Addictive Behaviors 29 (2004) 1735 1744

Adolescents responses to the gender valence of cigarette advertising imagery: The role of affect and the self-concept
William G. Shadela,*, Raymond Niaurab, David B. Abramsb
Department of Psychology, University of Pittsburgh, 130 North Bellefield Avenue, 5th floor, Pittsburgh, PA 15260, USA b Centers for Behavioral and Preventive Medicine, Brown Medical School and The Miriam Hospital, USA
a

Abstract The studies presented in this manuscript evaluated the role that affect and the self-concept play in adolescent never smokers reactions to the gender valence of cigarette advertising imagery. Study 1 (n=29; 59% female) revealed that adolescent females have more positive affective reactions to femalevalenced cigarette advertising imagery compared to male-valenced cigarette advertising imagery. Study 2 (n=101; 56% female) revealed that adolescent females viewed female-valenced cigarette advertising imagery as more relevant to their self-concepts compared to male-valenced cigarette advertising imagery. Across both studies, male adolescents did not respond differently as a function of the gender valence of cigarette advertising imagery. Thus, female-valenced cigarette advertising imagery may have specific effects on never smoking female adolescents by enhancing positive affect and suggesting that women who smoke hold the same characteristics as do the young women themselves. D 2004 Elsevier Ltd. All rights reserved.
Keywords: Adolescence; Smoking; Gender differences; Advertising; Self-concept; Affect

1. Introduction Smoking prevalence rates increased throughout the 1990s in both male and female adolescents; male adolescents smoking prevalence rates peaked at 37.7% in 1997 and female adolescents smoking prevalence rates peaked at 34.9% in 1999 (CDC, 2002a). Smoking prevalence rates among both male and female adolescents decreased from 1999 to 2002 (CDC, 2002a, 2002b; Johnston, OMalley, & Bachman, 2003), though a relatively large
* Corresponding author. Tel.: +1-412-383-2059; fax: +1-412-383-2041. E-mail address: wgs1@pitt.edu (W.G. Shadel). 0306-4603/$ see front matter D 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.addbeh.2004.03.042

1736

W.G. Shadel et al. / Addictive Behaviors 29 (2004) 17351744

percentage of adolescents of both genders can still be classified as current smokers (males at 29.2% and females at 27.7%; CDC, 2002a, 2002b). Additional gender differences have been observed with other smoking behaviors, for example, male adolescents are more likely to report having ever tried smoking compared to female adolescents (66.3% vs. 61.6%) and to report regular heavy smoking (>10 cigarettes/day on each day smoked for 30 previous days; 5.2% vs. 3.1%; CDC, 2002b). A greater proportion of male adolescents seem to experiment with smoking at earlier ages and initiate regular smoking at earlier ages compared to female adolescents (USDHHS, 2001). Thus, there is a clear need to uncover factors that contribute to differences in initiation between males and female adolescents, and also, which contribute to differences in their ongoing smoking behavior. Myriad factors are probably responsible for smoking initiation among adolescents, but a chief contributing factor may be exposure to cigarette advertising (USDHHS, 2001; Wakefield, Flay, Nichter, & Giovino, 2003). However, very little information is available about precisely how cigarette advertising might influence smoking among adolescents specifically (USDHHS, 2001; see also Shadel, Niaura, & Abrams, 2001). Studies that have examined cigarette advertising with reference to adolescent smoking have focused on the images displayed by the advertisements (Covell, 1992; Covell, Dion, & Dion, 1994; Krupka & Vener, 1992; Krupka, Vener, & Richmond, 1990; Romer & Jamieson, 2001; Shadel, Niaura, & Abrams, 2002, Shadel, Niaura, & Abrams, in press; Slovic, 2001). Cigarette advertisements that use extensive amounts of text are relatively rare; most advertisements present images with a single theme and minimal text describing cigarettes or the effects of smoking (Warner, 1985). Cigarette advertisements appear to be tailored differentially to males and females (USDHHS, 2001). Cigarette advertisements designed to appeal to females use images that emphasize issues that may be of special concern to women (USDHHS, 2001), for example, that smokers are lean, attractive, independent, and selfreliant (Krupka & Vener, 1992; Krupka et al., 1990). In contrast, cigarette advertising that is tailored to males may emphasize so-called masculine images (Krupka & Vener, 1992; Krupka et al., 1990; USDHHS, 2001). However, the reasons that such images may differentially influence smoking behavior adolescent males and females are not well understood (USDHHS, 2001). Several theory-driven mechanisms may account for the differential effects that gendervalenced imagery in cigarette advertisements may have on male and female adolescents smoking behaviors. First, affective responses to advertising are an important component of their persuasive efficacy in general (Geuens & DePelsmacker, 1998), and the positive affective reactions that cigarette advertising produces may in part drive adolescents decision to smoke (Romer & Jamieson, 2001; Slovic, 2001). As such, female-valenced advertising imagery (i.e., images that emphasize themes of relevance to women) would be expected to produce more positive affective reactions among female adolescents compared to advertising imagery that is male valenced (i.e., images that emphasize themes of relevance to men). Male adolescents would be expected to show more positive affective reactions to male-valenced imagery compared to female-valenced imagery. Second, the self-concept undergoes a significant degree of change during adolescence (Harter, 1999) and adolescents may look to outside sources, such as to cigarette advertisements, as they struggle to define and redefine themselves (Shadel et al.,

