A content analysis of social groups in prime-time Spanish-language television.

Historically, content analytic research has pointed to disparities in the representation of Latinos in U.S. media offerings in terms of both the sheer number of Latino characters (Greenberg, Mastro, & Brand, 2002) as well as the often-stereotypical nature of these depictions (Ramirez Berg, 2002).

Although a small but growing body of empirical studies documents the nature of Latino portrayals on U.S. television, little empirical research has examined the imagery presented in Spanish-language television. Research investigating the features of Spanish-language portrayals becomes particularly meaningful when the rapidly increasing viewership and popularity of Spanish-language television networks in the United States is recognized (Barnes & Jordan, 2005). Moreover, given that the programs aired on these networks are nearly exclusively produced outside the United States (Consoli, 2005; Univision, 2005) they are likely to differ from those on English-language networks, owing in part to differing cultural norms and ideals, including those associated with traditional gender and sex roles (Glascock & Ruggiero, 2004). When considered from the perspective of social identity theory, the characteristics of these portrayals become of consequence as these images would be implicated in processes of identity formation and social comparison (Harwood & Roy, 2005) among U.S. Latino viewers. Therefore, the present study applies a social identity theory framework in content analyzing a 1-week, random sample of Spanish-language television.

Spanish-Language Networks in the United States

The Spanish-language television industry has been steadily growing in the United States since the establishment of Univision (Stillig, 1995), which currently ranks as the most-watched Spanish-language network (Grillo & Bednarski, 2004). Indeed, since the inclusion of Spanish-language programs in the Nielsen rankings, Univision has emerged as the fifth most-watched television network in the United States ("Television en Fuego," 2006), due to some extent to reaching 98% of all U.S. Latino television households (Univision, 2005). Telemundo is second in Spanish-language television viewership, penetrating 92% of U.S. Latino television homes (Telemundo, 2005), followed by the Univision Communications-established Telefutura, which reaches 85% of Latino television households (Downey, 2005b). The final player in the Spanish-language television race is Azteca America (Downey, 2005a), currently reaching 77% of the total U.S. Latino population (Azteca America, 2005).

Telenovelas, (1) similar to soap operas, dominate the Spanish-language television networks and provide the principal basis for Univision's high ratings (Azteca America, 2005; Consoli, 2005; Downey, 2005b). Given their tremendous popularity, these programs have attracted the attention of the U.S., English-language networks. Notably, ABC, CBS, FOX, and NBC all have English-language telenovela projects in development, with plans to begin airing as early as 2006 ("Television en Fuego," 2006; "Telenovelas," 2006).

It is notable to point out that in addition to Univision's status as the most watched Spanish-language television channel, it also presents a threat to the major U.S. broadcast networks; frequently outranking at least one of the English-language networks among young adults (Barnes & Jordan, 2005). Univision further asserts that more U.S. Latinos watch its programming in every daypart than any of the four mainstream broadcast networks (Univision, 2005). When assessing media use among bilingual Latinos in the United States, this is meaningful as members of this segment of the population have the option of selecting from both English- and Spanish-language programming--vastly broadening the television offerings. In fact, research has revealed that whereas older viewers are likely to attend more to Spanish- than English-language programming (Barnes & Jordan, 2005) due to the fulfillment obtained from watching shows that represent their cultural background (Faber, O'Guinn, & Meyer, 1986), younger viewers increasingly choose Spanish-language programming based simply on the appeal of the content (Barnes & Jordan, 2005).

Focusing attention on Spanish-language television is additionally valuable because Latinos typically report among the highest levels of television consumption in the United States. Across all age groups, Latino American households watch more television in prime time and daytime than the average U.S. home (Nielsen, 2005). When these elevated rates of television consumption are coupled with theoretical research suggesting that media effects are more pronounced among frequent users of the media (Gerbner, Gross, Morgan, & Signorielli, 1980), the imagery to which this segment of the population is exposed becomes increasingly important; both in terms of viewing English- and Spanish-language television.

Portrayals of Latinos on English-Language U.S. Television

The frequency and nature of Latino portrayals on U.S. television has been an issue of longstanding concern--Latino representation in U.S. programming is rare and when present, these images often involve negative stereotypes (Greenberg et al., 2002). Indeed, during the fall 2003 season and across six broadcast networks, Latino characters on prime-time television programming accounted for only 6.5% of the character population (Children NOW, 2004), despite comprising approximately 12.5% of the U.S. population (U.S. Census, 2000). Among all Latino characters portrayed on television, only 11% held high-status jobs, with Latinos more likely to be seen in roles as domestic workers than any other racial/ethnic group (Children NOW, 2004).

Harwood and Anderson (2002) found that Latino television characters were deemed less attractive and less appropriately dressed than their White counterparts. Additionally, they were identified as the lowest in social attraction and fulfilled negative plot functions more so than Whites, indicating the malevolent nature of these characterizations. Mastro and Behm-Morawitz (2005) found that among men, Latinos were of lower job authority than Whites, and among women, Latinas were of lower social authority than Whites. Although not significant by Scheffe tests, chi-square tests revealed that Latinos were portrayed as lazier and had lower levels of intelligence than White characters. Moreover, compared to their White peers, Latino women possessed the lowest work ethic and demonstrated the greatest levels of verbal aggression, whereas Latino men displayed lower levels of articulate speech and greater levels of antagonism.

