Articles - UF Department of Statistics

Journal of Health Communication, Volume 6, pp. 11È29, 2001 Copyright Ó 2001 Taylor & Francis 1081-0730/01 $12.00 1 .00

Articles Measuring Exposure to Pro- and Anti-tobacco M arketing Among Adolescents : Intercorrelations Am ong M easures and Associations W ith Smoking Status JE N N IF E R B . U N G E R TESS BO LEY CRU Z D A R L E E N S C H U ST E R University of Southern California School of Medicine Los Angeles, California, USA

JU N E A . F L O R A Stanford Center for Research in Disease Prevention Stanford, California, USA

C . A N D E R SO N J O H N SO N University of Southern California School of Medicine Los Angeles, California, USA Exposure to tobacco -related marketing has been implicated as one of the risk factors for tobacco use among adolescents . However, tobacco -related marketing exposure has been measured in di†erent ways in di†erent studies, including per ceived pervasiveness , receptivity , recognition , recall, and a†ect. It is not known whether these measures represent one or more underlying constructs and how these underlying constructs are associated with adolescent smoking status . T his study analyzed data from 5 ,870 eighth -grade students in California, collected in 1996È 1997 as part of the Independent Evaluation of the California T obacco Control , Prevention, and Education Program . An exploratory factor analysis of multiple measures of tobacco -related marketing exposure revealed four distinct factors : per ceived pervasiveness of protobacco marketing , perceived pervasiveness of anti tobacco marketing, recognition of speciÐc antitobacco advertisements , and receptivity to protobacco marketing. Receptivity to protobacco marketing showed the strongest association with smoking status ; higher levels of receptivity were Collection of the data described in this article was supported by a contract from the California Department of Health Services, Tobacco Control Section (Contract 95-222998). Data analysis and manuscript preparation were supported by funds from the California Tobacco-Related Disease Research Program, Grant 7KT-0006. The analyses, interpretations, and conclusions are those of the authors, not the California Department of Health Services. Address correspondence to Jennifer B. Unger, Ph.D., Institute for Health Promotion and Disease Prevention Research, University of Southern California School of Medicine, 1540 Alcazar St., CHP 207, Los Angeles, CA 90089-9008.

11

12

J . B . Unger et al. associated with higher levels of smoking. T wo measures of exposure to antitobacco marketing (perceived pervasiveness of antitobacco marketing and recognition of spe ciÐc antitobacco ads ) were highest among established smokers and lowest among susceptible nonsmokers . T he same pattern was evident for perceived pervasiveness of protobacco marketing . Results suggest that exposure to tobacco -related market ing is a multidimensional construct , and each dimension may have a unique contri bution to the process of smoking initiation . Because adolescents are exposed to numerous pro- and antitobacco messages, it is important to develop antitobacco media campaigns that can successfully counter protobacco marketing e†orts . Poten tial strategies include targeting the susceptible nonsmokers who are at high risk for smoking and developing messages to decrease receptivity .

Although the prevalence of adult smoking in the United States has declined over the past decade, adolescent smoking prevalence continues to increase. Nationwide, 19% of eighth-grade students reported past-month smoking in 1998, as compared with 14% in 1991 (Johnston, OÏMalley, & Bachman, 1998). In 1998, 46% of eighth-grade students reported that they had tried smoking, and 9% were daily smokers (Johnston et al., 1998). The high prevalence of smoking among adolescents indicates an urgent need to determine the risk factors for adolescent smoking and to develop e†ective interventions and policies to prevent adolescent smoking. Iný uence of Tobacco -Related Marketing on Smoking

Numerous studies have implicated exposure to tobacco-related marketing as one of the risk factors for smoking, Historically, smoking prevalence in certain demographic groups has increased following the introduction of tobacco advertising campaigns targeting those groups. Smoking among young males increased following the 1912 introduction of Camel cigarettes, whereas smoking among young females increased during the 1920s following the ChesterÐeld and Lucky Strike womenÏs marketing campaigns (Pierce & Gilpin, 1995). More recently, smoking among youth increased following the introduction of Joe Camel in the 1980s (Pierce et al., 1991). In addition, individual di†erences in receptivity to tobacco marketing appear to explain some of the risk for adolescent smoking. Adolescents who have favorite tobacco advertisements or own tobacco promotional items are at increased risk of smoking initiation (Feighery, Borzekowski, Schooler, & Flora, 1998 ; Pierce, Gilpin et al., 1998). Other studies also have found associations between self-reported tobacco marketing exposure and adolescent smoking (Altman, Levine, Coeytaux, Slade & Ja†e, 1996 ; Arnett & Terhanian, 1998 ; Botvin, Goldberg, Botvin & Dusenbury 1993 ; Schooler, Feighery & Flora, 1996). These studies indicate that tobacco marketing may be an important factor in adolescent smoking initiation. Iný uence of Antitobacco Media on Smoking

