Female Roles in Television Advertising: Viewers' Use of Gender Role Cues in Appraising Stereotypic and Non-Stereotypic Role Portrayals Richard H. Kolbe, Washington State University, Washington Carl D. Langefeld, Indiana University, Indiana The study uses the Bem Sex-Role Inventory (BSRI) as both a self-rating and projective scale to predict viewer responses to stereotypic and non-stereotypic role portrayals in television conlmercials. Projective BSRI ratings of ad characters were significant predictors ofperceptual judgments about the ad character, advertisement, and product.
Differences between self-ratings and projective character ratings on the BSRI were also significant predictors of the ad perceptual judgments. Directions for future research in examining role stereotyping in advertising are offered. INTRODUCTION The depiction of female roles in television advertising has raised a number of provocative research questions. Research in this area has been fostered by the observations made by media analysts regarding the inconsonance of fenlale role portrayals relative to social norms.
Supporting these observations have been numerous content analyses which have pointed to the small number, poor quality, and limited breadth of roles afforded female characters in the medium relative to those held by females in real life (Courtney and Whipple 1974; Dominick and Rauch 1972; Gilly 1988; McArthur and Resko 1975; O'Donnell and O'Donnell 1978; Scheibe 1979; Schneider and Schneider 1979). The evidence suggests that advertisers have often used portrayals which can be labeled stereotypic female roles (e. g. , female as housewife, female as subservient to a male) as opposed to non-stereotypic roles (e. . , female as athlete, leader, business person). While the content of female roles in television advertising is well understood, the factors which influence viewers' perceptions of these roles has received less research attention. Central to this issue is the determination of which factors explain viewer responses to role portrayals. In addition, the implications of such judgments on perceptions about the advertisement and advertised product need to be considered. The main issue addressed in the current study is how viewers respond to stereotypic (8) and non-stereo typic (NON-S) role portrayals.
The basis of such responses is related to the manner in which an individual processes gender-related information -- a process likely rooted in an individual's own level of masculinity and femininity. If this relationship holds, then masculinity and femininity self ratings should be predictors of viewer perceptions of role portrayals and related attitudes toward the ad character, product, and the ad itself. Gender Processing The use of gender-related information to process and interpret stimuli is a substantial component of cognitive processing (Bern 1985).
From early in life, individuals categorize people, objects, and behaviors as masculine and feminine, usually with prescriptions as to their appropriateness for the individual's own gender (Bandura 1977; Fein et al. 1975; Kagan 1964; Kohlberg 1966; Lewis and Weinraub 1979; Mischel 1966; O'Bryant and Corder-Bolz 1978). For people we encounter in social interactions (perhaps including vicarious interactions via television), we frequently ascribe qualities of masculinity and femininity (two orthogonal, bipolar dimensions).
The propensity to use gender role cues to categorize others varies across individuals. Yet, gender remains an important classificational dimension for many individuals (Bern 1985). Gender-related processing has been considered in a number of marketing studies with only limited success. For example, Gentry and Haley (1984) were unsuccessful in using gender schema processing to predict ad recall. Schmitt, LeClerc and Dube-Rioux (1988) found attitude toward the ad did not differ between gender-orientation subject groups. These results contrast with the 65 sychological literature which has frequently found differences due to the gender orientation of subjects (cf. , Bern and Lenney 1976; Frable and Bern 1985; Moore, Graziano and Millar 1987; Moore and Rosenthal 1980; Quackenbush 1987). The evidence of gender related processing and perception formation would seem to be very much a part of the issues related to viewer's responses to female roles in television advertising. That is, there should be some cognitive response that stereotypic portrayals elicit from viewers, either in the direction of acceptance or rejection.
From this perspective, the current study regresses somewhat fron1 past research to address more basic issues related to gender role judgments about stereo typic and non-stereotypic role portrayals. In general, this study considers the perceptual judgments about stereotypic and non-stereotypic roles and their relationships to the gender judgments of these ad characterizations. Bern Sex-Role Inventory Bern has proposed the Bern Sex-Role Inventory (BSRI) as a means for appraising an individual's gender orientation through Masculinity and Femininity subscales (Bern 1974; 1981).