When taken together, content analytic research suggests that the manner in which the majority of Latinos are depicted in English-language media reflects several longstanding media stereotypes (Ramirez Berg, 2002). These images include representations of Latinos as criminals and deviants; as sexual provocateurs and objects of sexual desire; and as dimwits and targets of ridicule. Consequently, it is not surprising that Latinos in the United States are increasingly turning to the Spanish-language networks, but what can viewers expect to encounter when tuning in to the offerings on Spanish-language television?

Portrayals on U.S. Spanish-Language Television

To date, only one empirical study documents the content found on Spanish-language television (see Glascock & Ruggiero, 2004). In their analysis of three Spanish-language networks (Univision, Telemundo, and Azteca America), Glascock and Ruggiero (2004) provide preliminary insights into the Spanish-language television landscape. In so doing, their findings make a notable contribution by offering an initial look into the images provided in the Spanish-language, prime-time landscape. Specifically, Glascock and Ruggiero (2004) report that males (52%) slightly outnumbered females. Women were identified as younger, tended to have lighter skin and hair color, and were more provocatively dressed than men. In terms of occupations, men held more professional jobs whereas women held more support and service jobs, such as maids and waitresses. Moreover, women were portrayed as having more parental responsibilities than men; a finding they suggest is indicative of the tendency for media content to provide representations that maintain traditional gender and sex roles.

Among both sexes, lighter-skinned characters were more likely to be in major roles and of higher socioeconomic status; with darker-skinned characters more likely to be in supporting roles and of lower socioeconomic status (Glascock & Ruggiero, 2004). Additionally, lighter-skinned characters were depicted as younger than darker-skinned characters. The authors posit that these results, privileging lighter-skinned characters, point to existing real-world biases within the Latino communities that favor a fair complexion.

Although Glascock and Ruggiero's (2004) results are noteworthy, features of the design suggest that continued research is in order. First, the lack of a random sample of programming limits the ability to confidently generalize from their findings (Neuendorf, 2002). Second, inter-coder reliabilities do not consistently meet conventionally identified levels of acceptability (Krippendorff, 2002). Last, the categorical nature of many of the variables restricts analytical options. Accordingly, the present study addresses each of these points in order to expand on their foundational work.

Theoretical Implications

As the content analytic studies of U.S. and Spanish-language television networks indicate, Latino viewers attending to either type of media have the potential to be exposed to skewed representations as they relate to portrayals of different groups. When it comes to U.S. television programs, Latino viewers are likely to encounter negative or stereotypic representations of their ethnic group. When exposed to Spanish-language media, viewers may witness certain social groups (i.e., men, wealthy, young, & light skin characters) as dominant over other groups. Of course, it is not the duty of the networks to provide programming that strictly replicates or improves on the social world. Nonetheless, the potential outcomes of viewing should not be ignored. Social identity theory (Tajfel & Turner, 1979) provides an appropriate framework to assess the possible effects of exposure to these images on audience members' social identities and future intergroup interactions.

Social Identity Theory and the Media

Social identity theory (SIT) posits that people's identity is derived (in part) from their membership in different groups, which offer norms of thinking and behaving (Hogg & Abrams, 1988). Through the process of categorization, individuals psychologically compartmentalize people into groups in relation to the self such that people are identified as either in-group or out-group members (Abrams & Hogg, 1990; Brown, 1995; Hogg & Abrams, 1988). When the in-group is favorably compared to an out-group, self-concept and esteem are maintained and enhanced (Brown, 1995; Hogg & Abrams, 1988). Alternatively, self-esteem may be weakened when membership in a subordinate group confers a negative social identity (Abrams & Hogg, 1990; Hogg & Abrams, 1988). Such intergroup comparisons can be made on the basis of sex (Ambady, Paik, Steele, Owen-Smith, & Mitchell, 2004), complexion (Uhlmann, Dasgupta, Elgueta, Greenwald, & Swanson, 2002), age (Harwood, Giles, & Ryan, 1995), race/ethnicity (Mastro, 2003), and a variety of social groups and/or categories.

When a group fails to provide members with positive group distinctiveness, individuals may engage in the process of comparing a subgroup membership with other relevant subgroups on dimensions that are more advantageous (Gaertner & Dovidio, 2000). For instance, Uhlman et al. (2002) found that Latinos held more favorable attitudes toward light-skinned Latinos than toward dark-skinned Latinos. However, these differences did not arise when the overarching group (i.e., both light- and dark-complexion Latinos) was compared to Whites.

Notably, research has revealed that television messages can become incorporated into the ongoing processes of social comparison and identity formation (Harwood & Roy, 2005; Mastro, 2003). For example, in terms of depictions of ethnicity on English-language television, Latinos may perceive a negative status associated with their in-group, as their televised counterparts are often confined to unfavorable portrayals (Harwood & Anderson, 2002; Mastro & Behm-Morawitz, 2005). Upon failing to achieve positive group distinctiveness, Latino viewers may seek out television messages that provide positive group distinctiveness, possibly turning to the Spanish-language television networks.

In support of this identity-based viewing strategy, Harwood (1997; 1999) found that individuals preferred television messages that substantiated their social identities and provided grounds for the establishment of favorable group-based distinctions. Additional anecdotal evidence for such viewing preferences can be gleaned from the fact that the Spanish-language television networks are enjoying rapidly increasing popularity among bilingual Latinos, a demographic that has the opportunity to select between English- and Spanish-language television media (Barnes & Jordan, 2005). The social identity-based implications here are that in addition to merely preferring this content, audience members may learn the normative attributes and social positions of different groups in society. Thus, in attempts to reinforce and enhance identity needs through the use of Spanish-language media, Latino viewers may further acquire knowledge about group norms, social standing, and the like.