W.G. Shadel et al. / Addictive Behaviors 29 (2004) 17351744

1737

2001). Research has supported the notion that adolescent self-conflict (i.e., the degree to which an adolescent struggles with self-definition) moderates their reactions to cigarette advertising imagery (Shadel et al., in press), and facets of the adolescent self-concept have been shown to predict smoking initiation (Gibbons, Gerrard, Lando, & McGovern, 1991). Thus, because gender-valenced cigarette advertisements project images that may be differentially important to males and females, the more relevant cigarette advertising images are to female adolescents own self-concepts, the more likely they may be to be influenced by the advertisement. Similarly, the more relevant cigarette advertising images are to male adolescents own selfconcepts, the more likely they may be to be influenced by those advertisements. As such, female adolescents would be expected to perceive female-valenced advertising imagery as more relevant to their self-concepts compared to advertising imagery that is male valenced. Male adolescents would be expected to perceive male-valenced imagery as more relevant to their self-concept compared to female-valenced cigarette advertising imagery. The images from female-valenced cigarette advertisements and from male-valenced cigarette advertisements were presented to male and female never smoking adolescents in two studies. Study 1 (see Shadel et al., 2002) tested whether the gender valence of cigarette advertising imagery is associated with gender differences in never smoking adolescents affective responses to those images. Study 2 (see Shadel et al., in press) examined whether the gender valence of cigarette advertising imagery is associated with gender differences in never smoking adolescents perceptions of the relevance of the images to their self-concept. Support for our hypotheses would provide initial evidence suggesting psychological mechanisms through which the gender valence of cigarette advertising imagery differentially influences male and female never smoking adolescents. 1.1. Study 1 1.1.1. Participants Never smoking adolescents (never smoked or used tobacco products)1 were recruited using a professionally designed campaign. Advertisements asking individuals to participate in a

A never smoking sample of adolescents is critical to employ for two reasons. First, models of adolescent smoking and substance use (e.g., Flay & Petraitis, 1994; Shadel et al., 2001) suggest that exposure to advertising occurs in the scheme of more distal or background factors related to adolescent smoking and contributes unique variance to initial experimentation or initial puffing behavior. Second, because brief initial exposures to smoking may have an impact on sensitivity of nicotine receptors in the central nervous system (Pidoplichko, DeBiasi, Williams, & Dani, 1997) and lead to the development of symptoms of dependence (DiFranza et al., 2000), smoking even at low levels may cloud the picture of how cigarette advertising affects adolescents. As such, the effects of cigarette advertising among never smokers are important to investigate because said advertising may have its most potent impact before a more regular, dependent-like pattern of smoking behavior is in place (and smoking or smoking behaviors become increasingly governed by dependence processes). In other words, a never smoking sample of adolescents is critical to employ so as to understand the relatively pure effect of cigarette advertising on adolescents. Parental smoking was also an exclusion due to the findings that parental smoking is consistently a significant risk factor for youth smoking, and parental smoking could potentially differentially affect smoking behavior among adolescent males and females (UDSHHS, 2001).

