Gender Schema Theory As a Predictor of Ad Recall

James W. Gentry, Oklahoma State University
Debra A. Haley, Oklahoma State University
ABSTRACT - The basis for this research is drawn from Bem's (1981) Gender Schema Theory, which proposes that one's sexual self-concept affects how one structures items in memory. Further, this memory structure is thought to play an anticipatory role in the search for and assimilation of incoming information. Bem (1981) found that sex-typed individuals (masculine males or feminine females) were more likely than cross-typed individuals to recall words in clusters which were consistent with their sex-role orientation. The study found very little empirical support for Bem's (1981) version of the theory, but did find some support for another version presented by Markus et al. (1982).
[ to cite ]:
James W. Gentry and Debra A. Haley (1984) ,"Gender Schema Theory As a Predictor of Ad Recall", in NA - Advances in Consumer Research Volume 11, eds. Thomas C. Kinnear, Provo, UT : Association for Consumer Research, Pages: 259-264.

Advances in Consumer Research Volume 11, 1984      Pages 259-264

GENDER SCHEMA THEORY AS A PREDICTOR OF AD RECALL

James W. Gentry, Oklahoma State University

Debra A. Haley, Oklahoma State University

[The authors express their thanks to Nancy Allison, Marilyn Bamberger, Ray Fisk, Sandy Heh, Jo Anne Hopper, John Mowen, and Greg Pickett for their help in the study. Further, the authors appreciate the financial support provided by the Dean's Excellence Fund, College of Business Administration, Oklahoma State University.]

ABSTRACT -

The basis for this research is drawn from Bem's (1981) Gender Schema Theory, which proposes that one's sexual self-concept affects how one structures items in memory. Further, this memory structure is thought to play an anticipatory role in the search for and assimilation of incoming information. Bem (1981) found that sex-typed individuals (masculine males or feminine females) were more likely than cross-typed individuals to recall words in clusters which were consistent with their sex-role orientation. The study found very little empirical support for Bem's (1981) version of the theory, but did find some support for another version presented by Markus et al. (1982).

INTRODUCTION

For advertising to be effective, the receiver must (1) notice it, (2) process its content, and (3) favorably relate that content to his/her cognitive structure. Much effort is aimed at increasing the awareness of the advertisement, through extensive copy testing and through careful placement of the ad. Likewise, numerous studies have been made as to how people process information (such as the brand versus attribute processing literature). Far less is known about the manner by which individuals store advertising content in long term memory. One possible explanation (Krugman 1965) is that learning takes place through classical conditioning, a passive process. A more active view of the process is that the information is processed cognitively and is stored in long term memory as schemata (Bozinoff 1981). Schemata are the large sets of well structured cognition: that have been learned over time as experience accumulates. A schema functions as an anticipatory structure in that it determines the readiness to search for and to assimilate incoming information. This study will investigate how gender-based schemata relate to the recall of advertising.

Bem (1981) has proposed Gender Schema Theory as an explanation of how one's gender (masculine or feminine) affects the individual's cognitive structure (a network of associations that organizes and guides the individual's perception). Each individual may have many schemata, and the readiness with which an individual involves one schema rather than another is referred to as the cognitive availability of the schema (Nisbett and Ross 1980; Tversky and Kahneman 1973). Given the importance of one's sexual self-concept as children develop into adults, it is likely that the gender schema will be relatively available. This study involves an extrapolation of Bem's (1981) research into the applied field of advertising research. The stimuli in this research were mock print ads, and the subjects were asked to provide unaided recall measures. Gender schema theory should provide insight into how people process "gendered" ads and into why they respond to them differently, as found by Alreck, Settle, and Belch (1982).

One purpose of this study is fairly basic in nature; to provide further investigation of Gender Schema Theory's ability to explain information processing differences among individuals. The other purpose of the study is to provide insight into how individuals process the content of advertisements. Prior research has related personality (in this case, sex-role orientation) to buyer behavior (e.g., Alreck, Settle, and Belch 1982, Gentry and Doering 1977; and Vitz and Johnston 1965). In general, sex has predicted better than sex-role orientation (Alreck, Settle, and Belch 1982; Gentry and Doering 1977). The attempt to relate personality to behavior and behavioral intention variables appears to have had little theoretical support, as most models of consumer behavior hypothesize personality to affect information processing variables rather than output variables such as attitude, intention, or purchase. howard and Sheth (1969), for example, hypothesize that personality affects perceptual constructs such as stimulus ambiguity, overt search and attention. This study relates one's gender to the processing of information.