However, the imagery presented on Spanish-language television may not be offering entirely evenhanded portrayals of different groups (Glascock & Ruggiero, 2004); particularly in terms of strongly held gender-role attitudes (Mirande, 1997) which define men as strong, courageous, aggressive, emotionally detached breadwinners and women as passive, obedient, self-sacrificing homemakers (Gonzalez, 1982; Soto, 1983). That is, Spanish-language television may be teaching, or reinforcing, the group norms associated with each sex. Specifically, men and women may see different characteristics and attributes associated with their on-air counterparts and incorporate these into their cognitions about the two sexes. Given findings from Rivadeneyra and Ward (2005) revealing female Latino adolescents' television exposure to be positively associated with more traditional gender role attitudes (suggesting that viewers indeed learn about their own in-groups via media consumption), the importance of documenting the imagery presented in Spanish-language television is underscored. As such, the present study was designed to investigate this content. In particular, consistent with findings from existing research, the following hypotheses were developed regarding representations of men and women in Spanish-language television programming.

[H.sub.1a]: More men than women will hold professional occupations.

[H.sub.1b]: More women than men will be found working as homemakers.

It is important to recognize, however, that individuals are members of a variety of different groups (Gaertner & Dovidio, 2000) and identification with each of these groups (both in isolation and in combination) is likely to affect the messages they attend to and learn from the media (Harwood & Roy, 2005). For instance, an older male television viewer may not only learn norms associated with his sex from television, but also those associated with his age group. Moreover, attention and identification may be enhanced by the in-group specificity of the television model (in this example, an "older male" character as opposed to simply a "male" character or an "older" character). Therefore, for content analytic research to appropriately and effectively address the potential influence of exposure, consideration also should be paid to the intersections of different group memberships in terms of how they are represented. Accordingly, the present study extends its focus beyond merely documenting representations of men and women, but additionally examines the intersection of sex with age, skin complexion, and socioeconomic status. Importance is given to these groups as research has revealed each of these categories to be associated with meaningful group-based norms and values that may be associated with biases in intergroup attitudes and beliefs (see Gilens, 1996; Harwood et al., 1995; Uhlmann et al., 2002). To explore this issue, the following research question was developed.

[RQ.sub.1]: How do age, complexion, and socioeconomic status influence the manner in which male and female characters are represented on television?

Although tests of the proposed link between exposure to Spanish-language television and intergroup cognitions will require experimental examination, the present study offers an important first step by documenting the nature of the images that are likely to influence social identity and intergroup behaviors. Only by first empirically analyzing the portrayals depicted in Spanish-language television, can effects studies be formulated to assess the implications of exposure.

Method

A composite week of evening television (7-11 p.m. Monday-Sunday) was constructed from a random sample of programming across the four major Spanish-language networks (Azteca America, Telefutura, Telemundo, & Univision) using a random numbers table. The sample was recorded from October to December 2004. All programs were analyzed intact; accordingly, if a program began before 7 p.m. or ended after 11 p.m., it was included in its entirety.

Two student coders (one male and one female), fluent in Spanish, were trained on programming outside the actual sample until acceptable levels of inter-coder reliability were achieved. Although none of the specific episodes used for coder training or reliability assessments were contained in the actual sample, all programs selected for training were recorded from the four major U.S. Spanish-language networks. Consequently, these same programs (not episodes) appear in the final sample randomly selected for evaluation. Coders received a total of approximately 30 hours of joint training conducted by this study's authors over an 8-week period. Alongside these formal sessions, 2-4 hours per week of practice coding trials were required. Once appropriate levels of reliability were stably attained, evaluations of the actual sample began, with programs randomly assigned to each coder. Reliabilities were computed using Krippendorff's alpha and are reported individually below. These were assessed based on four, randomly selected episodes, which provided 39 units of analysis (i.e., characters) for each of the coders to evaluate.

Coders evaluated only the non-dubbed, scripted, fictional, entertainment programming within this random sample. Accordingly, programs such as telenovelas, comedies, parody shows, family dramas, courtroom/crime dramas, medical dramas, etc., would be viable to code, whereas reality-based programs and news magazine formats would be excluded from analyses.

Only the main characters were analyzed both at the character level and at the interaction level. Main characters were defined by playing an integral role in the development of the story line for the particular episode. At the character level, measures of the characters' attributes were based on the entire program. Alternatively, at the interaction level, only the character's first substantial interaction was coded to assess the relative position of the character. Consistent with existing research, a substantial interaction was defined as a conversation between at least two characters involving at least three turns (Mastro & Greenberg, 2000).

Character Level Attributes

Demographic information (i.e., sex, age, occupation, and income level) was coded for all main characters. Gender was the character's biological sex ([alpha] = .95) of either male or female. Age ([alpha] = .93) was recorded on a 5-point scale from 20s or younger(1) to 50s or older(5). Although categorizing age in this manner limits the ability to make more subtle distinctions between characters within the youngest as well as the oldest categories, it provides the most effective grouping to statistically assess the question s of interest to the present study. Occupation ([alpha] = .85) was defined as the character's primary responsibility, including: professional/business person, service provider, government/legal position (e.g., lawyer, judge, police officer), homemaker, or other. Income level ([alpha] = .80) of the character was rated from little/no wealth (1) to well-off/rich (5).

