe a slide presentation program about commercial advertising's promotion of unrealistic standards of thinness and attractiveness. Although the authors present only anecdotal results, they claim the program was highly successful. They attribute its success to the utilization of a popular media source (magazines) and the focus on broad societal factors that contribute to eating disorders and women's preoccupation with body image.
Connecting Media Literacy, Girls, and Gender Stereotyping. Dreyer (1994) notes that the massive educational reform movement begun in the 1980s has not taken seriously the psychosocial needs of children and adolescents, particularly when it comes to identity development, and the development of self-esteem. Because of the increasingly diverse needs and profiles of American families, both economically and culturally, it is becoming increasingly clear, according to many (Holtzman, 1995;Shamai & Coambs, 1992;Soriano & Soriano, 1994) that the public school setting must evolve into a place providing integrated social and psychological support services. This "school of the future" (Holtzman, 1995) will recognize culture and the larger forces at work in shaping educational and life success. In addition, recent focus on the particular psychosocial needs of both European American and minority adolescent girls, has prompted an array of school-based programs, such as Sistahs and Girls Can! (AAUW, 1993). Shamai and Coambs (1992) report on three inoculatory school-based programs that were essentially unsuccessful in changing attitudes and beliefs, one of which was focused on gender stereotyping. The authors suggest that the school, as an autonomous institution, cannot effect attitudinal change unless the society changes, since educational intervention programs typically conflict with the dominant values and messages of the students' culture. Since the media are sources of conflicting messages, Shamai and Coambs suggest that a holistic approach that includes other sources of influence might be more effective.
Application of this model to media literacy efforts has not been documented, but appears to hold promise as a teaching strategy. According to ELM, two routes to attitude change are possible: central route and peripheral route, the difference being in the amount of thoughtful consideration that occurs in response to persuasive communication. Peripheral route attitude change occurs without much cognitive assessment, is usually in response to a pleasing cue, and is usually an immediate, affectively-based, short-term route to attitude change. Attitude change of this type occurs quickly, but is not long-lasting and is vulnerable to conflicting information.
Mass media advertising is an example of periph~ral route processing. Central route processing requires significant cognitive elaboration and is more difficult to achieve, but lasts longer, is more resistant to counter persuasion, and affects behavior more (Brooks-Harris, et al, 1996). Unlike peripheral route change, central route change requires motivation to process the information, ability to process the information, and a preponderance of positive thoughts about the information.

The Present Study
The following conclusions were drawn on the basis of the existing literature: 1. Adolescent girls are particularly vulnerable to cultural messages that they are less valuable or capable than boys, and that much of their identity is connected to their attractiveness.
2. Many of these messages promote low self-esteem and traditional interactions with boys and men, particularly as they relate to sexual behavior, with negative consequences.
3. A powerful source of these messages and role models come from the media, in particular television, which promote a portrayal of men and women in traditional gender roles. This includes an emphasis on sexuality as a primary focus of interpersonal relationships.
4. Through mechanisms related to cultivation, needs and gratifications, and social learning, many girls assimilate and accept these media messages.
5. Media literacy education is an emerging field, and will likely be incorporated into educational curricula on a widespread basis in the near future.
6. Media literacy education is therefore a potential vehicle through which adolescents can be taught how to interpret and critique cultural messages about gender. It is proposed that lower self-esteem in girls, as it relates to social interactions with others, and the adoption of traditional gender attitudes, beliefs, and behaviors are connected, and that this combination makes girls vulnerable to engaging in traditional interactions with boys, which can potentially culminate in early sexual behavior.
Because it is widely recognized that: the mass media are among the most influential sources of gender stereotyping; that both genders are in need of support and guidance as they confront gender norms; and that media literacy educa,tion in the school setting has relatively untested potential for contributing to that support, a media literacy pilot intervention was developed. This intervention focuses on the effects of gender stereotyping in television on self-esteem and on social interactions with other-gender peers.
The present intervention incorporated the recommendations of Linz, et al. (1992) for tailoring a media literacy program to adolescents: utilizing materials that adolescents like and readily identify with; using measures specifically designed for adolescents; conducting the intervention in a school setting; and utilizing a 3-month follow-up to determine longer-term effects. The theoretical approach is based primarily on social learning theory and the cultivation model. A combination of methodologies was employed that encouraged an analysis of television as a vehicle of popular culture, that integrated inoculation methods of instruction with a cultural studies perspective. Teaching tenets from the Elaboration Likelihood Model, focusing on central route processing, informed the strategies that were used.
Although the study targets girls, boys also are adversely affected by gender stereotyping and their attitudes and responses to a media literacy program, especially one that confronts gender norms, are important. For this reason, and for practical purposes in dealing with structured school schedules, both boys and girls were included in the intervention.
The major dependent variables were: social self-esteem, attitudes towards women, traditional girl-boy interactions, and choice of ideal characteristics and television characters. The major independent variables were a workshop intervention, amount of previous media exposure, and gender.
The following hypotheses were tested: 1. There will be positive correlations among pre-intervention levels of the following variables: • traditional attitudes toward women • traditional girl-boy interactions • gender-stereotyped ideal traits • gender-stereotyped television characters • amount of television watched • belief in the degree to which television shows are realistic, and • for girls, there will be negative correlations between pre-intervention measures of the above variables and social self-esteem.
2. On the pretest measures, those with high exposure to media sources will endorse significantly more of the following than low media users: • traditional attitudes toward women • traditional girl-boy interactions • gender-stereotyped ideal traits · • gender stereotyped television characters • belief in the degree to which television shows are realistic, and • for girls, lower social self-esteem.

Workshop participants will show a greater change between pretest and
post-test measures than control participants. They will: • endorse significantly fewer traditional attitudes towards women, • choose less gender stereotyped ideal TV characters, • choose less gender stereotypic traits, and • believe less in the degree to which television shows are realistic.

Workshop participants will show a greater change between pretest and
follow-up measures than control participants for the variables listed in #3 above, as well as the following: • engage in fewer traditional girl-boy interactions, and • for girls, score higher on social self-esteem measure.
5. Workshop participants with high media exposure will show greater change for the variables listed in #4 above than those with low media exposure.
6. Girls will show greater change than boys will for the variables listed in #4 above.

Participants
The original sample of subjects was 149 high school students in 9 th grade and in 11 th grade at Westerly High School, Westerly, Rhode Island during the spring of 1997. All participants were from intact gym classes. All 9 th and 11 th grade students at this school are required to take a 6-week term of health class during their normal gym period, and each gym class is randomly assigned to a particular term. Of the seven gym classes that had health in the spring term, four were chosen as the experimental classes (N = 75, 31 boys and 44 girls), and three were chosen for the control group (N = 74, 41 boys and 33 girls). There were two 11 th grade classrooms, one experimental and one control. The 9 th grade classes were placed into experimental or control groups based on size, with the goal of obtaining equal sample sizes. All classes were lead by one of two teachers, a first-year Physical Education teacher and the Health Department/Physical Education head, who had been at that school for over 20 years.
The size of the classes varied, as can be seen in Table 1.
A pretest, post-test control group design was used, and all subjects received a pre-test, post-test, and follow-up test. The experimental classes received the media literacy workshop developed for this study, while the control classes received their regularly scheduled health class. Prior discussion with the teachers had indicated that the health class syllabus did not include media literacy, sex-roles, or stereotyping as topics.
Thirteen students did not take the pretest and so the following demographic data were collected from only 136 students, or 91.3% of the sample. Ages ranged from 13 to 17, with the average 9 th grade age of 14.8 years, and the average 11 th grade age of 16. 7 years. Ethnically, the sample was 94% European American, 1.3% Native American, 1.3% Asian American, and less than 1 % other ethnicities. Fifty-five percent reported being Catholic, 29% Protestant, and the remaining were of other faiths or did not report one. The average combined education level of parents was · 3.14, on a scale of 1 to 5, with 3 representing some college, and 4 representing college graduation. A family profile question revealed that 48.3% came from intact families, 16.1 % came from single-parented families, and 24.2% came from reconstituted families.
Chi square analyses showed no significant differences between control and experimental subjects on any demographic variable.