LITERATURE REVIEW

This study builds upon research in social psychology on sex-role orientation. Bem (1974) developed the concept of psychological androgyny, which challenged the traditional view that masculinity and femininity were on opposite ends of a continuum. Instead, Bem (1974) suggested that masculinity and femininity were two separate dimensions and that an individual could be high on both dimensions (or "androgynous"). Spence, Helmreich, and Stapp (1975) challenged the measurement approach advocated by Bem (1974) and presented an alternative, the Personal Attributes Questionnaire (PAQ). The concept of psychological androgyny proved to be an interesting one to many researchers, and a large literature developed (see Taylor and Hall (1982) for a good summary). In marketing, several studies have attempted to relate psychological androgyny to consumer behavior (Allison, Golden, Mullet, and Coogan 1980; Burns 1977; Gentry and Doering 1977, 1979; Gentry, Doering, and O'Brien 1977; Golden, Allison, and Clee 1979; Tucker 1976). As discussed earlier, most of the research investigated relationships between sex-role orientation and variables such as attitude or behavior.

Later research by Bem (1981) combined the research tone on psychological androgyny with that tone in the area of information processing. Heretofore, most of the literature in the area of consumer information processing has concentrated on how people process information (brand vs. attribute processing) and has not emphasized the use of personality variables to explain why people process information differently. Bem's (1981) Gender Schema Theory may provide a partial explanation.

Bem's (1981) studies have supported the expected role of the gender schema. She used the Bem Sex Role Inventory (BSRI: Bem 1974) to classify respondents as masculine, feminine, androgynous, or undifferentiated. Males classified as being masculine and females classified as feminine were grouped as being sex-typed. Males categorized as feminine and females categorized as masculine were classified as cross sex-typed. Individuals who scored high on both masculine and feminine measures were classified as androgynous. Those individuals scoring low on both masculinity and femininity were referred to as undifferentiated. Bem's research hypothesis dealt with sex-typed individuals, proposing that they do organize information in terms of the gender schema. This hypothesis was operationalized in the first study as sex-typed individuals would, after being presented with a sequence of 61 words, show more clustering of gender-relevant items in free recall than would non-sex typed individuals. In a second study, the hypothesis was operationalized as sex-typed individuals will respond more quickly when asked to make a dichotomous me/not me judgment about each of the 60 attributes of the BSRI. The results supported both of these hypotheses.

Markus, Crane, Bernstein, and Siladi (1982) supported Gender Schema Theory to some extent, as they found that masculine schematics recall masculine terms more readily while feminine schematics recall feminine items more readily. Contrary to Bem's (1981) hypothesis Markus et al. (1982) also found that androgynous individuals rec=!let masculine and feminine items readily. Whereas Bem (1981) reported results dealing only with the cluster of recalled items, Markus et al. found differences in recall itself.

A second body of literature relevant to this study is that on the frequency of recalling incongruent information. In a study investigating the role of gender in product choice, it is very possible to cross the gender of the product by the sex of the person presenting the product. For example, a male model may promote a product which is viewed as being feminine. Recall of such an advertisement may be due more to its incongruity rather than to its being consistent with the individual's gender schema. Some researchers (Bower, Black, and Turner 1979; Greenwald and Sakumura 1967; Hastie and Kumar 1979) have found that information which is novel or distinctive is likely to be recalled in free recall faster. Crocker (1981) suggests that the greater recall is due to the individual's need to process the information at a deeper level in order to "explain" the incongruence. The bulk of this research has involved abstract tasks such as the recall of word pairs.

This study will involve an extrapolation of earlier research to a more applied setting, the viewing of mock print ads. As discussed in the methodology section, the applied nature of the study presents problems in operationalizing some of the concepts. However, the more applied nature of the task should make the results of more value to advertising practitioners.

RESEARCH METHODOLOGY

When translated into an advertising context, Bem's (1981) results would lead us to infer that subjects will be able to recall better those advertisements which best fit their gender schema. A possible conflict may occur, though, due to the dynamics involved in gendered advertising. While sex-typed (masculine males and feminine females) individuals may be expected to recall advertisements consistent with their sex-type better than advertisements consistent with the opposite sex, it is not clear what the effect of incongruent advertisements would be. One example of an incongruent advertisement would involve the portrayal of a female model with a traditionally masculine product or of a male actor with a traditionally feminine product.