Physical attributes also were coded on a 5-point scale for all main characters. Body type ([alpha] = .80) indicated the character's physical fitness ranging from excessively thin (1) to obviously obese (5). Complexion ([alpha] = .89) was coded to indicate the character's skin tone such that fair/pale skin was coded as (1) and dark skin was coded as (5). Physical attractiveness ([alpha] = .85) was based on the extent to which the character was deemed generally physically unattractive (1) or appealing to others (5).

A character's speech was assessed on a 5-point scale ranging from inarticulate to articulate ([alpha] = .85). Inarticulate characters were unable to convey ideas to others based on their communication skills (1), whereas articulate characters were adept at clearly communicating their ideas to others (5). Motivation ([alpha] = .85) was measured to assess the character's initiative. A lazy character was wholly inactive and lacked drive or initiative to engage in any activity (1), and a motivated character was very active and exemplified great initiative (5). A character's level of respect ([alpha] = .87) also was coded ranging from ridicule (1 = target of social derision and criticism) to respect (5 = highly esteemed).

A character's intelligence ([alpha] = .93) ranged from unintelligent (1) to intelligent (5). Unintelligent characters displayed a lack of basic reasoning skills, whereas intelligent characters displayed intellectualism and advanced cognitive reasoning. Flirtatious behavior ([alpha] = .80) was coded according to the extent to which the character was coy and modest (1) versus enticing and playful (5) in social situations. Finally, characters' social authority ([alpha] = .89) and job authority ([alpha] = .93) also were coded. A character was of low social authority (1) if he/she received social advice from other characters and was of high social authority (5) if he/she gave social advice to others. Similarly, a character was deemed low in job authority (1) if he/she was subservient to others in the workplace and high in job authority (5) if he/she is in a position of control or power in the work environment.

Interaction Level Characteristics

Interaction level variables measured the features present in each main character's first substantial interaction. Conversation topic (e.g., health, love, politics, business, etc. in a 14-topic category scheme) was coded for all interactions ([alpha] = .97). Dominance ([alpha] = .84) in the interaction was coded on a 5-point scale (1 = character is subservient or conformed easily to the will of others; 5 = character is authoritative and assertive in nature). The physical aggressiveness ([alpha] = .90) of characters was assessed according to the extent to which the character used physical force, with (1) representing the lowest level of force (or absence thereof) and (5) indicating the highest degree of unnecessary physical aggression. Similarly, verbal aggression ([alpha] = .85) referred to the extent to which the character used hostile and abusive language, with a score of (1) designating the lowest degree of verbal aggression and a (5) indicating the highest.

The degree to which the character appeared to be a sex object ([alpha] = .89) in the interaction was measured on a 5-point scale indicating whether the character did not exhibit sexual objectification (1), or was strongly considered a sex object (i.e., his/her primary social role involved providing sexual pleasure to others present in the interaction; coded as 5). Sexual aggression ([alpha] = .85) identified the extent to which the character initiated sexual engagements on a 5-point scale ranging from not sexually aggressive (1) to actively aggressive sexually (5). Finally, in terms of attire ([alpha] = .90), conservative dress (1) was unassuming and revealed no skin whereas provocative dress (5) revealed excessive skin or was notably tightly fitted.

Analyses

In order to assess differences in depictions based on the biological sex of the character, chi-squares and t-tests were executed depending on the level of measurement of the variables. For further insights into findings from chi-squares, adjusted standardized residuals also were examined. In addition, because theory suggests that varying representations of group memberships have the potential to differentially influence identification and social identity concerns, 2 x 2 ANOVAs were used to examine the extent to which sex interacted with age, income level, and skin tone, in predicting the manner in which characters were depicted in Spanish-language programming.

Results

This 1-week sample of Spanish-language programming yielded a total of 36 scripted, fictional, entertainment programs; containing 308 main characters involved in substantial interactions. The remainder of the programs randomly sampled to construct the composite week fell outside the programming genres of interest. Each of the coded programs was identified as an hour-length, regularly occurring, soap-opera/telenovela. Fifteen were aired on Telemundo, 10 on Azteca America, another 10 on Univision, and one on Telefutura. (2)

Among the characters depicted on these shows, 40.6% (n = 125) appeared on Telemundo, 33.8% (n = 104) on Univision, 22.7% (n = 70) on Azteca America, and 2.9% (n = 9) on Telefutura. Most commonly, characters were identified in their 20s (45.5%, n = 140), with the slight majority of characters found to be female (52.3%, n = 161). T-tests revealed the women (M = 2.89, SD = 1.22) to be significantly younger (t= 2.18, df= 308, r = .12, p < .05) than the men (M = 3.19, SD= 1.16).

Characters were most often depicted discussing domestic issues/family matters (n = 111, 36.0%). Although no specific occupation predominated (as the most commonly noted occupation was that of "other"), characters were found most frequently in business/professional positions (n = 43, 14.0%) followed by jobs in service positions (n = 23, 7.5%), and occupations in the legal system (n = 11, 3.6%). Additionally, chi-squares (excluding the category of "other") revealed significant differences in occupation based on the sex of the character ([chi square] = 25.82, df = 3, Cramer's V = .54, p < .01), with men more frequently found in business/professional positions (n = 31, 58.5%) and government/legal positions (n = 11, 20.8%) than women (n = 12, 33.3% and n = 0, respectively). Women were more often found to be homemakers (n = 11, 30.6%) and in service positions (n = 13, 36.1%) compared with their male counterparts (n = 1, 1.9% and n = 10, 18.9%, respectively). Examinations of the adjusted standardized residuals revealed significant differences from expected values for men (2.3) and women (-2.3) on the category of business/professional occupations. Differences from expected values for men (2.9) and women (-2.9) also emerged for government/legal positions. Finally, men (-3.9) and women (3.9) were revealed to differ significantly from expected values on the category of homemaker. Significant differences were not revealed for service occupations.