Instruments
Demographic and Television Exposure Information. This survey, constructed for the present study, obtained demographic information such as ethnicity, age, gender, family structure, parents' educational level, and religious identification.
Questions on attitudes and behavior regarding television were modeled from those used by Saris (1994). These questions determine amount and content of television watched, parents' viewing habits, restrictions on use, the reality of television, ideal male/female characteristics, and choice of ideal TV characters. A final question asks how much knowledge about heterosexual relationships is gained from television.
This survey was used in the pretest, and a simplified version, including only questions dealing with television reality and ideal character traits, were distributed again at the post-test and follow-up. Appendix A includes the survey, and the method of scoring each item is provided in the answer space for each item.
Attitudes Toward Women Scale for Adolescents (A WSA). This instrument, developed by Galambos, Peterson, Richards, and Gitelson (1985), is a 12-item scale that measures the degree to which traditional attitudes are held toward women. This scale was chosen because of its brevity, its contemporary construction, and because it was designed specifically for adolescents. Seven items are phrased to reflect a traditional view of women such as "Swearing is worse for a girl than for a boy," and "Boys are better leaders than girls." Five items reflect contemporary views of women, such as "On the average, girls are as smart as boys," and "It is all right for a girl to ask a boy out on a date." Eight items deal with adolescent issues and th~ remaining four deal with adult issues. Items are scored on a 4-point Likert scale ranging from 1 = "agree strongly" to 4 = "disagree strongly." Scores can range from 12 to 48; a high score ( after reverse scoring on 5 items) reflects less traditional attitudes.
Reliability assessments indicate that the Cronbach alpha coefficients average . 78 for boys and . 72 for girls, a high level of consistency considering the length of the scale (Galambos et al, 1985). Test-retest data indicate that over a 3-year period, the reliability for girls was .54 (.73 for males) between 6th and 7th grades, and .46 (.60 for males) between 7th and 8th grades. These correlation coefficients may indicate real changes in attitudes given the lengthy interval of time between testing (Beere, 1990).
For girls, the stability between 6th and 8th grades is greater than between 6th and 7th grades, indicating that there may be a temporary change in attitudes for girls in 7th grade.
Concurrent validity with other sex-role inventory scales is high, and the developers of the measure found, as predicted, that boys and those from rural, working-class communities consistently expressed more traditional views. Also, as predicted, more traditional attitudes in girls were correlated with measures of low selfesteem. This scale was distributed at the pretest, post-test , and follow-up sessions.
See Appendix B.
Texas Social Behavior Inventory (TSBI). This scale was developed by Helmreich and Stapp (1974) as a 16-item objective measure of an individual's feelings of social self-worth or social competence. The current measure is a short form of the original 32-item scale, and it correlates .97 with the longer original form. Items are answered on a 5-point Likert scale ranging from "a" = "very much like me" to "e" = "not at all like me." The original phrasing "very characteristic of me" was reworded for the present study to be more understandable to young adolescents and to be consistent with the other scales being employed. Also, the word "people" was qualified to read "people my age" for the same reason. Total scores can range from 16 to 64; higher scores indicate greater social self-esteem.
Items address degree of self-confidence in groups of people, fear of speaking to strangers , and security in social situations. Although the scale appears to also measure social skill, Robinson, Shaver, and Wrightsman (1990) suggest that this scale is probably best used as a measure of self-esteem in social situations or environments.
Alternate-form reliability of the total 32-item scale is .89, and a Cronbach alpha of .92 has been reported. Although no test-retest correlations are available, the present study showed a test-retest correlation of .81. Convergent validity is indicated by a significant relationship to locus of control and performance self-esteem(. 76).
The TSBI correlates .40 with academic self-esteem and .39 with academic social selfesteem Hehnreich and Stapp (1974) reported that the TSBI correlated .81 with "masculinity" for men, and .83 with "masculinity" for women. It correlated .42 with "femininity" for men and .44 with "femininity" for women. "Masculinity" and "femininity" have since been interpreted as "agency" and "expressiveness." This scale was distributed during the pretest, post-test, and follow-up. See Appendix C.

Adolescent Traditional Interactions Scale (ATIS). This scale was developed
for the present study and is designed to assess traditional role behaviors that girls and boys might engage in when interacting with other gender peers. Nineteen statements are answered on a 4-point Likert scale ranging from "a"= "describes me very well" to "d" = "does not describe me at all." Two versions, one for boys and one for girls, was used. Except for one item, the items were identical except for interchanging the words "boy" and girl." Sample items include: " Ifl want a boy(girl) to notice me, I will try to dress in a sexy way around him (her);" and "I act more awkward and self-conscious when I am around popular boys (girls) than when I am around popular girls (boys)." For girls, the statements reflect stereotyped behaviors such as acting in ways that are deferential and passive, that stress physical and sexual attractiveness, and that emphasize the importance of gaining the approval and attention of boys. Higher scores indicate less traditional behaviors for girls. For boys, the statements reflect stereotyped behaviors such as acting in ways that are dominating, a lack of concern about physical and sexual attractiveness, and a lack of concern about gaining the approval of girls. The scale is reversed in values for boys, and higher scores also A total of three ''yes" or "no" responses (1 = "no; 2 = ''yes") were possible, with a total range of scores from 3 to 6. Higher scores reflect higher critical viewing skill development. This measure was taken at the 3-month follow-up only. See Appendix F.
Process Evaluations. This measure, developed for the present study, was a 5question survey measuring class satisfaction, and was distributed and handed in at the end of each-session. The first 3 questions asked for evaluations of how much the student learned, how much they liked the topic and the things done in class.
Responses were on · a 4-point Likert scale from 4 = "a lot" or ''very much" to 1 = "nothing" or "not at all." A range of scores between 3 and 12 was possible, with the higher score indicating more satisfaction. The other 3 questions asked for qualitative comments about what was liked best and least about the workshop session. A modified version that included the same questions about the workshop in general was substituted at the last class, following post-test administration. See Appendix G.
Procedure Pre-intervention. Prior to conducting the study, a focus group of four 9th grade adolescents representative of the sample was formed. In addition to pilot-testing the assessment devices, topics explored included attitudes and beliefs about gender, cross-gender interactions, and media use. The information gleaned was used in developing teaching strategies.
Following acceptance of the project by the school principal, a collaborative meeting was held with the two teachers whose classes were involved, in order to foster acceptance and support for the workshop. The rationale for, and the logistics of, the study were explained, questions were answered, and input for workshop planning was requested by the investigator. Reviewed were the negative consequences of gender stereotyped behavior, especially for girls, and especially as it relates to health concerns, school performance, and interpersonal relationships, as well as how the media contributes to these problems. In addition, the preventative and empowering aspects of skills and awareness training were emphasized.
The supervisory teacher requested that the two teachers describe the workshop briefly to the experimental groups and distribute parent and student consent forms, shown in Appendix H, along with an explanatory letter from the investigator, one month prior to the start of the intervention. The investigator was unable to verify consistency in the initial introduction of the workshop to the students, but was told that it was extremely brief in all cases. Since the emphasis of the intervention was on gender stereotyping in the media and not on sexual behavior, the letter, shown in Appendix I, was carefully worded so as not to mislead parents into interpreting the intervention as sex education. All students were given student consent forms, but were reminded that those whose parents have not consented do not need to complete it.
Students were assured that participation was voluntary, and were given the option of attending the regular health class. All students returned the consent forms, and all participated in the workshop.
Intervention. As is typical for interventions in school settings with intact classrooms, this intervention was a nonequivalent control group design. Although participants could not be randomly assigned to intervention groups, students are randomly assigned to gym classes, providing some degree of external validity. The time span for major media literacy programs typically ranges from a single-day, 6-hour workshop to year-long programs that cover approximately 30 to 50 hours (Brown, 1991). Positive results have been obtained from several shorter programs, such as those developed by Dorothy and Jerome Singer (in Brown, 1991;Singer, Singer, & Zuckerman, 1981), who offered eight 40-minute lessons over a 4- week period to children in grades K through 5. Each lesson covered a different topic, such as action and violence, special effects, commercials, reality of television, stereotypes, etc. The program consisted of a pretest session, 2 lessons a week for 4 weeks, followed by a post-test session. The lessons utilized discussion and activities, homework assignments, and short videos 7 to 11 minutes long to accompany each lesson.
This intervention incorporated many of the Singers' design features and implementation strategies. Because the participants were older than the children in the Singer programs, and because this workshop addressed only one aspect of media literacy, that is, the prevalence, antecedents, and consequences of gender stereotyping, this workshop was of shorter duration. It included 7 46-minute sessions; each class met at the same time of day, and typically twice per week. The design, shown in Table 2 The major teaching strategy involved showing a short, 7 to IO-minute, themebased video clip follo~ed by guided critique, discussion, and relevant activities.
Interactive strategies, such as group discussion, games and role-plays were emphasized to engage students in the process and reinforce learning. Homework focused on evaluating television viewing behavior of the participants and their families, and gender stereotyping in the student's personal world. Students were often asked to engage their parents in some of the homework activities and to share with them what they were learning. At the end of each session, short process evaluations, stapled to the session objectives, were completed and placed anonymously in a large envelope.
The three requirements of the Elaboration Likelihood Model, as used by Heesacker, et al (1996) for affecting attitude change were employed throughout. First, to increase motivation to cognitively process the session material: Second, to increase the students' ability to process the information: Scale, in that order. To ensure anonymity, each student was given an identification number from a class list that was used to match the pretest, post-test, and follow-up measures. The list that matches name and number was not seen by the researcher, but was kept with the class teacher. At each testing session, the teacher reminded the students what their numbers were. The students were assured that the teachers would never see the data and the researcher would not see the list matching name and number.
Following the pretests, a brief introduction to the program was given. For process and impact evaluation purposes, and to foster a sense of empowerment and commitment to the workshop, two students from each of the experimental classes were asked to volunteer to serve as representatives to provide input to the researcher throughout the intervention. Any questions, concerns, requests for topics to be covered, complaints, etc., were to be directed to the representatives and reported to the researcher at the start of each class. None was ever given. The researcher was available after each class to address any concerns or issues that arose as a result of the workshop, but no issues were raised. Session 1 included a description of social psychology and a homework assignment that involved monitoring their television use and naming appealing traits of favorite characters. Each student was given a folder in which to keep notes and handouts, entitled "Media Scope: A Media Skills Workshop," that was to be brought to each class.