Two hypotheses stated in alternative form will be investigated in this studs:

H1: Sex-typed consumers will recall more frequently those advertisements consistent with their traditional sex-role than will cross-typed individuals.

H2: Incongruent ads will be recalled more frequently by sex-typed individuals than by cross-typed individuals.

Thus sex-typed individuals are hypothesized to recall consistent and incongruent advertisements more easily. Ads consistent with sex-role orientation will be recalled more easily because they are processed efficiently (and thus stored in long term memory) through the use of their gender schema. Incongruent ads are expected to be recalled more easily because sex-typed individuals will find the level of incongruence greater and will have to process the information at a deeper cognitive level.

Procedures Overview

Student subjects were categorized as to their gender and then asked to evaluate a series of print ads. After seeing all of the ads, they asked to provide unaided recall measures of the ads.

Development of the Stimuli

The stimuli in the study were ads portraying different products and members of different sexes. Masculine, feminine, and sexually neutral products were chosen, drawing upon the prior research by Gentry and Doering (1977) for help in the categorization of products. They had students subjects rate a wide variety of products and leisure activities as to their masculinity/femininity and the majority of the products included in our pretest were ones included in that study. In the pretest, an undergraduate consumer graduate behavior class (n = 40) was asked to rate a list of 30 products and activities on a seven-point masculine-feminine scale. The eight most feminine, eight most masculine, and four most sexually neutral stimuli were selected for the next round of pretesting. Pictures were taken of the twenty products, and a graduate consumer behavior class (a = 15) was asked to rate those stimuli on the same masculine/feminine scale. Finally the 15 stimuli used in the study were selected on the basis of their ratings; these stimuli are listed below.

Differential experience with the products chosen is consistent with the notions underlying Gender Schema Theory. However, differential experience with brands within a product class is not. In order to control for brand familiarity (and its expected effect on ad recall), artificial brand names were used. Four-letter nonsense words were generated randomly, modified so that they resembled possible brand names, and then evaluated by an undergraduate consumer behavior class (n = 40) on good/bad and masculine/feminine scales. The brand names selected are given below (with the associated stimulus):

Masculine Stimuli             Feminine Stimuli         Neutral Stimuli

Zorb Football                  Cafi Blusher              Denk Deodorant

Mikz Aftershave              Bafs Hairspray          Belb Swimming Pool

Helc Pick-up Truck        Casp Ballet Studio     Bapt Bicycles (his and hers)

Pobb X-Rated Movie     Surq Yarn

Fews Shotgun                  Midd Ice Skates

Gerr Motor Oil                 Porl Egg Beater

The mixing of a male (female) model with a feminine (masculine) stimulus constitutes the incongruent ads necessary in order to test the second hypothesis. Other forms of incongruity (such as color vs. black and white ads, unusual product usage, or more vivid product descriptions in some ads) could be implemented, but the mixing of the sex of the model with the gender of the product constitutes the most relevant form of incongruity in the area of sex-role orientation. A male and a female were selected to be shown with the gendered stimuli. Both are young, attractive doctoral students. In order to control for variations in the masculinity or the femininity of the ads across stimuli, it was decided to seep the human components of the ads constant. The male model was pictured with three masculine stimuli (football, pick-up truck and aftershave) and three feminine stimuli (ballet studio, yarn, and hairspray). The female model was pictured with the other six gendered stimuli.

Captions of about 25 words (range: 20-30) were written to go with the pictures. The content of the message included one mention of the brand name, the mention of the price, and encouragement to try the product or activity. A pretest (n = 15) of the captioned pictures was made before the final 15 ads were selected.

Sample

The study was conducted using students (a = 168) in undergraduate classes. Given the exploratory nature of this study, it was believed that student subjects would be able to provide a sufficient first test of the hypotheses. The intent of this study is to test theory application, not effects application (Calder Phillips, and Tybout 1981).

A second reason for the use of student subjects relates to a measurement problem. Sex-typed individuals are relatively more common than cross sex-typed individuals regardless of the age of the subjects, but cross sex-typed individuals are especially rare among those in their forties or older and in those without at least some college education (Alreck, Settle, and Belch 1982). Consequently, the selection of college students as subjects should result in having a larger proportion of cross sex-typed individuals in the sample.