Because the precise nature of media representations has the potential to influence aspects of social identity, analyses were next conducted to specifically examine the extent to which sex interacted with age, socioeconomic status, and skin tone in predicting character attributes and interaction characteristics. Thus, 2 x 2 analyses of variance were used. Accordingly, median splits were created for the variables age, income level, and complexion, to allow for use as predictors in these two-way ANOVAs.

Character Level Attributes

Age. In examinations of social authority, a significant sex by age interaction was revealed F(1, 80) = 6.1 7, [[eta].sup.2] = .06, p < .025, indicating that older men demonstrated the highest level of social authority whereas older women exhibited the lowest level of social authority (see Table 1 for descriptives). This difference was significant by simple effects tests F(1, 56) = 23.11, p < .01. The same pattern was revealed in assessments of job authority F(1, 80) = 5.01, [[eta].sup.2] = .05, p < .05. Older men were characterized with the highest level of job authority and older women were depicted with the lowest level of job authority (Table 1). This difference was again significant by simple effects tests F(1,56) = 18.75, p < .01.

No interaction effects were revealed in the sex by age analyses of variance examining body type F(1, 294) = 0.00, p > .05; attractiveness F(1, 294) = 0.16, p > .05; articulation F(1,294) = 0.04, p > .05; motivation F(1, 294) = 0.48, p > .05; intelligence F(1, 293) = 2.50, p > .05; level of respect F(1, 294) = 1.11, p > .05; or flirtatiousness F(1, 293) = 0.01, p > .05.

Income Level. Examinations of the influence of sex and income level on the manner in which characters were depicted on Spanish-language programming demonstrated disparities in terms of the body type of the character. Specifically, ANOVA findings revealed a significant sex by income level interaction for body type F(1, 300) = 3.83, [[eta].sup.2] = .01, p = .05, such that women with higher income levels were depicted with the slimmest physique whereas men with the highest income level were portrayed with the heaviest body type. Simple effects tests revealed this difference to be significant F(1, 141) = 12.71, p < .01. Descriptive statistics can be found in Table 1. No other sex by socioeconomic status differences emerged.

Complexion. Analysis of variance results revealed significant sex by complexion interactions in predicting both intelligence F(1, 295) = 6.08, [[eta].sup.2] = .02, p < .01 and articulation F(1, 295) = 3.70, [[eta].sup.2] = .01, p = .055. On the measure of intelligence, simple effects tests revealed men with a fair complexion to be significantly more intelligent than women with a fair complexion F(1, 136) = 5.40, p < .025. See Table 1 for descriptive statistics. In terms of articulate speech, men with a fairer skin tone were judged most articulate whereas fair-skinned females were deemed least articulate. This difference was significant by simple effects tests F(1, 136) = 5.11, p = .025. No other significant interactions emerged for examinations of the influence of sex and complexion on character attributes.

Interaction Level Characteristics

Age. No significant findings emerged on examinations of the influence of sex and age on any of the interaction characteristics. This included: dominance F(1, 293) = 0.01, p > .05; physical aggression F(1, 70) = 1.13, p > .05; verbal aggression F(1, 70) = 0.34, p > .05; sexual objectification F(1, 70) = 0.01, p >.05; sexual aggression F(1, 70) = 0.00, p > .05; or attire F(1, 70) = 1.94, p > .05.

Income Level. The socioeconomic status of characters interacted with sex in predicting the extent to which characters were depicted wearing provocative attire F(1, 80) = 21.15, [[eta].sup.2] = .22, p < .01. Specifically, men of higher income levels were found to dress most conservatively whereas women of a higher income level dressed the most provocatively. This difference was statistically significant based on simple effects tests F(1, 48) = 23.44, p < .01. See Table 1 for descriptive statistics. Income level and sex also interacted in predicting dominance F(1, 80) = 6.86, [[eta].sup.2] = .07, p < .01 such that men of higher socioeconomic standing were the most dominant but women in the upper income range were most submissive. Simple effects tests revealed this difference to be significant F(1, 48) = 31.38, p < .01. No other significant findings on income level were revealed.

Complexion. Sex and complexion interacted in predicting evaluations of verbal aggression F(1, 298) = 4.54, [[eta].sup.2] = .01, p < .05. In particular, men with darker complexions were deemed significantly more verbally aggressive than women with darker complexions F(1, 160) = 8.83, p < .01 based on simple effects tests. See Table 1 for descriptive statistics. No other significant ANOVA results emerged for examinations of the influence of sex and complexion on interaction characteristics.

Discussion

Research investigating the relationship between media exposure and subsequent perceptions about the characteristics of different groups supports the assertion that television viewing has a measurable influence on consumers' cognitions (Armstrong, Neuendorf, & Brentar, 1992; Mastro & Kopacz, 2006). This is consequential when considering the findings from the present study, which reveal that different formulas are used when depicting men and women in Spanish-language programming. Consistent with results from Glascock and Ruggiero (2004), although a nearly equivalent number of male and female characters were represented, women tended to be younger than their male counterparts. Men also appeared in business/professional settings more frequently, whereas women were more often seen in roles as homemakers. The correspondence between these outcomes and those from Glascock and Ruggiero (2004) illustrates some stability in the Spanish television offerings, at least in this regard.