Session 2. Gender socialization concepts were introduced in Session 2 and
were a foundational theme throughout the remaining weeks. Props included gender stereotyped toys, greeting cards, and a book of classic children's literature. A ''Name that Stereotype" game was played. An introduction to media literacy was given, followed by a video created for this workshop that consisted of a compilation of images on children's television, particularly cartoons. Images of violence in cartoons directed at boys, stereotyped portrayal of gender roles, and the overall lack oflead female cartoon characters was put in the context of early gender socialization. A homework assignment was given that included interacting with parents or significant others, in the hopes that by explaining concepts learned in the workshop to others, internal acceptance or endorsement of the concepts would be enhanced.
Session 3. During Session 3, television as a commercial enterprise, including its persuasive, entertaining, and informational features, were reviewed. Students were asked to discuss their favorite shows in terms of these three features. Commercial advertisements, and in particular, the beauty industry, were explored. Popular magazine advertising was also highlighted through a group activity using sample teen magazines. Most of Slim Hopes, a film that focuses on the negative portrayals of women in advertising (Kilbourne, 1993), was shown and discussed . . For homework, the students were asked to take an inventory of beauty products in their home, and critically analyze 4 commercials.

Session 4.
The antecedents and consequences of stereotyping were discussed during Session 4. Students went on a guided-imagery "Gender Journey" (Orenstein, 1994), and discussed how life would be different if they were born the other sex. A short lecture was given noting general research findings regarding the negative effect of traditional sex roles for both men and women. Stereotyped characters on daytime and primetime television programs were discussed. A video created for this workshop, comprised of segments from various situation comedies, was shown and discussed. Homework included listing negative personal effects of stereotyping on the lives of the students.
Session 5. After a class discussion of how stereotyping can hurt the students personally, Session 5 focused on music television, its portrayal of gender stereotypes, and the emphasis on sexuality paired with violence. Segments from the "Dreamworlds If' (Jhally, 1993) video were shown, and students were asked to fill out a brief survey during the film to encourage conceptual understanding of the film's major points. Because of the intensity of the images on this video, time was left for discussion and processing of its emotional impact.
Session 6. The final session of the workshop focused on distinguishing positive from negative images and interactions on television and focused on dramatic programming and interpersonal behavior. The reality of television messages was a primary theme and how these messages affect one's own heterosexual relationships. completed. Before collecting the post-test, the investigator asked the students to mark on the first page how many of the sessions each student had missed. The data from anyone who missed more than one session was not included in analyses of data.
Following the post-test, the researcher reviewed the salient points of the workshop, discussed goals for the future, and ended with a relaxed, open-ended group discussion.
An evaluation form similar to the weekly process evaluations was distributed to be filled out and returned at the end of class.
The students from the control group classes, who had been attending regular health classes during the intervention period, also received the post-test measures.
Following the last session, the researcher met with the two teachers, who had been present during each class, to receive input and an informal evaluation, and to thank them for their participation. (Please list)." All adjectives or phrases listed were compared to a list of adjectives compiled from the Bern Sex Role Inventory (BSRI) (Bern, 1974)  and the logged distribution was used in data analyses.
The fifth dependent measure, amount of television watched, was assessed by one question on the same survey, asking for a categorical response to how much television was watched each day, with 1 = less than 1 hour a day to 4 = 5 or more hours per day. The mean score was 2.26 (SD= .886). There were no significant differences in television viewing between groups or genders. Music television use, as described above, was also included as a dependent variable. Although this statistic is nonorthogonal with television use, its particular relevance to gender stereotyping prompted inclusion in some of the analyses .
The final measure, belief in the degree to which television shows are realistic, was measured by a question on the same survey, " Do you think the way most television portrayals of the way men and women treat each other are real or accurate?'' Scores ranged from 4 = "very real" to 1 = "not at all real," with a mean score of2.38.
There were no significant differences at pretest based on experimental/control group or gender. Fifty-four percent of the responses were "not very real," 32.9% were "sort ofreal," and 3% each said "very real" or "not at all real." Originally, choice of television characters who were considered ideal women or men was to be included as a dependent measure but, for the following reasons, was These findings indicate a biased sample in the remaining results.
For female TV character, 39% of the responses were missing or sarcastic, but this was spread evenly between boys and girls, with 39% of each sex not responding.
There were only two perceived sarcastic replies for this question, both from boys.
Finally, there was a much wider variety of responses for female character from girls, with 50% of male responders (N = 23) giving either Pamela Anderson or Jennifer McCarthy as a response.
In addition to the primary dependent variables and music television use, one other question about media and interpersonal interactions was included in preintervention analyses. To the question "How much has television taught you about how to act with the other sex?" 5 Likert-scaled responses were possible, with 1 = "all" to 5 = "none." Assumptions were met, and the mean response was 3.19, or slightly less than "some." Only one student said "all" and 6 students (or 4% of the sample) said "none." Media Profile .
A factorial MANOV A with Bonferroni adjustment revealed some significant differences between girls and boys in their media habits or beliefs (F = 4.19, df = 11, 107,p < .001, eta 2 =.301) (see Table 5). The average number of televisions in the home was 3.4, with 45% of the families owning 4 or more TVs, and 6% owning 6 or more. No family was without a TV. Boys' families owned significantly more TVs Fully 68% of the students owned their own television (N = 87), but this was more true of boys; 61 % of the girls did not own their own TV versus only 38% of boys. As would be expected, the amount of television watched was positively correlated with owning a television (r = .256, p = .003), but there were no significant gender differences in amount of television or music television watched. There were also no differences in media use based on grade level, religion, or parent education level. On average, students reported watching between 2 and 3 hours of television a day, and less than 2 hours of music television a day. Girls liked watching MTV more than boys (F = 5.45, df= 1,117,p = .021, eta 2 = .045). One person reported never having watched music television, and 3 reported not watching television. Table 5 shows means of relevant media use indicators .
The students' responses about the reality of television were, on average, neutral. They reported that interactions between characters are slightly realistic (M = 2.36, range = 1-4, with 2 = "sort of real" and 3 = "not very real"), that the people on television seem slightly realistic ( M = 2.37, with 2 = "somewhat like real people" and 3 = "not very much like real people"), and that the things that happen to them on television are slightly less realistic (M = 2.60, with 2 = ''yes, usually" and 3 = "not usually") . They also reported that television had taught them some amount about how to act with the other sex (M = 3.19, range= 1-5, with 3 = "some" and 4 = "not much"). Table 6 shows media profile means by sex for each group and the 7 classes.
While not considered in the overall analyses, magazine use contributes information to a media use profile. Girls reported enjoying looking at magazines more than did boys (F = 10.3 75, df = 1,117, p = .002, eta 2 = .081 ). In tallying the two favorite magazines of both girls and boys, girls named 130 titles and boys named 99.  Table 7.