Data Collection Procedures

Undergraduate consumer behavior classes were used in the data collection. Two classes (n = 85) saw nine ads (no incongruent ads included) and two other classes (a - 83) saw 15 ads (including male model/feminine stimulus and female model/masculine stimulus ads). The order in which the ads were viewed was randomized before each showing of the slides to guard against possible recency or primacy effects. [These are not expected as the rank order correlation between the order of seeing the ads in the pretest and the order of recalling them was nonsignificant (r = -.01).]

Before being shown the slides, the students were asked to complete the Personal Attributes Questionnaire (PAQ: Spence, Helmreich, and Stapp 1974), which has been used more frequently in consumer behavior studies (Allison, Golden, Mullet, and Coogan 1980; Burns 1977; Gentry and Doering 1977; Tucker 1976) than has the BSRI. Respondent scores on the PAQ were used to identify sex-typed and cross sex-typed individuals. While this procedure is similar to that used by Bem (1981), it should be noted that such a procedure has several critics. For example, Sirgy (1982) states that standardized, "clinical" personality measures should not be used as indicators of sex-role self-concepts. Also, Spence and Helmreich (1981) and Pyke and Graham (1983) discuss conceptual problems with the use of the BSRI to categorize subject as to their sex-role orientation Bem (1981b) argued that her detractors and misinterpreted how the BSRI was being used.

There is confusion in the literature as to the categories used to classify respondents. Bem (1981) used four categories: undifferentiated (those who scored low on both the masculine and feminine dimensions), sex-typed (masculine males and feminine females), cross sex-typed (feminine males and masculine females), and androgynous (those who scored high on both dimensions). On the other hand, Markus et al. (1982) used four different categories: low androgynous (defined in the same way as undifferentiated above), masculine schematics (those who scored significantly higher on the masculine dimension, regardless of sex), feminine schematics (those who scored significantly higher on the feminine dimension, regardless of sex), and androgynous. Also, different statistical methods were used to classify the respondent's gender. Bem (1981) divided her subjects at the median for each dimension; for example, those higher than the median on both dimensions -were classified as being androgynous. [This procedure is identical to the one recommended by Spence, Helmreich and Stapp (1974) for use with their PAQ. This procedure is the one which we used in this study.] Then the individual's sex was taken into consideration in order to determine if, for example, high masculine-low feminine respondents were sex-typed (males) or cross sex-typed (females). On the other hand, Markus, et al. (1982) did not break their data at the median of each dimension, but rather classified respondents as masculine schematics if their responses to masculine items were significantly greater statistically (p < .05) than their responses to the feminine items. Sex was not included explicitly, although the vast majority of masculine schematics were male (42 of 49) and the vast majority of the feminine schematics female (25 of 30).

Given the differences found between masculine and feminine schematics (Markus et al. 1982; Crane and Markus 1982), the combination of masculine males and feminine females into the "sex-typed-' category (Bem 1981) seers to be questionable. Consequently, we will investigate recall by people in six different categories: undifferentiated (or low androgynous), masculine males, feminine females, feminine males, masculine females, and androgynous. Thus, our categorization encompasses both that of Bem (1981) and that of Markus et al. (1982)

The subjects were asked to evaluate the ads on the following dimensions as they viewed chem: good/Dad, appealing/ unappealing, feminine/masculine, progressive/regressive, unattractive/ attractive, and boring/interesting. The main purpose of these evaluations was to lend credence to the cover story that this session was for on evaluating mock ads. Also, the masculine/feminine ratings served as one manipulation check for the stimulus categorizations.

The subjects were asked next to recall the products which they had been shown, in an unaided manner ("Please list all of the products which you have been shown.") Then they were asked to rate the stimulus (as opposed to the advertisement for the product) on a seven-point masculine/feminine scale as a second manipulation check of the stimulus categorizations. The students in the classes which were not shown the incongruent ads were asked to rate the six products involved in those novel ads on a seven-point masculine-feminine scale. This measurement provides a means of investigating the effect of the inconsistently gendered model on the product's image when these ratings are compared to those by the students seeing the novel ads.