Although the documentation of these purely gender-based differences is meaningful, it is additionally valuable to identify variations in gender portrayals based on intersections with important group memberships (i.e., age, socioeconomic status, and skin tone). This serves two primary purposes. First, offering greater detail regarding the content of Spanish-language television allows for more meaningful comparisons to be drawn between Spanish- and English-language television fare. Second, by providing greater specificity in terms of the characterizations of men and women in Spanish-language television, the manner in which viewers' identity-based outcomes may be affected can be better understood.

Spanish- Versus English-Language Television

Given that a mere 6.5% of the English-language television population is Latino (Children NOW, 2004), for Spanish-speaking Latino viewers looking to find images of their ethnic in-group the obvious choice is Spanish-language television. Despite this difference in sheer frequency, however, several of the characterizations consumers might encounter on Spanish television appear to parallel those on English-language television. Consistent with portrayals of Latinos (Mastro & Behm-Morawitz, 2005) and Whites (Signorielli & Bacue, 1999) on English-language television, youth is emphasized. Indeed, few portrayals of older age groups appear in either programming outlet. In addition, whether viewing Spanish- or English-language television, the important role of the family in the lives of Latinos is emphasized. Results from the current study reveal that characters on Spanish-language television were most frequently engaged in discussions related to family matters; with women commonly serving in roles as homemakers. Similarly, when depicted on English-language television, Latino (as well as White) characters are regularly seen as family members (Mastro & Behm-Morawitz, 2005). However, in terms of topics of conversation found on English-language television, crime ranks at the top of the list for Latino men (Mastro & Behm-Morawitz, 2005). From a social identity perspective, it would not be unreasonable to argue that in order to manage identity needs, Latino viewers may migrate toward Spanish-language television simply to avoid the negative focus on criminality that often accompanies portrayals of their ethnic group in English-language media.

Similarities between Spanish- and English-language television are also revealed when examining a variety of character-based attributes. Findings from the present study indicate that older women display the lowest levels of social authority on Spanish-language television, whereas older men exhibit the highest. This is not unlike what viewers can expect to encounter when watching English-language television; with Latino female characters demonstrating lower levels of social authority than their White counterparts (Mastro & Behm-Morawiz, 2005). In terms of job authority much the same pattern emerges. Consistent with Spanish-language programming, on English-language television older male characters are seen more often in the workplace (Signorielli & Bacue, 1999) and are more authoritative in the workplace than are female characters (Mastro & Behm-Morawiz, 2005). From a social identity perspective, the implications of these findings for Latino women are particularly consequential. Regardless of whether they attend to English- or Spanish-language television, Latino women encounter messages that have the potential to diminish their status both in terms of social authority as well as with regard to job authority. This may be of even greater concern for older women whose on-air counterparts exhibited the lowest levels of both social and job authority on television. Accordingly, exposure to these subordinate images of one's in-group may ultimately have a negative influence on the self-concept of women viewers (older women in particular) due to the inability to achieve positive group distinctiveness.

Spanish-Language Television and Possible Identity-Based Outcomes

This study's results reveal male characters with lighter skin tone to be more intelligent, articulate, and verbally aggressive than their darker on-air counterparts. Alternatively, women with lighter skin tone were less intelligent, less articulate, and more verbally aggressive than were women with darker skin tones. These findings partially support the notion that a preference exists for lighter skin tone in Latin America and among Latinos in the United States (e.g., Farley, 1999; Wade, 1997). Thus, Latino viewers may infer from Spanish-language television that men with fair skin enjoy a greater potential for success and accomplishment in life due to the inherent possession (or at least the attainability) of these desirable features. However, the disparity between men and women in terms of the attributes associated with lighter complexions leaves open the possibility that assumptions based on skin tone may not universally hold.

In terms of the potential for Spanish-language television content to reinforce traditional gender roles, which define Latino men as tough, assertive, and controlled whereas women are submissive and deferential (Gonzalez, 1982; Soto, 1983), some support for this assertion was revealed. Wealthy men were characterized with the heaviest body type, the most conservative attire, and the most dominant interaction style. Alternatively, wealthy women were the slimmest, most provocatively dressed, and the most submissive characters on Spanish-language television. Although not wholly representative of traditional Latino sex roles, these findings appear to suggest somewhat of a reliance on these norms, particularly among those who have achieved economic success in life. Accordingly, from a social identity perspective, consumption may lead to the reinforcement of these attributes among viewers who identify with the characters. Furthermore, viewers aspiring to similar ends may attempt to emulate these features.

Taken together, the findings from the current study provide a picture of the overarching imagery viewers are likely to encounter when tuning in to Spanish-language television. From the perspective of social identity theory, such data are meaningful as they offer insights into the possible influence of exposure on group-based outcomes. From this perspective, one might expect that Spanish-language television viewers may begin to associate features such as power, social influence, career success, intelligence, and competence, with men--particularly older men. Conversely, for women, intelligence and articulate speech may come to be considered secondary to qualities such as attractiveness and youth.

Although it is recognized that the media are only one of myriad factors that contribute to the development and maintenance of group identities, when considering that Spanish-language media acts both as a source of information for Latinos as well as a resource to promote positive social identity ("Making," 2006), its importance should not be ignored. Indeed, empirical research substantiates the claim that media messages become incorporated into the process of identity formation, indicating that audience members use media to create environments that support their group identity; oftentimes beyond that which exists in society (Harwood, 1999). Given the heightened state of ethnic and racial awareness provoked by recent immigration reform proposals in the United States, the importance of the media in the lives of U.S. Latinos may be particularly consequential at present. Of course, linking media exposure with viewer outcomes will require effects studies which include both audience level as well as content level characteristics (Shoemaker & Reese, 1996), however, the results provided here offer the necessary first step in this research process.