Hypotheses
Hypothesis 1 The first hypothesis predicted relationships among the six primary dependent measures. It was proposed that measures of gender stereotyping, the A WSA and the ATIS, would correlate with each other and with lower self-esteem in girls, and with some characteristics of media use. Two-tailed bivariate Pearson correlation coefficients, using SPSS software, revealed several significant relationships, as shown in Table 8. Attitudes toward women correlated positively with traditional adolescent interactions, so that the more traditional the attitudes towards women, the more traditional were their reported interactions with other gender peers (r = .285, p = .001). Both attitudes toward women (r = -.387,p < .001) and traditional interactions (r = -.218,p = .023) correlated negatively with stereotyped ideal traits, so that the less traditional the attitudes and the interactions with other gender peers, the less stereotyped descriptors were used to describe the ideal man and/or woman.
Adolescent traditional interactions also correlated with the belief in the reality of how · people treat each other on television, or reality of interactions (r = -.218, p = .012), with the more traditional the reported interactions, the more the students believed in the reality of television interactions. Social self-esteem correlated negatively with media use, with the more television (r = -.234, p = .006) and music television (r = .246,p = .004) watched, the lower the reported social self-esteem When the overall bivariate correlations were split by sex, the following differences were found, as shown in Table 9. The overall relationship between attitudes toward women and traditional interactions was significant only for girls; the more traditional girls' interactions with boys, the more traditional their attitudes toward women (r = .328, p = 005). Also, the relationship between social self-esteem and media use was only significant for girls. The more television and music television girls watched, the lower their social self-esteem (TV r = -.319, p = .006; music television r = -.367, p = .002). Social self-esteem in girls was also correlated with traditional attitudes (r = .23,p = .03) and traditional interactions with boys (r = .347, p = .003), with the more traditional the attitudes and reported interactions, the lower the social self-esteem in girls. Social self-esteem was not correlated with any pretest variable for boys. The overall relationship between traditional interactions and belief in the reality of television interactions was significant only for girls (r = .297, p = .011). However, more belief in the reality of television interactions was reported as more music television was watched for boys (r = -.247,p = 049). Both boys and girls who reported more traditional attitudes toward women used a higher percentage of stereotyped descriptors for ideal male/female (girls r = -.248,p = .03; boys r = -.267, p = .045).
One final variable was included in the correlation analyses, how much was reported being learned from television about how to interact with the other sex.
Overall, amount of media use (TV r = -.279,p = .001, music television r = -.263,p = .002) correlated with amount learned, with the more watched, the more learned. When girls and boys were looked at separately, this relationship was significant only for boys (TV r = .492,p < .001; music television r = .405,p = .006). In addition to amount of media use, the more students believed in the reality of interactions on TV, the more they reported learning about how to act with the other sex overall (r = .225,p = .009), but significantly so only for girls (r = .243, p = .04). Finally, for girls only, the more traditional their reported interactions with boys, the more they reported learning about how to act with boys from television (r = .316, p = .007).

Hypothesis 2
The second hypothesis predicted that those high in media use would exhibit more traditional gender beliefs, attitudes, and behaviors than those low in media use, that high media girls would exhibit lower social self-esteem, and that belief in the reality of television interactions would be higher. Tables 10 and 11 show the results of these analyses.
To determine media use, total scores for television and music television use were split into 3 groups: high, medium, and low. Low media users were 43% of the sample (N = 59) and were those who watched up to 3 hours of television and/or music television daily. Medium media users were 31 % of the sample (N = 38) and were those who watched 4 hours daily. High media users were 26% of the sample (N = 36) and were those who watched 5 or more hours a day (total N = 133 combinations of gender and media use, as predicted (see Table 11 ).
There was a near significant difference (F = 2.86, df = 2,127,p = .06, eta 2 = .043) in belief in the reality of interactions and media group, with those in the high group (M = 2.27) believing more in the reality of television than those in the low media group (M = 2.55). As can be seen in Table 10, similar trends also existed for AWSA and ATI scores, with low media users less traditional in their attitudes and interactions than high media users.
Because of a lower N, a separate MANOV A was run on percent of stereotyped ideal traits. While significant differences existed for gender, as previously reported, there was no media group significance (F = 1.85, df= 1,102,p = .162 ns). There was, however, a significant interaction (F = 3.158 df= 2,102,p = .047, eta 2 = .058); the girls in all media groups stereotyping significantly less (M = .22 low, M = .24 medium, and M = .23 high) than boys in the high media group (M = .48).

Hypothesis 3
It was hypothesized that the effect of the intervention workshop would be reflected in significantly greater change scores between pretest and post-test for the experimental group than for the control group. Scores were screened, using SPSS software, for group differences, in order to assure that a gain score analysis design would be appropriate. According to Arvey and Cole (1989), a gain score analysis is appropriate only if pretest and post-test correlations are over .50. This was true for most dependent variables, as shown in Table 12.
Of the students who took the pretest, 128 were also present for the post-test (65 experimental and 63 control), and 122 provided complete data for the analyses.
Although many students were absent from one of the other 5 workshop sessions, all participants who did not miss more than one workshop session, and who completed both pretest and post-test measures were included in the analyses. According to the recommendation of Campbell and Stanley (1963), ideally the experimental condition requires subjects to be present at all the sessions, while the control condition does not, and the differential attendance could produce mortality in one group and not the other, thereby biasing the experimental sample in favor of the "conscientious and healthy" (p.16). But, because 1) a similar theme was carried throughout the sessions, 2) review was an element each session, and 3) because external validity is enhanced by more closely simulating a normal school-based curriculum which includes absences, all students who only missed one session were included.  Table 13 shows post-test change data for each dependent variable, and also follow-up change data, as described below. Table 14 shows the same data for the stereotyped ideal trait dependent variable.

Hypothesis 4
It was hypothesized that change due to the intervention would be stable and that distal effects that could be measured at a 3-month follow-up, and would include changes in self-esteem and traditional interactions. As mentioned, the 3-month follow-up testing conditions were markedly different from pre/post-test measures, as students were casually gathered on the gymnasium bleachers waiting for their eye exams. A total of82 completed the follow-up measures, and of those, 74 provided complete data for a pre/follow-up change analysis ( 43 experimental students and 31 controls). As on the post-test data, the AWSA distribution was normal and standard • scores were able to be used and a log transformation was done on the stereotyped ideal traits data A factorial MANOV A revealed no significant differences in change scores between intervention/control groups (F = 1.72, df= 4, 67,p = .155 ns) on the AWSA, ATI, TSBI or reality of television interaction scores measures. Self-esteem means and standard deviations across all three time periods and by group are shown in Figure 1.
Though nonsignificant, both experimental and control boys also showed more change for both the TSBI and AWSA than did girls at both post-test and follow-up (see Table   13).
A factorial ANOVA revealed significant change at follow-up between intervention/control groups in percentage of stereotyped traits used to describe ideals (F = 6.82, df= 1,42,p = .012, eta 2 = .14), but it was not in the expected direction.
Experimental subjects showed a mean change of -.014 and control subjects showed a .094 change, indicating that control subjects used stereotyped descriptors about 9% less at follow-up than at pretest, but experimental subjects used them about 1.5% more at follow-up than at pretest. There were no significant sex or interaction effects. Of note is the low number of subjects who opted to answer this question (N = 60 out of 74, or 81 %) on both pretest and follow-up, possibly biasing the results (see Table 14).

Hypothesis 5
It was hypothesized that at follow-up, high media users would show more  Table 15.

Hypothesis 6
It was hypothesized that after 3 months, workshop girls would have changed more on the dependent measures than workshop boys. However, as noted above, the factorial MANOV A run for media group showed no significant differences between change scores for boys and girls. This was also true for pretest/post-test change scores (F = .482, df= 4,57,p = .74 ns), and for both sets of scores for stereotyped ideal traits.