RESULTS

Cronbach alphas were computed for the responses to the PAQ. Both the masculine scale (v = .72) and the feminine scale (aL = .75) exhibited a sufficient level of reliability. Using the procedure recommended by Spence, Helmreich and Stapp (1974), the medians on the masculine scale (median = 53) and the feminine scale (median = 32) of the PAQ were the breakpoints in the categorization of the subjects' genders. There were 31 androgynous subjects (21 males and 10 females), 38 undifferentiated subjects (10 males and 28 females), 51 high masculine/low feminine subjects (28 males and 23 females), and 48 high feminine/low masculine subjects (27 males and 21 females).

The ratings of the masculinity/femininity of the advertisements and of the products themselves, as shown in Table 1, support the a priori categorization of the products. For those products in which the sex of the model was consistent with the a priori categorization of its render, the ratings of both the advertisements and of the products are strongly masculine for the "masculine-- products> strongly feminine for the "feminine-' products, and very close to the midpoint of the scale for the "neutral" products. Also, there is no differentiation between the masculinity/femininity racing of the advertisements and of the products for these products.

However, when the model shown in the ad is not with the a priori categorization of the product, the advertisements were not rated as being as masculine (or feminine) as were the advertisements for which the sex of the model was consistent with the product's gender. The ratings of the products themselves by those subjects who saw the advertisements were more masculine (or feminine), though not as extreme as the rating of the products' masculinity/femininity by those who did not see the product. Thus, the ratings of the products by those who did not see the advertisements provide strong support for our a priori categorization. The ratings of the advertisements indicate that one can change the perceived gender of the product by using a model whose sex is inconsistent with the traditional gender. The masculinity/femininity of the advertisements was significantly (p < .05) more neutral than the product perceptions (from those who did not see the ads) for all six products. Further, those seeing the ads rated the products themselves more neutral in the case of all six products, although only three (shotgun, motor oil, and hairspray) were rated as being significantly (p < .05) more neutral.

TABLE 1

MASCULINITY/FEMINITY RATINGS OF THE ADVERTISEMENTS AND OF THE PRODUCTS

As was the case in the pretest, the correlation between the order that the ad was seen and the order in which it -was recalled (if at all) was .005, indicating that there were no primacy or recency effects.

The percentage of each of the six groups (undifferentiated, masculine males, feminine females, feminine males, masculine females, and androgynous) that recalled each advertisements is shown in Table 2. When differences among all the groups were investigated using chi square analysis, only two products were found to be recalled differentially: blusher, x2 (5 d.f.) = 19.3, p < .01; and deodorant, x2 (5 d.f.) = 12.1, p < .05. In the case of blusher, the explanation was that females (feminine females, masculine females, and undifferentiated--the majority of those were female) recalled the ad more frequently than males. In the case of deodorant, those who were classified as nongendered (undifferentiated or androgynous) were more likely to recall the ad. The first finding does not support Gender Schema Theory, since it would suggest that sex and not gender is related to recall. The second finding may be more consistent with Gender Schema Theory, as non-gendered individuals were more likely to recall a neutral product. However, a high percentage of masculine males were also likely to recall the deodorant ad, casting further doubt upon the ability of Gender Schema Theory to predict recall.

TABLE 2

PERCENTAGE RECALLING EACH ADVERTISEMENT

As ad recall is a questionable operationalization of information processing, two other dependent variables were investigated. The first was the ease of recall; i.e., whether the advertisement was recalled early. This was operationalized as the advertisement being recalled among the first three ads (or among the first five ads for the incongruent ads since they were always part of a 15-ad presentation instead of a nine-ad presentation). The results dealing with the ease of recall are shown in Table 3. A chi-square analysis was made for each product; the only product for which there were significant differences across gender in the ease of recall was the football advertisement [X2 (5 d.f.) - 10.8, p < .1]. Masculine respondents, regardless of sex, were more likely to recall early the football advertisement.