References

Abrams, D., & Hogg, M. A. (1990). An introduction to the social identity approach. In D. Abrams & M. Hogg (Eds.), Social identity theory: Constructive and critical advances (pp. 1-9). Hemel Hempstead, England: Harvester Wheatsheaf.

Ambady, N., Paik, S. K., Steele, J., Owen-Smith, A., & Mitchell, J. P. (2004). Deflecting negative self-relevant stereotype activation: The effects of individuation. Journal of Experimental Social Psychology, 40, 401-409.

Armstrong, G., Neuendorf, K., & Brentar, J. (1992). TV entertainment, news, and racial perceptions of college students, Journal of Communication, 42, 153-176.

Azteca America (2005). Azteca America. Retrieved September 26, 2005, from http://www.aztecaamerica.com/corporate/

Barnes, B., & Jordan, M. (2005, May 2). "Big four" TV networks get a wake-up call--in Spanish. Wall Street Journal, p. B1.

Becker, A. (2005). Spanish language expansion: New broadcast shows include novelas, dramas, sports and news. Broadcasting & Cable, 135, 32.

Brown, R. (1995). Prejudice: Its social psychology. Oxford, England: Blackwell. Children NOW (2004). Fall colors 2003-04: Prime-time diversity report. Retrieved November 5, 2004, from http://www.childrenandmedia.org

Consoli, J. (2005). Latino nets add daypart. MediaWeek, 15, 8-10.

Downey, K., (2005a). Little Azteca grows up: Nielsen status means it now plays with the big boys. Broadcasting & Cable, 135, 20.

Downey, K. (2005b). Look who's talking: Usually reticent Univision finally starts to tell its story. Broadcasting & Cable, 135, 20.

Faber, R. J., O'Guinn, T. C., & Meyer, T. P. (1986). Diversity in the ethnic media audience: A study of Spanish language broadcast preference in the U.S. International Journal of Intercultural Relations, 10, 347-359. Farley, C. J. (1999, July 5). Latin music goes pop. Time, 153, 75-79.

Gaertner, S., & Dovidio, J. (2000). Reducing intergroup bias: The common ingroup identity model. Hove, England: Psychology Press.

Gerbner, G., Gross, L., Morgan, M., & Signorielli, N. (1980). The "mainstreaming" of America: Violence profile no. 11. Journal of Communication, 30, 212-231.

Gilens, M. (1996). Race and poverty in America: Public misperceptions and the American news media. Public Opinion Quarterly, 60, 515-541.

Glascock, J., & Ruggiero, T. E. (2004). Representations of class and gender on primetime Spanish language television in the United States. Communication Quarterly, 52, 390-402.

Gonzalez, A. (1982). Sex roles of the traditional Mexican family: A comparison of Chicano and Anglo students' attitudes. Journal of Cross-Cultural Psychology, 13, 330-339.

Greenberg, B., Mastro, D., & Brand, J. (2002). Minorities and the mass media: Television into the 21st century. In J. Bryant & D. Zillmann (Eds.), Media effects: Advances in theory and research (pp. 333-351). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Grillo, J. B., & Bednarski, P. I. (2004). Advertisers' Spanish lesson. Broadcasting & Cable, 134, 22-24.

Harwood, J. (1997). Viewing age: Lifespan identity and television viewing choices. Journal of Broadcasting & Electronic Media, 41, 203-213.

Harwood, J. (1999). Age identification, social identity gratifications, and television viewing. Journal of Broadcasting & Electronic Media, 43, 123-136.

Harwood, J., & Anderson, K. (2002). The presence and portrayal of social groups on prime-time television. Communication Reports, 15, 81-97.

Harwood, J., Giles, H., & Ryan, E. B. (1995). Aging, communication, and intergroup theory: Social identity and intergenerational communication. In J. F. Nussbaum & J. Coupland (Eds.), Handbook of communication and aging research (pp. 133-159). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Harwood, J., & Roy, A. (2005). Social identity theory and mass communication research. In J. Harwood & H. Giles (Eds.), Intergroup communication: Multiple perspectives (pp. 189-211). New York: Peter Lang.

Hogg, M. A., & Abrams, D. (1988). Social identifications: A social psychology of intergroup relations and group processes. New York: Routledge.

Krippendorff, K. (2002). Content analysis: An introduction to its methodology. Beverly Hills, CA: Sage.

Making their feelings known. (2006, April 1). Economist, 378, 26-27.

Mastro, D. (2003). A social identity approach to understanding the impact of television messages. Communication Monographs, 70, 98-113.

Mastro, D., & Behm-Morawitz, E. (2005). Latino representation on primetime television. Journalism and Mass Communication Quarterly, 82, 110-130.

Mastro, D., & Greenberg, B. (2000). The portrayal of racial minorities on prime time television. Journal of Broadcasting & Electronic Media, 44, 690-703.

Mastro, D., & Kopacz, M. (2006). Media representations of race, prototypicality, and policy reasoning: An application of self-categorization theory. Journal of Broadcasting & Electronic Media, 50, 305-322.