Additional Analyses
Analyses of Covariance. Although most pretest/post-test correlations were high, as shown in Table 12, a gain score analysis assumes that they are perfect (Arvey & Cole, 1989), and so the data were re-analyzed as a post-test only design. Separate univariate analyses of covariance (ANCOV As) were run on all dependent post-test measures, holding pretest scores as a covariate for each analysis. Once again, there were no significant differences on any dependent variable. Also, to explore the possible impact of the different teachers, and different grade levels, multivariate analyses of covariance (MANCOVAs) were rerun controlling for these factors. There were no significant differences in change scores depending on grade level or which teacher managed the classroom.
Analysis by Classroom. Another problem possibly affecting the validity of this type of design is the use of intact classrooms. As will be described below, the four    A separate ANOV A revealed no significant differences in change scores for stereotyped ideal descriptors.
Stereotyped Ideal Descriptors. In exploring the use of adjectives used to describe ideal male and female traits, adjectives were also categorized as to whether they referred to physical characteristics, and to whether they referred to specifically sexual descriptors. Those who used primarily(~ 75%) physical descriptors, such as "nice legs," "good build," "blue eyes," "nice figure," etc., were separated from those who used fewer than 75% physical descriptors, and from those who used none. There was no difference between girls and boys in male ideal descriptors, and there was no overall difference in using primarily physical traits in describing ideal males versus females. This was because fewer girls used primarily physical descriptors to describe their ideal female, so no overall difference between ideal male and ideal female descriptions appeared.
Paired samples t-tests on just workshop boys showed no significant change in usage of physical descriptors at post-test, although the mean usage increased. Paired samples t-tests on girls showed a significant decrease at post-test for use of primarily physical descriptors for females (t = 2.24, df = 33, p = .032), but a significant increase at follow-up from pretest levels (t = 3.13, df= 21,p = .005). There was no significant change for controls.
Of note is the distribution of specifically sexual descriptors used. Words that referred to sexualized body parts or to sexual status were included in this category, such as "big breasts," "nice ass," "sexually active," and "slutty." These words were used almost exclusively by boys to describe ideal females, although the majority of both sexes did not use them. There were no significant differences between experimental ~nd control groups, and both groups increased between pretest and posttest, with only experimental boys maintaining this higher level at follow-up. For example, the last class of the day was also the largest class (N = 27).
Qualitative observations suggested that the smaller the class, and the earlier in the day, the more attentive and involved the students were . This was supported by process evaluation satisfaction ratings, which revealed a significant inverse relationship between level of satisfaction and class size. The mean satisfaction rating, based on an average of 3 questions with a Likert scale range of from 1 to 4, with 4 indicating the most satisfaction, was lower the larger the class size (r = -.625,p = .001). The 3 questions asked how much the student learned, how much the topic was liked, and how much the things done in class were liked. .506,p = .01), and liking the things done in each class less~= .532,p = .004) than the smallest class. Of note is a similar trend, although only approaching significance, in satisfaction between the smallest and second largest classes for question 2 (p = .07) and question 3 (p = .06). Supporting these findings are mean differences in total workshop satisfaction, a similar set of questions asking students to evaluate the workshop as a whole, shown in Table 20. Again, the smallest class showed the most overall satisfaction (M = 3.76), significantly more so than the largest class (M = 3.11).
The largest class was also the class with the more unsupportive teacher. Both teachers involved were asked to introduce the investigator and to otherwise not play an active role unless disciplinary measures were necessary. From a qualitative standpoint, the two teachers appeared to manage the classroom very differently and were supportive of the workshop to different degrees. The teacher of the largest class was in her first year .of teaching, displayed a more informal class management style, had more trouble managing students, and did not emphasize the importance or value of the workshop as much as the other teacher, who taught the other three experimental class groups.
Quantitative results of the overall process evaluation for the workshop as a whole revealed increased satisfaction in a response-by-class pattern similar to the session evaluations. The final evaluation asked for the participant's gender, and, as shown in Table 21, girls consistently rated the workshop higher than boys, with an overall mean of3.63 for girls and 3.29 for boys.
Qualitative comments were overwhelmingly positive, and can be summarized as follows. To the question, "What was the thing you like best?" the majority response from both girls and boys was ''the videos." Other popular choices were the discussions and role playing. Several girls made comments about what they learned, such as "I liked learning about how the television stereotypes people. You never really think about it until it is pointed out," and "I never realized how much the media is involved in my life." Many left the question "What did you like least?'' blank or said "nothing." Otherwise, the most frequent response was "the surveys." Some also mentioned the lectures, and some noted the homework. The few comments of a substantive nature mostly referred to uncomfortable aspects of the video "Dreamworlds 11 " Although . watching guys' reactions to television-that helped me learn about my classmates." Table 22 lists sample responses to the 3 process evaluation questions.

DISCUSSION
The primary goal of this study was to test the impact of a gender-focused media skills workshop in a high school class setting on the attitudes, beliefs, and behaviors of adolescents. It was proposed that greater media exposure is related to traditional ideas about gender and that these are related to low social self-esteem in girls. It was further proposed that a workshop dealing with media literacy and gender issues, especially as these relate to interactions with other gender peers, would be effective in changing some of these attitudes, beliefs, and behaviors, and would result in higher social self-esteem scores in girls. Although participants in the workshop did not.show significant change on any of the primary dependent measures, a number of findings support relationships among variables, and the qualitative results are encouraging.

Pre-Intervention Relationships
This sample of adolescents from a middle-sized town in Rhode Island watched an average of 1 to 2 hours of television a day, and less of music television. Although boys owned their own TV more often than girls, they also reported watching television with their parents more than did girls. Even though boys watched slightly more TV than girls, girls reported watching more MTV, a highly potent source of stereotyped Images. Girls reported more parental limits placed on their viewing and girls talked more with their parents about what they watched. It is possible that this is an indication of more protective behavior in general toward girls, but also may reflect girls' choices of programs. MTV and programs such as Beverly Hills 90210, daytime soaps, or daytime talk shows, such as Jenny Jones, more often deal with topics that have to do with sexuality than do situation comedies or sports programs, and are watched more by girls than boys.
Correlational relationships and questions about media habits revealed interesting profiles that both support previous :findings and provide intriguing questions for future research. The profile oftraditionality found elsewhere is supported here, and suggests some degree of construct validity between measures.
Girls, in general, were less traditional in their gender attitudes and beliefs than were boys. Girls and boys with more traditional attitudes about gender endorse more traditional interactions, and the more students believed in the reality of television interactions, the more traditional were their reported interactions with other gender peers. The more traditional, the more "ideal" males and females were described in a traditional way. Thus, these measures provide a constellation of traditional role indicators.
The pattern of responses for the dependent variable dropped from the analyses, choice of TV character that most fit one's ideal list of traits, also supports these findings. Boys were far more likely to not respond to this question or to offer sarcastic responses, especially for choice of ideal male. The nonresponders were significantly more traditional in their attitudes and interactions. One of the hallmarks of traditional hegemonic masculinity is a rejection of any identification with feminine attitudes, beliefs or behaviors, and thus a strong rejection of homosexuality. The notable resistance to identifying a male character as an admired ideal might indicate a homophobic response by more traditional boys. Seven of the 30 boys, or 23%, who did not respond or who responded sarcastically were clear in their attempt to reject any association with homosexuality, saying ''I'm not gay." Of primary interest are indications from this research that level of media exposure is likely related in some way to traditional ideas about gender. Although not statistically significant in this sample, all dependent measure means were in the expected direction depending on extent of media use. As can be seen in Table 10, low media users had more contemporary attitudes toward women, more contemporary approaches to interacting with other gender peers, used fewer stereotyped descriptors in describing their ideal male and female, believed less in the reality of television interactions, and had higher social self-esteem than high media users. This was true for both boys and girls.
However, as expected, low social self-esteem was more strongly related to media exposure, especially music television, for girls. Confirming previous findings, low social self-esteem in girls was also strongly related to endorsement of traditional gender interactions and to traditional attitudes toward women. Although traditionality and media exposure are both linked to lower social self-esteem in girls, no direction of causality can be inferred. Also, traditionality and media exposure were not correlated with each other, suggesting that there may not be a direct link between the two. The nature of these relationships should be the focus of future research.
Perhaps low self-esteem is the determining variable here , rather than the stereotyped nature of media images. A needs and gratifications interpretation would suggest that students with low social self-esteem, perhaps due partly to a traditional framework about gender roles, seek out more media time as a form of diversion or soc_ ial integration. According to Rubin (1984), these two motivations for viewing serve different needs and gratify in different ways. Diversion provides emotional release and an escape from reality, and social integration may help one overcome loneliness or provide parasocial relationships with characters. When these characters are portrayed attractively in stereotyped ways, low social self-esteem girls may find a level of reinforcement for traditional gender ideas. One can envision from here the creation of a negative cycle, grounded in needs and gratifications as well as social learning theories, of a modeling of traditional gender roles, leading to lower social self-esteem, leading to a need for diversion and social integration found in the seductive and vicarious reinforcement of television.
A third motivation for viewing, according to needs and gratifications theory, is for informational purposes. Media consumption for boys in this sample does not appear to be linked with their social self-esteem, but it does appear to be related to learning about interactions with girls. A strong correlation of media exposure, television and music television, with how much boys say they learn about how to act with the other gender, indicates that an important use of the media for boys is for informational purposes. It is likely that girls have additional resources for gaining this type of information that boys might not take advantage of, such as teen magazines, and intimate conversations with other girls. This coupled with the finding that the more boys watch music television, the more they believe in the reality of interactions on television, is a potential cause for concern given the association of music television and sexualized violence that some researchers are exploring.
The cultivation model is a primary theoretical :framework that explains media impacts in a reversed direction. Using this mode~ traditional gender attitudes, beliefs, and behaviors are cultivated in heavy viewers of stereotyped programming.
Eventually, through repeated exposure, these messages become mainstreamed into the dominant belief system represented in the media. Therefore, low social self-esteem in girls can be cultivated through heavy exposure to traditional gender roles. This model is further supported by significant relationships in this sample of girls between traditional attitudes and interactions and both the belief in the reality of televised interactions, and the belief that television has taught them a lot about how to act with the other gender.
High media use is also associated in this sample with lower social self-esteem in boys, although less so than for girls, which is not easily explained using a cultivation model in the face of evidence that male roles on television support dominant cultural norms for men. What seems more reasonable is that individuals with high self-esteem are busier seeking out activities that may be more selfactualizing than watching television, supporting a needs and gratifications model.
Van Evra (1998) has offered an integrated theoretical model for explaining media effects that fits the data gathered here, and may be useful in explaining the absence of more statistically significant findings. Van Evra proposes that four factors be considered in evaluating media effects. First, the motivation for viewing must be considered, an idea endorsed by needs and gratifications theory. Second, the perceived reality of television contributes to how much is internalized. Third, aniount of viewing will moderate effects, as suggested by cultivation theory. Finally, the number of information alternatives that are available to the viewer will determine whether the messages are cultivated, or how much is endorsed or modeled.
The motivation for viewing in this sample and the choices of programming may be different for those with different social self-esteem levels, or for those with different viewing goals. Boys may be seeking information and girls may be seeking diversion or parasocial interaction. Second, one of the standard goals of media literacy has been to teach children that television is not real, and to show them the technical and psychological aspects involved in producing images that mimic real life.
As adolescents, this · sample may be able to discern when messages are realistic, possibly explaining why responses on these survey questions were mostly neutral.
Better questions perhaps would have been about their desire to model television images, or about the persuasive properties of advertising or of documentary/informational types of programming.
Third, the amount of viewing in this sample appears to be related to certain ideas about gender. As shown in Table 10, heavier viewers were more traditional and had lower social self-esteem. Had the sample been larger, these trends may have reached significance.
It is possible, as Brown and Newcomer (1991) suggest in their analysis of how the proportion of"sexy" programming adolescents watch might be related to sexual behavior, that the amount of media exposure is less important in determining ideas about gender than the type of shows watched. Boys appeared to watch more comedies, such as Seinfeld, The Simpsons, and Home Improvement, police shows, such as New York Undercover, and sports programs. Girls were more likely to choose prime-time domestic dramas, or "night-time soaps," dealing with young adults in relationships, such as "Beverly Hills 90210, Party of 5, and Sunset Beach. Future research might investigate the relationships between gender ideas and type of programming watched.