The other alternative operationalization of information processing dealt with the sequence of the recalled advertisements. If gender schema affect the manner in which information is processed and then stored in memory, products of the same gender should be recalled together. The results dealing with the order of recall are shown in Table 4. Analyses of variance for the mean percentage of sequential pairs recalled found no significant differences in how the different gendered individuals recalled any of the types of advertisements. However, one consistent trend (though not statistically significant) does appear in the results shown in Table 4: masculine females are more likely to recall in sequence advertisements of masculine products than are feminine males (masculine product, male model: 35% vs. 17%; masculine product, female model: 28% vs. 18%; masculine product, male or female model: 23% vs. 6%). In fact, masculine females were more likely to recall masculine product advertisements in sequence than any other group. They also recalled female products with male models in sequence more frequently than feminine males (26% vs. 15%), but not neutral products (27% vs. 28%) or feminine products with female models (24% vs. 22%). Though not as marked, the same patterns are noticeable between masculine females and the other gender classifications as well.

TABLE 3

EASE OF RECALL MEASURE: PERCENTAGE RECALLING EACH ADVERTISEMENT AMONG THE FIRST THREE RECALLED

Hypothesis 1 posits Bem's (1981) operationalization of Gender Schema Theory, that sex-typed individuals would be more likely to recall gendered schema advertisements than would cross sex-typed individuals. To test this hypothesis, masculine males and feminine females were grouped together, as were feminine males and masculine females. The analyses found no significant differences for any of the products for any of the dependent variables, thus providing little support for Bem's (1981) operationalization.

Hypothesis 2 suggests that gendered individuals will recall the incongruent advertisements more frequently because these will process the information at a deeper level. None of the six products in the incongruent advertisements were recalled significantly more frequently, earlier, or in sequence by sex-typed rather than cross sex-typed individuals.

Markus et al. (1982) suggested that differences between masculine and feminine schematics should be investigated, rather than between sex-typed and cross-typed individuals. For example, for all three masculine, male-model products, recall was greater for masculine males than for feminine males and greater for masculine females than for feminine females. In order to investigate this viewpoint, we grouped masculine males with masculine females and compares their recall to that of the combination of feminine females and feminine males. Chi-square analyses were performed for each product, and the only (marginally) significant differences were found for deodorant (X2 < 3.5, p < .1) and footballs (X2 = 2.3, p < .15). In both cases, masculine schematics were more likely to recall the ads than were feminine schematics. When ease of recall is investigated rather than recall frequency, significant differences were found for two products: footballs, X2 (1 d.f.) = 9.9, p < .01; and egg beaters, x2 (1 d.f.) = 3.9, 2 < .1. Masculine respondents were more likely to recall early the football advertisement while feminine respondents were more likely to recall early the egg beater ad. No significant differences were found between masculine and feminine respondents in terms of more frequent sequencing of gendered products. However, the patterns noted above (that masculine females recall masculine products in sequence more than feminine males) is consistent with the Markus et al. (1982) version of Gender Schema Theory. On the other hand, Markus et al. would predict also that masculine males would cluster male stimuli more than feminine females; no such trends are noted. In summary, while the evidence does not constitute strong support for the Markus et al. (1982) view of Gender Schema Theory, clearly there is more support for their version than for Bem's version. All of the significant findings, except that for recall frequency for deodorant, are consistent with the Markus et al. version.

TABLE 4

MEAN PERCENTAGE OF SEQUENTIALLY RECALLED PAIRS, CLUSTERED ON BASIS OF GENDER

The results with the blusher product in Table 2 (that females recalled the ad more frequently than did males) suggest that sex may be a better predictor than gender. Sex is certainly a much simpler variable to measure. When the recall of the ads was related to the individuals' sexes, greater recall of blusher was the only strongly significant result. However, one marginal relationship (X2 (1 d.f.) = 2.2, p < .15) was found for yarn, which was recalled more frequently by females than males. Moreover, no significant differences were found by sex for the other two dependent variables: ease of recall and sequencing of recall.

From a marketing point of view, segmentation by sex is quite customary. Gender Schema Theory might suggest that we should segment within each sex. Consequently, we compared the recall of masculine males with that of feminine males and the recall of feminine females with that of masculine females. No differences in recall frequency were found between the two groups of females, and only one difference was found between the two male groups. Masculine males were more likely (X2 = 3.5, p < .1) to recall the deodorant advertisement than were feminine vales. This result does not conform to Gender Schema Theory's predictions since the product involved is a neutral one. However, when ease of recall (Table 3) is investigated, we note several significant relationships: males recalling football ads early, x2 (1 d.f.) = 3.2, p < .1; females recalling football ads early, x2 (1 t.f.) = 5.9, p < .025; males recalling blusher ads early, x2 (1 t.f.) = 3.1, 2 < .1; and females recalling egg beater ads early, * (1 d.f.) = 5.4, 2 < .025. Except for the blusher ads, the individuals whose gender was consistent with that of the product were more likely to recall the ads early. For the blusher ad, masculine males were more likely to recall the at early than were feminine males. In terms of the sequencing of ads (Table 4), masculine females consistently sequence masculine products more than feminine females (though, as discussed earlier, the differences are not statistically significant). In summary, these results do provide some justification for investigating further the effects of gender within a sex.