McAnany, E. G., & La Pastina, A. C. (1994). Telenovela audiences: A review and methodological critique of Latin American research. Communication Research, 21, 828-849.

Mirande, A. (1997). Hombres y machos: Masculinity and Latino culture. Boulder, CO: Westview.

Neuendorf, K. A. (2002). The content analysis guidebook. Thousand Oaks, CA: Sage.

Nielsen Media Research (2005). Hispanic-American television audience. Retrieved August 28, 2005, from http://www.nielsenmedia.com/ethnicmeasure/hispanic-american/ index Hisp.html

Ramirez Berg, C. (2002). Latino images in film: Stereotypes, subversions, and resistance. Austin: University of Texas Press.

Rivadeneyra, R., & Ward, L. M. (2005). From Ally McBeal to Sabado Gigante: Contributions of television viewing to the gender role attitudes of Latino adolescents. Journal of Adolescent Research, 20, 453-475.

Shoemaker, P., & Reese, S. (1996). Mediating the message: Theories of influences on mass media content (pp. 253-260). White Plains, NY: Longman.

Signorielli, N., & Bacue, A. (1999). Recognition and respect: A content analysis of prime-time television characters across three decades. Sex Roles, 40, 527-544.

Soto, E. (1983). Sex-role traditionalism and assertiveness in Puerto Rican women living in the United States. Journal of Community Psychology, 11, 346-354.

Stillig, E. A. (1995). The history of Spanish language television in the United States and the rise of Mexican international syndication strategies in the Americas. The Howard Journal of Communications, 6, 231-249.

Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In S. Worchel & W. G. Austin (Eds.), The social psychology of intergroup relations (pp. 33-47). Monterey, CA: Brooks-Cole.

Telemundo Communications Group (2005). Informacion corporativa. Retrieved September 26, 2005, from http://www.telemundo.com/telemundo/2449852/detail.html

Telenovelas have lots of fans at other networks (February 24, 2006). Los Angeles Times. Retrieved February 24, 2006, from http://www.latimes.com/entertainment/news/tv/ la-fi-telenovela23feb23,1,83094.story?coll=la-entnews-tv

Television en fuego (February 17, 2006). Los Angeles Times. Retrieved February 17, 2006, from http://www.latimes.com/news/printedition/opinion/ la-ed-spanish17feb17,1,4061267.story?ctrack=1&cset=true

Uhlmann, E., Dasgupta, N., Elgueta, A., Greenwald, A. G., & Swanson, J. (2002). Subgroup prejudice based on skin color among Hispanics in the United States and Latin America. Social Cognition, 20, 198-225.

Univision Communications (2005). Univision. Retrieved September 26, 2005, from http://www.univision.net/corp/en/index.jsp

U. S. Census (2000). Retrieved May 1, 2005 from http://factfinder.census.gov/servlet/SAFF People?_sse=on

Wade, P. (1997). Race and ethnicity in Latin America. London: Pluto Press.

Notes

(1) Telenovelas, similar to U.S. soap operas, air on a daily basis, but have a definitive ending-point, after 20 weeks or approximately 150 episodes (Downey, 2005b; McAnany & La Pastina, 1994). Univision and Telemundo differ in their approach to producing and airing this particular genre. Although the telenovelas featured in Univision's programming are produced in Mexico, Telemundo has increasingly invested resources in the recruitment of producers, writers, and actors to create novelas depicting the Latino experience in the United States (Becker, 2005).

(2) As an Univision Communications--owned and -established network, Telefutura's programming strategy is to offer alternate genres from those being offered on the other major Spanish-language networks for every daypart (Univision, 2005). For instance, while Univision and Telemundo broadcast telenovelas in prime-time programming, Telefutura airs mainstream, Hollywood movies. In doing so, Univision prevents losing its own audience to Telefutura while expanding its viewership.

Dana E. Mastro (Ph.D., Michigan State University) is an Associate Professor in the Department of Communication at the University of Arizona. Her research interests include media effects and racial/ethnic stereotyping.

Michelle Ortiz (M.A., University of Arizona) is a doctoral student in Communication at the University of Arizona. Her research interests include intergroup processes and media effects.

Table 1
ANOVA Descriptive Statistics for Sex by Age, Income Level,
and Complexion Interactions on Character Attributes
and Interaction Characteristics

                                           Men               Women

Dependent Variable                   N     M      SD    N     M      SD

Social Authority     Younger         13   3.77   0.83   13   3.46   0.78
                     Older           36   4.06   0.92   22   2.67   0.65
Job Authority        Younger         13   3.46   0.78   13   3.15   0.80
                     Older           36   4.31   0.95   22   3.00   0.74
Body Type            Higher Income   71   2.37   0.57   72   2.06   0.47
                     Lower Income    74   2.22   0.71   87   2.17   0.59
Attire               Higher Income   32   1.13   0.34   18   2.13   0.99
                     Lower Income    17   1.35   0.49   17   1.18   0.39
Dominance            Higher Income   32   4.69   0.47   18   3.63   0.52
                     Lower Income    17   4.47   0.51   17   4.12   0.60
Intelligence         Dark Skin       92   3.95   0.72   69   4.07   0.60
                     Fair Skin       51   4.02   0.55   87   3.77   0.64
Articulation         Dark Skin       92   4.29   0.72   69   4.33   0.56
                     Fair Skin       51   4.57   0.54   87   4.31   0.70
Verbal Aggression    Dark Skin       92   1.58   0.70   70   1.29   0.49
                     Fair Skin       51   1.63   0.66   89   1.66   0.69