Results of the Workshop Intervention
Exposure to the media literacy workshop dealing with gender stereotyping developed for this project did not effect changes in gender attitudes, beliefs, and behaviors in this sample of adolescents, and had no significant effects on the social self-esteem of girls. No overall change score was significant except for two unexpected findings: that boys in the workshop classes experienced a drop in social self-esteem and the workshop students stereotyped more three months later than the control students.
However, there were indications from the critical viewing skill assessment at post-test that some learning took place. Although statistically not significant, both workshop boys and workshop girls responded that the characters in the post-test comic video clip did not act like they would have in real life, or that they did not act like they should have acted. Workshop participants would have rewritten the script using a less stereotyped plot than control students. Based on the qualitative comments that accompanied the Likert-scaled responses, however, the interpretation of the questions was diverse, and likely flattened the response. For example, many responded that indeed many men and women do act like the ones in the clip, although those behaviors were judged as silly or inappropriate. The two questions asking for judgments about how they should have acted were purposefully open-ended, without referring specifically to stereotyping, and the qualitative responses again were varied. Because it was a comedy, many did not take the behaviors seriously. Perhaps if more pointed questions were used, and the choice of video clip was not a comedy, the responses would have been clearer to interpret.
The true effects of this workshop are difficult to uncover. Problems with theoretical approach, with methodological and validity issues, and with features of the sample may obscure positive impacts. The inoculation approach, also known as impact mediation (Hobbs, 1998), is a skill-based, protectionist one in which students are taught to resist the negative impact of specific messages, such as those having to do with violence or gender stereotyping. Akin to traditional approaches to sex education, these methods have been criticized as being elitist, instructor-focused, and largely ineffective. As Hobbs (1998) notes, "When media literacy skills are positioned in opposition to media culture, the quality of the instruction is compromised . . . (S)tudents are unresponsive to the inference that they are helpless victims of media influence who need to be rescued from the excesses and evils of their interest in popular culture"(p.19).
Some have suggested abandoning this traditional theoretical approach to media literacy education (Brown, 1998b ). This study used the inoculation approach, which, while most feasible given the short-term nature of the intervention, seemed to have produced resistance in the students, as suggested by the increase in gender stereotyping found at follow-up. As part of the identity process, a separate "adolescent culture" (Shamai & Coambs, 1992) becomes well established and is resistant to change. Gender role adherence is typically intensified during adolescence, bringing into question the developmental timing of introducing a topic that attempts to change the sexual scripts they are learning from that culture. Perhaps broaching gender role issues would be more successful with pre-adolescents, who are more flexible in their ideas about gender. Also, as Buckingham (1993, in Hobbs, 1998 noted, the protectionist stance ignores adolescents' emotional engagement with the media, and tries to replace the genuine enjoyment they experience with negative cymc1sm.
Aspects of the Elaboration Likelihood Method used in developing the workshop that should produce central route change were not emphasized enough.
According to this method, central, rather than peripheral, route change is most effective in attaining permanent attitude change, but is more difficult to attain because significant cognitive appraisal is required. Motivation, ability, and favorable thoughts resulting from the cognitive appraisal must be present for central route change to take place. In the current study, the motivation to cognitively process the type of information presented was likely not high, especially for boys, who were more resistant to the program than were girls. Favorable thoughts about the message topic were probably not fostered enough, given the implication that an adult "expert" was critiquing "adolescent culture." The effectiveness of persuasive communication depends on three factors: the source of the persuasive message, the content of the message itself, and the personal involvement of the audience (Brehm & Kassin, 1993). First, the source should be credible and likable, with interpersonal similarity a key factor. A workshop run by peers, a team consisting of a man and a woman, or someone more closely identified with adolescent culture might increase attitude change. Second, if the message content is too discrepant from the attitudes of the audience, change will not occur.
Attending to a message and then accepting it is more likely when the message confirms existing beliefs. Thus, for boys, a persuasive message that contradicts existing ideas about male/female relationships may be resisted, especially if that message is perceived as judgmental. Reactance theory suggests that information may be resisted when it threatens perceived freedom or threatens a fundamental view of self (Zimbardo & Liepe, 1991). The basic message from this workshop for boys, that their attitudes, beliefs and behaviors about gender relationships may be unhealthy, could have been perceived as a threat to their freedom or to their fundamental view of self. Establishing the specific goals for promoting critical analyses of gender roles and media messages should include a consideration of baseline attitudes of both boy and girl adolescents.
Finally, the message should have personal relevance to the lives of the target audience. Young adolescents perhaps do not easily recognize the personal negative impact of gender stereotyping. Encouraging more experiential activities and allowing adolescents more opportunity to arrive at their own conclusions might have more lasting impact than an inoculation approach. Engaging in behaviors that conform to new cognitions can act as a form of self-persuasion; that is, to reduce discomfort caused by a discrepancy between their behaviors and pre-existing attitudes, students may modify their attitudes without the direct persuasion involved in an inoculation approach (Festinger & Carlsmith, 1959). If students were asked to teach others, such as parents, about gender stereotyping, were asked to engage in extensive reversed-role activities, or engage in other behaviors that support the goals of the workshop, it is possible that the personal relevance of the message would have been better experienced. Although these approaches were taken in the present workshop, the intervention was too brief to produce enough impact. These techniques should be explored in future workshop designs.
Better known in Europe is a media literacy approach called "discriminating responsiveness," by .which students are taught analytical thinking skills to be used independently. This approach encourages free, active participation on an individual level. It is an interactive model of sender and receiver, whereby individual receivers interpret media messages in terms of their own experiences, needs, and associations (Brown, 1998b). Because meaning is situated in the receiver, not in the content, students are empowered to come to their own conclusions about such things as the negative impact of gender stereotyping in television.
Combining this approach with considerations of who the change agent is appears more promising. Incorporating media literacy efforts into the normal school curriculum is an obvious necessity, given the degree to which youth culture is mediadriven. Using a discriminating responsiveness approach on the part of teachers would require teacher training, but would decrease the effect of individual agendas of teachers who will undoubtedly have differing opinions about what is most important about media literacy.
Although primary analyses did not indicate significant changes as a result of the workshop, several indications suggest some impact. Girls significantly decreased their use of stereotyped descriptors at post-test, but significantly increased their usage at follow-up. One interpretation of this finding is an immediate positive effect of the workshop on girls, but without lasting value. Another interpretation is that, at followup, resistance to filling out yet another survey, plus the uncontrolled, jovial environment of the follow-up setting influenced results in an unexpected way. It is possible that some challenging of traditional role expectations and some self-scrutiny was engaged in by workshop participants, which may have produced a level of discomfort that was ·resented when it resurfaced 3 months later at follow-up.
Of interest is the increased number of stereotyped descriptors used by experimental boys at post-test and follow-up, and the greater number of sexual descriptors used by experimental boys than control boys at follow-up. While one immediate assumption is that the workshop had a negative impact on boys, who displayed their resistance through reversed responses, it is clear that they were hearing some message. It is proposed that distal effects may be the most important ones, and resistance to an inoculation approach dealing with gender, perhaps a topic of discomfort for adolescent boys in generai might translate in later years into a heightened sensitivity to this issue.
There is another finding suggesting that intervention resistance influenced results. The two smaller classes, classes 1 and 2, were much more attentive than the two larger classes, and it was the two smaller classes that stereotyped 13% and 10.6% less at post-test (than at pretest), while the two larger classes stereotyped 15.3% and 14.4% more at post-test. By follow-up, the reverse occurred.
An intriguing finding is the drop in social self-esteem experienced by experimental workshop students, especially the boys. The scores for all boys dropped between pretest and post-test, indicating one of three things. Internal validity may be compromised by the effects of pretesting, which could have sensitized all subjects to issues of gender stereotyping and interacting with other gendered peers. This could have a more negative impact on the boys than on girls. Second, a common history, such as developmental changes, or end-of-the-year worries about exams, summer plans, etc., could have had a differentially negative impact on boys.
Since these changes were not statistically significant, however, they could be due to chance. It is possible that the changes in self-esteem could involve instability of · the .measure itself Although the TSBI test-retest correlation in this study was high (.81), no further reliability data was available. The construct validity of the measure is also a potential problem, as the measure may reflect level of social skill or competence rather than self-esteem. Refining the construct of social self-esteem and finding a more discrete measure might enlighten the findings from this study.
Finally, although chance findings were possible, at follow-up the control boys' social self-esteem raised slightly, but experimental boys experienced a significant drop. The same pattern was true for girls. Again, follow-up conditions were not ideal, and the sample size was smaller, but this pattern seems to suggest that something inherent in the workshop affected the social self-esteem of its participants in a negative way, perhaps by introducing questions about what they had previously taken for granted.
It is easier to speculate about this finding for boys than for girls. It is possible that boys' newfound awareness of the effects of stereotyping on girls might contribute to feelings of self-doubt about identifying with traditional masculinity. This is enhanced by the finding that low media use boys, who perhaps get the least reinforcement for adhering to a traditional role, experienced the greatest drop in social self-esteem among the experimental boys.
The findings for girls are more difficult to interpret, but it is possible that any discussion about gender roles during this time of identity formation is unsettling, and may provide another reason for beginning these discussions at an earlier age and continuing them throughout the school years.