DISCUSSION

Most of the previous work in marketing that has investigated the concept of one's masculinity/femininity has related gender to attitudes or to purchase behavior. However, the theoretical models of consumer behavior, such as the Howard and Sheth (1969) motel, have hypothesized that personality affects information processing variables directly and affects output variables such as attitude and purchase only indirectly. Gender Schema Theory provides a theoretical base for predicting how one's gender will affect recall of advertisement.

The results of this study do not support Bem's (1981) hypothesis that sex-typed individuals will recall the information better than cross sex-typed individuals. Bem's (1981) classification scheme of grouping masculine males with feminine females and grouping masculine females with feminine males thus does not seem to have much explanatory value in terms of how people process advertisements. On the other hand, the results of the study provide some support for Markus et al.'s (1982) alternative interpretation that gender schema operate between masculine schematics and feminine schematics. Further, the results dealing with the ease of recall suggest that it might be beneficial to extent the classification scheme so that we can investigate differences in processing by different gendered individuals of the same sex.

Stuteville (1971) hypothesized that male acceptance of female-oriented products will occur much slower than female acceptance of vale-oriented products. One would expect masculine females to adopt masculine products before feminine females and expect feminine males to adopt feminine products before masculine males. Further, according to the hierarchy of effects model, individuals need to be cognitively aware of the product before accepting it. Thus the results of this study are supportive of Stuteville's contention, as (1) masculine females seem to sequence advertisements for masculine products more than did feminine females, (2) masculine females recalled the football advertisement more easily than did feminine females, but (3) masculine males recalled the blusher advertisement more easily than did feminine males. The results provide justification for further investigation of the role one's gender plays in one's processing of the information contained in advertisements.

SELECTED REFERENCES

A complete listing may be obtained from the authors.

Alreck, Pamela L., Settle, Robert B. and Belch, Michael A. (1982), "Who Responds to 'Gendered' Ads, and How?" Journal of Advertising Research, 22 (No. 2, April-May), 25-32.

Bem, Sandra Lipsitz (1974), "The Measurement of Psychological Androgyny, Journal of Consulting and Clinical Psychology, 42, 155-162.

Bem, Sandra Lipsitz (1981a), "Gender Schema Theory: A Cognitive Account of Sex-Typing, Psychological Review, 88 (No. 45. 354-364.

Bem, Sandra Lipsitz (1981b), "The BSRI and Gender Schema Theory: A Reply to Spence and Helmreich," Psychological Review, 88 (No. 4), 369-371.

Gentry, James W. and Doering, Mildred (1977), "Masculinity-Femininity Related to Consumer Choice, in Contemporary Marketing Thought, eds. Barnett A. Greenberg and Danny N. Bellinger, Chicago, IL: American Marketing Association, 423-427.

Markus, Bazel, Crane, Narie, Bernstein, Stan and Silati, Michael (1982), "Self-Schemas and Gender," Journal of Personality and Social Psychology, 42 (No. 1), 38-50.

Pyke, Sandra U. and Graham, J. Martin (1983), "Gender Schema Theory and Androgyny: A Critique and Elaboration," International Journal of Women's Studies, 6 (No. 1, January-February), 3-17.

Spence, Janet T., Helmreich, Robert and Stapp, Joy (1974), "The Personal Attributes Questionnaire: A Measure of Sex Role Stereotypes and Masculinity-Femininity," Journal Supplement Abstract Service Catalog of Selected Documents in Psychology (Ms. No. 617), 4, 43.

Stuteville, John R. (1971), Sexually Polarized Products and Advertising Strategy," Journal of Retailing, 45 (No. 2), 3-13.

Taylor, Marylee C. and Hall, Judith A. (1982), "Psychological Androgyny: Theories, Methods, and Conclusions," Psychological Bulletin, 92 (No. 2), 347-366.

----------------------------------------