The standard approach researchers have taken in using the BSRI is to classifY individuals into one of four gender orientation categories via sample-based median splits on Masculinity (M) and Femininity (F) dimensions. Individuals high on both M and F are called androgynous; high M and low F individuals are masculine; low M and high Fare feminine; and those low on both dimensions are undifferentiated. Males classified as masculine are called "sex-typed,"as are feminine females. Factor analyses of the BSRI indicate that a more internally consistent and parsimonious scale can be achieved with only one-half of the original items (Bern 1981).
The short form BSRI, which contains 10 masculine, feminine, and neutral items, was used in the current study. As mentioned previously, users of the BSRI have traditionally classified subjects into one of four gender orientation categories. However, examination of the methods used to create the two subscales and their empirical distributions suggest that these scales do not have natural categories, but instead approximate a multivariate normal distribution. A median split, a convenient and commonly used method for classifying sub ects, forces the separation of many similar observations near the median into distinct categories for which gender schema theory (Bern 1985) predicts different resul ts. For example, there is little difference between a M or F score of 49 and 51 (scores which are well within the measurement error of the BSRI); yet, the use of a cutpoint of 50 would indicate that the individuals who possess these scores would be markedly, and in our opinion artificially, different. Although intuitively appealing, the categorizing technique does not take advantage of the ordinal nature of the data and sacrifices statistical power.
This suggests the use of F and M as continuous variables; however, this approach comes at the expense of the traditional interpretation of the nomenclature (i. e. , feminine, n1asculine, androgynous, and undifferentiated). Consistent with this view, Cook (1985) points to other weaknesses of the median-split method. Cook's review of the BSRI literature indicates that the four median-split categories are often used without adequate theoretical justification and largely serve as convenient labels.
Cook suggests there is a need to address the effects of M and F individually. "[A]nyexperimental effects may be entirely attributable to one of the two dimensions, for exanlple nlasculinity. This overriding power would make levels of the other variable, and the classification by extension, largely superfluous" (Cook 1985, p. 104). Her recommendation is to give consideration to alternative uses of the scale (including difference scores, interactions, etc. ) to help explain research phenomenon.
A final rationale for the use of F and M as continuous variables is that Bern's Gender Schema Theory (1985) essentially predicts only the responses of sex-typed and androgynous individuals, leaving two other groups' behaviors unexplained. In total, these reasons point to the need to consider alternative methods of analysis of M and F. Current Research The current study uses the BSRI as both a projective instrument (used to rate ad character gender orientations) and a self-rating scale.
While the BSRI is designed to be a self-rating scale, it has been used as a projective scale in at least two other studies (Kolbe 1983; Peevers 1979). These measures are used as predictor 66 variables for perceptual judgments of the character (Pchar)' advertisement (Pad)' and advertised product (P prod). The research questions that arise from this exploratory investigation of gender schema and female roles in television advertising are as follows: 1. In general, do stereotypic ad characters obtain a less positive perceptual rating thap non-stereotypic characters? 2.
Do projective ratings of the BSRI differ for various character portrayals? 3. Are stereotypic fen1ale character roles rated as highly fenlinine? 4. Are non-stereotypic female character roles rated as more masculine than stereotypic female characters? 5. Does the gender of the observer influence the perceptions of characters or the BSRI projective rating of the ad characters? 6. Are the BSRI masculinity and femininity character ratings predictive of Pchar' Pad' and Pprod? 7. Does the absolute difference between BSRI projective character ratings and BSRI self-ratings predict P char' Pad' and Pprod?
METHOD Stimulus Ad Selection Off-air television advertisements were used to present stereotypic and non-stereotypic role portrayals. This differs from other studies which have typically used print advertisements. The ability to see the character, hear her speak, and observe behavior and mannerisms offers the viewer more input as to the personality of the individual appearing in the ad than could a print advertisement. This provides the respondent with more information upon which to make attitudinal and gender orientation judgments.