Qualitative Results
The most valuable outcome evaluation may be the qualitative responses of the students. Process evaluation results clearly support the use of small classrooms, as a direct, linear relationship existed between satisfaction ratings and class size. In spite of a lack of significant change scores and some reactions, particularly by boys, indicating resistance to the intervention, qualitative observations indicate the opposite.
Students were, for the most part, engaged, interested, and inquisitive. Many active debates took place, and watching videos was a universal highlight in each class. The process evaluation comments were almost all positive, particularly from girls, who rated the workshop higher in general than did boys. The aspects that were liked least are features that would not be present if this were a curricular addition, such as surveys and evaluations. Although lectures were mentioned as being disliked, the classes were highly interactive and made use of varied teaching strategies.

Class reactions to the various topics provided indications of students' values,
and provided impromptu teaching opportunities. For example, one video segment was a clip from the TV sitcom, Ellen, which dealt with a woman grappling with the realization, acceptance, and public acknowledgment that she was lesbian. The clip offered many opportunities to discuss stereotypes, including their particular effectiveness in comedic formats. Many of the students initially reacted with discomfort, in the form of joking, distancing comments, but after serious discussion about their responses and how stereotypes help maintain distance from people we are not comfortable identifying with, the joking diminished or stopped.
The question arises as to measurement techniques. Effecting change in constructs such as self-esteem and attitudes toward women may be a more subtle process than could be revealed by the dependent measures employed in this study.
Qualitative data may perhaps be the best early indicator of effects. Distal effects are most difficult to assess, but are surely the most valuable. It is possible that six hours of intense exposure to gender concepts, illustrated in active and engaging ways, did indeed leave some sort of lasting impression, and will provide a level of "inoculation" that may not be manifested until early adulthood . Longitudinal designs are of value here.

Validity Issues
External validity issues include possible instrument reactivity, and reactivity to the experimental arrangement. The students were very much aware of the researcher's role as a data collector as well as teacher. If media literacy were incorporated into the normal curriculum, resistance to intervention might not be a factor. Also, the differential attitude of the two teachers was perhaps a factor in some of the negative responses from class 4. This issue speaks to the need for the interventionist to assess readiness for change (Backer, 1995) and engage in an appropriate "unfreezing process" with the target organization. Media literacy is not yet a well-accepted concept in mainstream public education, any more than is education about gender stereotyping. Readiness involves assessing the existing cognitions, capacity and motivation for change, and expectations of the target organization or individual. It has been well documented (Fennema, 1993) that institutional consideration .of unfair, stereotyped treatment of girls is rare. As was experienced by this investigator, a lack of motivation for change can have significant detrimental impacts. A lack ofreadiness for change on the part of the teacher(s) perhaps was recognized by students, particularly in class 4, leading to heightened student resistance. Careful organizational assessment and the implementation of readiness enhancement measures may be as important as the intervention itself: and should be a focus of future program planning.
Internal validity problems, common in school-based interventions, were prevalent. While a control group design controlled for differences in history, maturation, mortality, and testing, a primary threat to internal validity involved different intrasession histories. Random assignment of participants was not possible.
An interaction effect of selection with history and maturation may be involved because the control group had a qualitatively different experience -a health class .
Different intrasession history among the experimental groups included two fire drills, a field trip, and Junior prom day, during which all but three of the 11th grade workshop members were absent. Each class met at its own time of day, and the dynamics in the class that met for the last time period of the day, for instance, was qualitatively different from that of the morning classes; it was more disruptive and students displayed less motivation . End of day fatigue and "spring fever," intensely apparent after 1 :30 p.m. with windows open in late May, were possible factors that influenced the quality of workshop experience .
For these reasons, treating participants individually is somewhat misleading , and a more appropnate approach, if the N were larger, would be to evaluate intact classrooms. On the other hand, these differences are present in any school setting and may actually increase external validity at the same time that internal validity is compromised.

Conclusions
It is becoming clear that media literacy is a necessary adjunct to the school curriculum. The movement to integrate these critical thinking skills into public schooling is gaining momentum, as can be seen in the increased number of statewide initiatives, such as in Massachusetts. At the same time the inoculation approach is being challenged as ineffectual, in sex education and other venues where resistance might be an issue (Kubey, 1998;Shamai & Coambs, 1992).
Also at issue is the well being of adolescents. Investigating empirically the protective and risk factors that contribute to the successful negotiation of identity development for both boys and girls is cruciaL as it is integral to their success academically and socially. As Dreyer (1994) suggests, educational reform movements must include the psychosocial needs of children and adolescents. This is particularly true for adolescent girls, who are negotiating both the world of adolescence as well as the world of inequality. Minority and poor girls face even more serious challenges, as they must also face issues of discrimination and economic disadvantage.
This study focused on how traditional gender attitudes, beliefs, and behaviors relate to social self-esteem and interactions with other gender peers, and how the media contributes to these. It was assumed that low social self-esteem in girls combined with traditional ideas about how to relate to boys, paired with media reinforcement, might be a formula for negative outcomes. But media images are complex and often present contradictory messages. Magazine articles and TV shows with contemporary images of women provide additionaL alternative information for adolescents about gender roles, but the messages are mixed with traditional ones. This seems to be particularly true of stereotyped images that are portrayed in highly attractive ways that obscure the real value (or lack of value) of their roles as women.
The show, Ally McBeal, highly popular with adolescent girls, provides an example of mixed messages to girls_to be both successful and assertive in a competitive legal environment, and to be waif-like, fragile, childlike, attractive to men, and, as is Ally's most prominent feature, emotionally labile.
There is active debate about how media literacy programs, well established in many other countries, should be configured, who should teach them, and how they should be incorporated into a school curriculum (Hobbs, 1998). Some, such as Lewis and Jhally (1998), recommend taldng an activist stance, teaching children that first and foremost, a political agenda underlies all media messages, and viewers should consider the source before they consider the message. While it is important to raise concerns about the representation of gender roles, as well as minorities, in the media, perhaps separating media literacy in the schools from gender stereotyping or other political issues is an approach that may avoid resistance by adolescents. We must find the best way to help adolescents make self-initiated connections between media and gender roles.   (Bern, 1974) P = Personal Attributes Questionnaire (Spence, Helmreich, & Stapp, 1974)  c Higher score reflects less belief in reality of TV interactions (1 = ''very real" to 4 = "not real at all") d Higher score reflects less learned from TV about how to act with opposite sex (1 = "all" to 5 = "none")  -.279** -.263**  2.82 (4.46) .807 (3.14) 1.13 (7.27) .006 (5.91) -.176 (7.55) 1.38 (9.45) .733 (6.08) .750 (5.66) N=74 Group Mean (SD) 1.60 (3.80) .580 (6.52) . 767 (8.69) .741      ]=intact family, 2=single-parented, 3=reconstituted, 4=other If your mother or father does not live with you, how often do you see him or her?    In your opinion, what are the characteristics of the ideal or "perfect" man? (please list)