1738

W.G. Shadel et al. / Addictive Behaviors 29 (2004) 17351744

study of How students ages 1117 think about different kinds of advertising were printed in monthly parenting magazines, displayed as fliers by local youth organizations and high schools, and shown as slides prior to films being shown in a multiplex movie theater. Inclusion criteria were as follows: age between 11 and 17 years; and provision (by at least one parent) of written informed consent and (by the adolescent) written informed assent. Exclusion criteria were as follows: ever smoking by the adolescent or parent; and presence of any medical or psychological condition that would make participating in the particulars of the study session difficult or inadvisable. Following completion of all study procedures (described below), each participant was provided with a US$50.00 gift certificate to a local music and video store and print materials (tailored to youth) regarding the deleterious effects of smoking on health (National Institutes on Drug Abuse, 2000). 1.1.2. Procedure Details of the study procedures are described by Shadel et al. (2002). After completing a baseline demographic and mood assessment (see below), participants evaluated the images taken directly from 24 cigarette advertisements, 24 antismoking advertisements, and 24 neutral advertisements (i.e., distracter stimuli). All advertisements were derived from print media sources (e.g., magazines, billboards) and in rotation during Fall 1999 and Winter 2000. The images were digitally altered so as to remove all cigarette product references to them (see below and Shadel et al., 2002) in order to reduce biases in their responses that may be brought about by presence of cigarettes in the images (see Amos, Gray, Currie, & Elton, 1997). Small groups of adolescents rated each image (presented as slides at a distance of 15 feet) on self-report items and described each image using single word adjectives. Subjects were also asked to try to identify what product from a list that the image was used to advertise. The slide images were presented in random order between different groups of adolescents. The process for selecting gender-valenced cigarette advertising images was as follows. First, only the 11 cigarette advertising images that no more than 30% of participants correctly identified as been taken from cigarette advertisements were considered. Second, two raters (one male, one female) independently sorted each of the 11 images into one of three categories: female valenced (n=3; i.e., tailored for a female audience specifically); male valenced (i.e., n=3; tailored to a male audience specifically; and gender neutral (n=2; i.e., tailored to neither males or females). Images were considered a member of one of these three categories if agreement between the raters was 100%; if raters categorizations did not agree with one another, those images were classified as gender neutral (n=5). Third, images were selected for analysis only if at least one human model was present. This process was used, then, to derive two cigarette advertising images of clear male valence and two of clear female valence. The two female-valenced images were taken from advertisements that promoted a brand of cigarettes smoked by adolescent females more than males, and one of the male-valenced images was taken from an advertisement that promoted a brand of cigarettes smoked by adolescent males more than females (CDC, 1994). Although no adolescent data were available on smoking prevalence rates for the other male-valenced image, studies of adult smokers indicated that the image was taken from advertisements that

W.G. Shadel et al. / Addictive Behaviors 29 (2004) 17351744

1739

promoted a cigarette brand that was smoked by males more than females (CDC, 1990). None of these four images was correctly identified as having been taken from a cigarette advertisement by more than 27% of the sample; there were no gender differences in percent correctly identified for either male-valenced ( P =.67) or female-valenced ( P =.98) images. Percent correctly identified was not related to the dependent variables for either males ( Ps=.93) or for females ( Ps=.84). Affect was assessed on two occasions. First participants responded to the following question at baseline, How do you feel today? using five options (very sad, sad, in the middle, happy, and very happy), that were later assigned values (15, respectively) for analysis. Higher numbers reflected more positive affect. This measure of affect was used as a covariate in later analyses (baseline affect). For each advertising image, participants responded to the following question, How does this picture make you feel? using the five response options noted above. Responses within each image type were averaged to produce an affective response score for each image type, either female valence or male valence (affective response).