Television advertisements used in this study were selected by a pretest employing expert judges. The two judges, one male and one female, who are marketing professors trained in promotion and advertising, evaluated the role portrayals in 49 television advertisements. The judges evaluated the ads for the purposes of: (1) identifying ads with distinctive major characters (one which had a 10 second or longer appearance in the ad with one or more lines of dialogue); (2) rating the general femininity and stereotypic qualities of ajor female ad characters; and (3) rating the masculinity and femininity of the major character with key items selected from the Bern Sex-Role Inventory (those items with the highest eigenvalues in factor analyses of the scale (Bern 1981)). In total, these dimensions were used to identify ads which contained character roles that were distinctive, either stereotypic or non-stereotypic, and possessing personality characteristics typical of stereotypic and non-stereotypic individuals.
Distinctiveness of the character role was important in assuring that subjects would identify and attend to major characters while viewing the ads. Such identification and attention capabilities were necessary in order for subjects to adequately make judgments about the characters. Based on the judges' ratings, four advertIsements were selected for use in this research. In both types of ads (stereotypic and non-stereotypic), one ad contained a female character appearing alone, while the second ad had a female/male dyad, with the female character having the major role.
The male/female interaction represents a more dynamic character portrayal than a single female character speaking to the camera and as such poses a more distinct role portrayal for subjects to analyze. The stereotypic role presentations were contained in laundry detergent and dishwashing liquid ads. The detergent commercial featured a woman who washed the shirt of her truck driving husband. Both a male and female appeared in this ad. The dishwashing liquid ad had a single female character who spoke directly to the camera. The dishwashing liquid ad character was the only person in the ad.
Non-stereotypic ads were for a major dog food brand and decaffeinated coffee. The dog food ad featured a female kennel owner, the only character in the ad, who spoke to the camera. The coffee ad featured a female scuba diver who was served coffee by her husband on board a boat. Experimental Sessions The ads were shown to undergraduate students enrolled in the introduc tory marketing course at a northwestern university. A total of 426 subjects, 67 in groups of 75-100, participated in the study. As over 30% of all students at the university take this course, a wide range of majors are represented.
The use of student subjects was considered appropriate as they are a group which should be responsive to gender role portrayals. The students were told that the premise of this study was to determine how individuals view television commercials, particularly in regard to charact~r portrayals. The students were told they would be viewing a series of commercials and asked to make sonle candid judgments about the ads. After viewing each commercial, subjects selected the individual whom they perceived to be the major ad character. Subjects reported how often they had seen the ad and rated the major character on the BSRI.
Perceptual judgments about the character, ad, and product were obtained with 7-point semantic differential scales anchored by irritating/not irritating, unpleasant/pleasant, dull/dynamic, depressing/uplifting, offensive/not offensive, and not enjoyable/enjoyable. These items were borrowed and adapted from the A ad literature (cf. , Gardner, Mitchell and Russo 1985; MacKenzie, Lutz and Belch 1986). After viewing all four commercials, subjects then rated themselves on the BSRI and provided general demographic information (age, sex, marital status, citizenship).
The results reported here focus on the projective BSRI ratings, perceptual measures, and BSRI self-rating. RESULTS BSRI Short Fornl Usage The short form BSRI was used in this study. The 30 item short form has obvious advantages over the 60 item Original BSRI in terms of administrative time and parsimony. Bern (1979; 1981) proposed the shortened version as a means of addressing criticisms of the psychometric weaknesses of the original form. The resulting 30 item scale increases the internal consistency and orthogonality of the F and M scales (Bern 1981).
The social desirability of the BSRI adjectives in the two scales was also balanced, which was a weakness of the original BSRI. Thus, the short form represents a refinement of the inventory (Bern 1981). A confirmatory analysis of the psychometric properties of the short form BSRI supports the internal consistency of the F and M scales. Cronbach's Alphas for the self-rating use of the BSRI F and M scales were . 90 and . 84, respectively. Projective BSRI reliabilities ranged from . 91 to . 94 on the F scale and . 88 to . 91 on the M scale. These reliabilities were considered ,;::-. g indicators of the internal consistency of the _scales under both application situations. Perceptions of Ad Characters, Advertisements, and Products Stereotypic characters did not have consistently lower perceptual judgments than non-stereotypic characters in this study (see Table 1). In total, stereotypic ad characters were not viewed negatively as response averages were above the midpoint of the scale. Results of analyses of variances suggest that there were significant differences between the commercial perceptual judgments of P char (p=. 003), Pad (p