Scoring for these four questions described in Methods section
What one TV character is most like that ideal man?
In your opinion, what are the characteristics of the ideal or "perfect" woman? (please list) What one TV character is most like that ideal woman? If a boy answers a question wrong in school, I am more likely to correct him than ifhe were a girl.

15.
Ifl see a sexy girl who I don't know personally, I will often make a comment or a gesture to let her know that I think she is "hot."* 16.
It is very important that my friends know it when I have a girlfriend.
17. Ifl want a girl to notice me, I will try to dress in a sexy way around her.

18.
Where girls are concerned, I try harder to look attractive than to act smart.

19.
I usually try to flirt with girls that I want to like me instead of telling them that I like them. · Items 1, 2, 5-7, and 15 are reverse-scored.

APPENDIXE
Critical Viewing Skills Critique (Part 1) Take a few minutes to think about the video clip you just saw and then answer the following questions. Take as much space as you need (you may use the back of the paper.) 1 Did the man act differently than he would have in real life? _Yes, totally_ Sort of _ Not really _ Not at all_ I didn't notice Please explain:

2.
Did the woman act differently than she would have in real life? _Yes, totally_ Sort of _ Not really _ Not at all _ I didn't notice Please explain:

3.
Pretend you are from another planet where everyone is the same sex -"male" and "female" do not exist. You have never seen human beings before. You are going to visit Earth disguised as either a man or a woman. In preparation, you see this film clip. Based on what you see, would you rather come as the man or the woman? Why?

4.
Did the man act the way you think a man should act in that situation? _Yes, totally_ Sort of _Not really _Not at all_ I didn't notice Why?

5.
Did the woman act the way you think a woman should act in that situation? _Yes, totally_ Sort of _ Not really _ Not at all _ I didn't notice Why?
6. If you were re-writing this scene, how would you change what happened?  My child, ________ _, has been asked to take part in the Critical Media-Use Skills Workshop research project described below. The researcher, Barbara Silver, will explain the project to him/her in detail. My child is free to ask questions at any time. Ifl have questions at any time, I can contact Ms. Silver (401-874-2193), or her faculty'advisor, Professor Bernice Lott (401-874-4248), anonymously, if I wish.
I understand that this is a 7-lesson workshop that will take place during the month of May in health class. During this time, my child will watch short clips from television shows that are popular with adolescents, and will engage in discussions and activities designed to teach children to better understand the kinds of messages the media sends. I also understand that there are no risks or discomforts in this workshop. My child will benefit by becoming more educated about the television industry and aware of some effects the media has on his/her personal life.
My child's part in this study is confidentia~ and his/her name will not appear on any measure taken. All records are to be kept by the researcher and any forms filled out will be identifiable by number only. I also understand that the decision to participate is up to me and my child. My child is free to withdraw from the study at any time with no negative consequences whatsoever. If my child wishes to quit, he/ she simply informs Ms. Silver at the beginning of class, or by calling her at 401-874-2193. My child can then join the regular health class.
IfI am not satisfied with the way this study is performed, I may discuss my complaints with Ms. Silver or Professor Lott, as noted above. In addition, I may contact the office of the Vice Provost for Research, 70 Lower College Road, University ofRhode Island, Kingston, Rhode Island, telephone: (401) 874-2635.
I have read the Consent Form. My questions have been answered. My · signature on this form means that I understand the information and that my child may participate in this study.

Signature of Parent _____________ _ Date ____ _
Printed Name ---------------- I have been asked to take part in the Critical Media-Use Skills Workshop research project described below. The researcher, Barbara Silver, will explain the project to me in detail. I should feel free to ask Ms. Silver questions at any time. IfI wish, I can also contact Ms. Silver by telephone (401-874-2193), as well as her faculty advisor, Professor Bernice Lott (401-874-4248), anonymously, ifl wish.
I understand that this is a 7-lesson workshop that will take place during May in health class. During this time, I will watch short clips from TV shows that are popular with adolescents, and will engage in activities and discussions designed to teach me to better understand the kinds of messages the media sends. I also understand that there are no risks or discomforts in this workshop. I will benefit by becoming more educated about the television industry and aware of some effects the media has on my personal life.
My part in this study is confidentia~ and my name will not appear on any measure taken. All records are to be kept by the researcher and any forms filled out will be identifiable by number only. I also understand that the decision to participate is up to me. I am free to withdraw from the study at any time with no negative consequene:y whatsoever. lfl wish to quit, I simply inform Ms. Silver at the beginning of class, or by calling her at 401-874-2193. I can then join the regular health class.
If I am not satisfied with the way this study is performed, I may discuss my complaints with Ms. Silver or Professor Lott, as noted above. In addition, I may contact the office of the Vice Provost for Research, 70 Lower College Road, University ofRhode Island, Kingston, Rhode Island, telephone: (401) 874-2635.
I have read the Consent Form. My questions have been answered. My signature on this form means that I understand the information and that I agree to participate in this study.

Signature of Student _____________ _ Date ----
Printed Name -------------  Thanks to the happy hours spent with the books, many children have developed a lifelong love of reading.

Over one billion LITfLE GOLDEN BooKS
have reached the hands of children . We salute the talented authors " and artists who create the booksand also the readers, young and old, who have enjoyed the picture books with the famous gold~foil binding.  Adopting traditional, stereotyped gender roles has some benefits. But it also has many negative effects. Here are just some of the negative effects found through research that happen more to those who adopt these roles than to those who do not: For males: Note: Much recent research indicates that heavy TV viewing is related to traditional attitudes about gender roles.

Session 5 "Dreamworlds 2"
1. Music videos are advertisements made to sell CDs and tapes. What is one thing they know will get our attention through the "clutter" of other ads? a. good music b. sex c.
great camera shots d.
seeing a video over and over again 2. Why is the title of the film "Dreamworlds"?
3. What do the producers of many music videos make the women of the Dreamworld be like? What are some of their characteristics?
4. Name a female artist who had to change her image to fit the "Dreamworld woman" image in order to sell her music.

5.
When we look at a· person only in parts or fragments -like just their breasts or legs -we lo_ se respect for them as a whole person. What does this mean?
6. Many music videos don't show bad treatment or even disrespect of women. But it's there enough so that when we do see it, it: a. seems normal b.
really stands out c.
makes us angry d.
sells more records 7. What does presenting women as objects to be looked at and touched have to do with sexual violence? Remember that the images you see are exaggerated to make them more interesting, exciting, or attractive.
Know that sex and violence are shown in unrealistic ways in the media -both have tremendous consequences that we usually don't hear much about .
Ask yourself if you are being true to yourself or if you are fqllowing peer or social pressure to fit into some kind of "image." ' Don't do or be something that makes you feel uncomfortable just because you see it alot in the media.
Most of the reason boys and girls act so differently is because we are taught to. We are alot more alike than we are different -we are just expected to behave in certain ways .
Don 't worship beauty too much -it is a shallow way to judge ourselves. In your lives, the most wonderful people you will know will probably not be the most beautiful.
Expand your horizons -respect yourself and others for your true selves -don't get stuck in narrow, stereotyped ways of thinking about how to be male or female!

CONCLu,grofJ
The media is fun as long as we tealize itg mru:gagru: can be vety ~eH:uagive and ate not always healthy. Remembet, if the media ghowed life as it teally ig, we wouldn't bothet watchingll