2. Results A 2 (gender)2 (image type: male valence; female valence) ANCOVA was used to analyze the affective response data (baseline affect was included as a covariate). Analyses of participants affective responses to the images revealed a significant interaction of gender and image type ( F=5.88, df =1,26, P<.02). As expected, female adolescents reported a greater level of positive affect in response to the female-valenced images compared to the malevalenced images ( P<.0001). Contrary to expectations, no differences in affective response were found between image types for male adolescents ( P =.35). Entry of age and ethnicity (White vs. non-White) as additional covariates did not change these results. Fig. 1A presents the adjusted means and standard errors for participants affective responses. In order to evaluate the possibility that cigarette advertising imagery in general generated more negative affect in males compared to females (which potentially could explain the above results), we examined gender differences in affective responses to the seven cigarette advertising images that were either gender neutral or gender valenced with no human model displayed (as described previously). No significant gender differences were found ( Ps>.24). 2.1. Study 2 2.1.1. Participants Participants were 101 never smoking adolescents recruited using the same media advertising strategy as in Study 1. Inclusion/exclusion criteria, informed consent/assent procedures, and compensation were the same as in Study 1. This sample was 56% female (n=58) and their average age was 14.0 (S.D.=2.0); 70% of the sample was Caucasian, 14% were African American, 9% were Hispanic, and 7% were of other ethnicity. The specific procedures of this study are described in Shadel et al. (in press).

1740

W.G. Shadel et al. / Addictive Behaviors 29 (2004) 17351744

Fig. 1. (A) Means and standard errors of male and female adolescents affective responses to female- and malevalenced images. (B) Means and standard errors of male and female adolescents self-relevance scores to femaleand male-valenced images.

2.1.2. Procedure The imagery stimuli for this study were taken from 11 cigarette advertisements and 12 antismoking advertisements (i.e., image only, no product information displayed) that could not be correctly identified as advertising a particular product by more than 30% of the sample in Study 1 (see Shadel et al., 2002). These advertising images were loaded onto an IBM ThinkPad 300 as bitmap files and were presented using the MediaLab 2000 v 2.1 (Empirisoft Research Software, Philadelphia, PA) experimental design software. Each participant was seated at the computer screen at a distance of 15 inches and was presented with each image individually (in random order between participants) under the instruction to think about how the image was relevant to their self-concept (i.e., reminds you of yourself). Once they had made a determination of the self-relevance of the image, they were instructed to depress either a labeled yes (i.e., reminds me of myself) or no (i.e., does not remind me of myself) on a computer keypad. Only responses to the two femalevalenced images and two male-valenced images (same edited, gender-valenced images used in Study 1) were used in this analysis. A self-relevance score was derived from these procedures as follows. Yes responses to each of the two images within image valence type were coded as 1 and no responses to

W.G. Shadel et al. / Addictive Behaviors 29 (2004) 17351744

1741

each of the two images within image type were coded as 0. The yes and no responses within each image type were averaged to yield a score of the probability that female- and male-valenced images were relevant to the self-concept. Higher scores (0.01.0) reflected a greater degree of relevance between the image and adolescents self-concept. 2.1.3. Results Male and female self-relevance scores were analyzed in a 2 (gender)2 (image type: male valence; female valence) ANOVA. Analyses of participants responses to the images, as expected, revealed a significant interaction of gender and image type ( F=5.36, df =1,99, P<.02). As expected, female adolescents reported a greater degree of self-relevance for female-valenced images compared to the male-valenced images ( P<.0001). Contrary to expectations, however, no differences in imagery self-relevance was found between image types for male adolescents ( P =.26). Entry of age and ethnic group (White vs. non-White) as covariates did not change these basic results. Fig. 1B presents the means and standard errors for participants self-relevance scores. To evaluate whether a gender-specific response bias could explain these significant results, we analyzed for gender differences in self-relevance scores for the remaining cigarette advertisements that were classified as gender neutral or had no human model present. No significant gender differences were found ( P >.21).

3. Discussion The results of these studies, taken together, provide initial evidence for possible psychological mechanisms through which cigarette advertising imagery may exert its effects on adolescent female never smokers. Our results could not be accounted for by a response bias in either gender to cigarette advertising imagery in general (cf. Covell et al., 1994), nor could the results be accounted for by differences in prior exposure to cigarette advertisements or to recognizing the presented images as being taken from cigarette advertisements. Rather, the specific female valence of cigarette advertising imagery was the active ingredient in producing our significant results. As such, these findings may provide an initial explanation as to why cigarette advertising campaigns, putatively targeted at females, have been historically associated with increases in smoking behavior among women (e.g., Pierce, Lee, & Gilpin, 1994; USDHHS, 2001). Adolescent females showed more positive affective reactions to female-valenced cigarette advertising imagery compared to male-valenced imagery. These findings underscore the important role that cigarette advertising imagery has in provoking positive affective reactions among adolescents (Slovic, 2001). Positive affective reactions may be associated with increases in smoking (Slovic, 2001). The findings from Study 2 underscore the importance of the self-concept in understanding adolescent females reactions to the gender valence of cigarette advertising imagery (see Shadel et al., 2001). These results are also conceptually consistent with studies that have found that the more similar adolescents in general see their self-concepts as similar to the self-concepts of what they perceive as typical smokers, the more likely they are to initiate smoking (Gibbons et al., 1991) and those studies that have

1742

W.G. Shadel et al. / Addictive Behaviors 29 (2004) 17351744

found relationships between other features of adolescents developing self-concepts (i.e., selfconflict) and their responses to cigarette advertisements (Shadel et al., in press). As such, the degree to which a cigarette advertisement displays an image that a never smoking female adolescent sees as relevant to her self-concept may be associated with increases her in smoking behavior. Male adolescent never smokers did not respond differently to male-valenced imagery across either of the two indices studied (compared to female-valenced imagery). There are at least two explanations for these results. First, some other non-image-based facet of cigarette advertising (e.g., textual descriptions of cigarettes) may influence never smoking adolescent males. Second, it may be male and female adolescents respond to gender valence of cigarette advertising imagery not in quantitatively different ways, but in qualitatively different ways. Self-concept and affect may be important dimensions for female adolescents, whereas some other advertising image factor not assessed in this study may influence male adolescents. Future research needs to address this question in detail. Limitations to the work presented in this manuscript need to be considered. First, the use of only four cigarette advertising images (two of each gender valence type) that display human models may restrict the generalizability of the findings. Other cigarette advertisements use no human models and our results may not be applicable to these cigarette advertisements or to other gender-valenced cigarette advertisements. Second, we only evaluated adolescents reactions to the images in cigarette advertisements, and not the actual advertisements. Our rationale was that the presence of a cigarette in the advertisements would have biased adolescents responses (see Amos et al., 1997). Still, adolescents responses to the advertisements themselves may differ. Third, the small sample size of Study 1 may limit the generalizability of those findings, and the fact that all participants in both studies were never smokers also limits the generalizability of the findings but provides an opportunity for future research to examine the moderating effect of smoking status on adolescents reactions to cigarette advertising. Fourth, our main dependent measures were composed of a single item each. Measures with a greater number of items could have improved reliability of those measures and may have produced stronger results. Finally, even though our findings were not tied to smoking intentions or to smoking behavior, they do suggest that mechanisms that have been shown to have an impact on these variables have a gender-specific component with respect to cigarette advertising that is worthy of future investigation. For example, future work could investigate what particular features of gender-valenced advertising (e.g., facial expression of model used in the advertisement or physical attributes of the models) adolescents find appealing and why (i.e., identify the active ingredients within a gendervalenced advertisement). Such work may permit a more detailed understanding of how cigarette advertisements might influence adolescents smoking behavior and perhaps help to limit their potential impact on smoking initiation.

Acknowledgements This research was supported by grant DA12350 from the National Institute on Drug Abuse.

W.G. Shadel et al. / Addictive Behaviors 29 (2004) 17351744

1743

The authors wish to thank Santina Ficara, Jennifer Wear George, and Megan Dombloski for their assistance in executing the study procedures and with data management and coding. A version of the work reported in this manuscript was presented in April 2002 at the annual meeting of the Society of Behavioral Medicine, Washington, DC.

References
Amos, A., Gray, D., Currie, C., & Elton, R. (1997). Healthy or druggie? Self-image, ideal image, and smoking behavior. Social Science Medicine, 45, 847 858. Centers for Disease Control (CDC) (1994). Changes in the cigarette brand preferences of adolescent smokers United States, 1989 1993. Morbidity and Mortality Weekly Report, 43, 577 581. CDC (1990). Cigarette brand use among adult smokersUnited States 1986. Morbidity and Mortality Weekly Report, 39, 671 673. CDC (2002a). Trends in cigarette smoking in high school studentsUnited States 1991 2001. Morbidity and Mortality Weekly Report, 51, 409 412. CDC (2002b). Youth risk behavior surveillanceUnited States, 2001. Morbidity and Mortality Weekly Report, 51(SS04), 1 64. Covell, K. (1992). The appeal of image advertisements: Age, gender, and product differences. Journal of Early Adolescence, 12, 46 60. Covell, K., Dion, K. L., & Dion, K. K. (1994). Gender differences in evaluations of tobacco and alcohol advertisements. Canadian Journal of Behavioural Science, 26, 404 420. DiFranza, J., Rigotti, N., McNeill, A. D., Ockene, J. K., Savageau, J. A., St Cyr, D., & Coleman, M. (2000). Initial symptoms of nicotine dependence in adolescents. Tobacco Control, 9, 313 319. Flay, B., & Petraitis, J. (1994). The theory of triadic influence: A new theory of health behavior with implications for preventive interventions. In G. L. Albrecht (Ed.), Advances in medical sociology. Vol. 4: A reconsideration of models of health behavior change ( pp. 19 44). Greenwich, CT: JAI. Geuens, M., & DePelsmacker, P. (1998). Feelings evoked by warm, erotic, humorous, or non-emotional print advertisements for alcoholic beverages. Academy of Marketing Science Review, 98 (Available: http:// www.amsreview.org/amsrev/theory/geuens01-98.html). Gibbons, F. X., Gerrard, M., Lando, H. A., & McGovern, P. G. (1991). Social comparison and smoking cessation: The role of the typical smoker. Journal of Experimental Social Psychology, 27, 239 258. Harter, S. (1999). The construction of the self: A developmental perspective. New York: Guilford Press. Johnston, L. D., OMalley, P. M., & Bachman, J. G. (2003). Monitoring the future national survey results on drug use, 1975 2002. Volume I: Secondary school students. Bethesda, MD: National Institute on Drug Abuse, NIH Publication No. 03-5375. Krupka, L. R., & Vener, A. M. (1992). Gender differences in drug (prescription, non-prescription, alcohol, and tobacco) advertising: Trends and implications. Journal of Drug Issues, 22, 339 360. Krupka, L. R., Vener, A. M., & Richmond, G. (1990). Tobacco advertising in gender-oriented popular magazines. Journal of Drug Education, 20, 15 29. National Institute on Drug Abuse (2000). Mind over matter: The brains response to nicotine, NIH Publication:00-4248. National Institutes of Health Rockville, MD. Pidoplichko, V. I., DeBiasi, M., Williams, J. T., & Dani, J. A. (1997). Nicotine activates and desensitizes midbrain dopamine neurons. Nature, 390(6658), 401 404. Pierce, J. P., Lee, L., & Gilpin, E. A. (1994). Smoking initiation by adolescent girls, 1944 1988: An association with targeted advertising. Journal of the American Medical Association, 271, 608 611. Romer, D., & Jamieson, P. (2001). Advertising, smoker imagery, and the diffusion of smoking behavior. In P. Slovic (Ed.), Smoking: Risk, perception, and policy ( pp. 127 155). Thousand Oaks, CA: Sage. Shadel, W. G., Niaura, R., & Abrams, D. B. (2001). How do adolescents process smoking and anti-smoking

1744

W.G. Shadel et al. / Addictive Behaviors 29 (2004) 17351744

advertising? A social cognitive analysis with implications for smoking initiation. Review of General Psychoogy, 5, 429 444. Shadel, W. G., Niaura, R., & Abrams, D. B. (2002). Adolescents responses to the imagery displayed in smoking and anti-smoking advertisements. Psychology of Addictive Behaviors, 16, 173 176. Shadel, W. G., Niaura, R., & Abrams, D. B. (in press). Who am I? The role of self-conflict in adolescents responses to cigarette advertising. Journal of Behavioral Medicine. Slovic, P. (2001). Cigarette smokers: Rational actors or rational fools? In P. Slovic (Ed.), Smoking: Risk, perception, and policy ( pp. 97 124). Thousand Oaks, CA: Sage. USDHHS (2001). Women and smoking: A report of the surgeon general. Atlanta, GA: USDHHS, CDC, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. Wakefield, M., Flay, B., Nichter, M., & Giovino, G. (2003). Role of the media in influencing trajectories of youth smoking. Addiction, 98(S1), 79 103. Warner, K. (1985). Tobacco industry response to public health concern: A content analysis of cigarette ads. Health Education Quarterly, 12, 115 127.