Because protobacco marketing is so ubiquitous (Ammerman & Nolden, 1995), numerous counteradvertising campaigns have attempted to counteract the e†ects of tobacco marketing on smoking behavior. Some of these mass media campaigns have been e†ective in reducing or preventing adolescent smoking behavior (Hafstad & Aaro, 1997 ; Hafstad, Aaro, & Langmark, 1996 ; Hafstad et al., 1997 ; Worden et al., 1996). However, other studies have found no association between exposure to antitobacco media campaigns and adolescent smoking behavior or intentions to quit smoking (Bauman, La Prelle, Brown, Koch, & Padgett, 1991 ; Murray, Prokhorov, & Harty, 1994 ; Popham, Potter, Hetrick & Muthen, 1991). Although not all studies of antitobacco media have found signiÐcant e†ects, the research described

Adolescent Exposure to Pro - and Anti-tobacco Marketing

13

above indicates that both protobacco marketing and antitobacco media campaigns may inÑuence adolescent smoking. Measures of Exposure to Tobacco -Related Media

Despite the growing evidence of an association between tobacco-related marketing and adolescent smoking, there is not a clear consensus about how to measure exposure to tobacco-related marketing. Which aspects of protobacco marketing make adolescents more likely to smoke, and which aspects of antitobacco marketing make them less likely to smoke ? Exposure to tobacco-related marketing potentially could be deÐned as (1) noticing the message (e.g., recognition and recall measures), (2) making judgments, about how pervasive the message is, (3) changing attitudes or behaviors in response to the message (e.g., wanting or owning tobacco promotional items), or (4) expressing positive a†ect toward a message (e.g., liking or preference). Many di†erent measures have been used, as described below. Recognition and recall of speciÐc ads . One measure of exposure to tobaccorelated marketing involves asking respondents to identify or recall information about speciÐc advertisements. Respondents are given a picture or description of an ad and are asked to provide information that they would know only if they had seen the ad, such as the brand name or slogan. For protobacco ads, the experimenter can show respondents pictures of ads (with all references to the brand name obscured) and ask them to identify the brand being advertised (Unger, Johnson, & Rohrbach, 1995). For antitobacco ads, which typically do not mention speciÐc brands, the experimenter can ask questions about the meaning of the ads to verify comprehension (Independent Evaluation Consortium, 1998). For example, in the Independent Evaluation of the California Tobacco Control, Prevention, and Education Program (Independent Evaluation Consortium, 1998), respondents were given a short written description of an antitobacco ad and were asked to identify the central message of the ad. This method can measure recognition (whether the respondent reports having seen the ad) and validated recognition (whether the respondent correctly comprehended the ad). One potential limitation of recognition measures is the accuracy of responses. That is, subjects may recall a similar ad, or they may claim to have seen an ad in an attempt to ““save faceÏÏ with experimenters. The use of a follow-up question to validate recognition is a more conservative and rigorous approach. Perceived pervasiveness of pro - and antitobacco marketing . Another indicator of exposure to tobacco-related marketing is the respondentÏs perception of the pervasiveness of this marketing. To measure this, Schooler, Feighery, and Flora (1996) asked respondents to rate how frequently they see or hear pro- or antitobacco advertisements on television, radio, or billboards or at events or stores. These measures typically are rated on a Likert-type scale, ranging from ““a lotÏÏ or ““oftenÏÏ to ““never.ÏÏ An advantage of this method is that these measures are not dependent on the content of speciÐc ads ; therefore, these measures do not become outdated as advertising campaigns change, and responses can be compared over time. Unfortunately, because these measures do not prompt memories of speciÐc ads, respondents may fail to retrieve from memory all the di†erent tobacco-related ads they have seen. In addition, these measures require respondents to make subjective judgments about the meanings of terms such as ““a lotÏÏ or ““hardly ever.ÏÏ Respondents who have strong negative opinions about protobacco marketing may rate protobacco marketing as more pervasive, regardless of the number of ads they

14

J . B . Unger et al.

have actually seen. Respondents with strong antitobacco attitudes may be more attentive to antitobacco ads and may therefore rate them as more pervasive. More fundamentally, this measure may fail to account for the quality of exposure or the audienceÏs level of attention, which may engender quite di†erent behavioral outcomes. In a validation study of adultsÏ self-reported television viewing behavior (Bechtel, Achelpohl, & Akers, 1971), respondents frequently were shown to be completely or partially engaged in other tasks during the times when they claimed to have been watching television. This may be true for self-reported viewing of billboards, magazines, and other locations where tobacco advertising exists as well. Hence, this measure may lack validity since inÑated estimates of true exposure are highly possible. Receptivity to tobacco marketing . Receptivity describes the successive stages of attachment to tobacco-related products. McGuire (1985) describes three elements of receptivity : (1) exposure to the message, (2) attention to and comprehension of the message, and (3) cognitive/a†ective response to the message. A similar sequence of stages is described by Peter and Olson (1990) : (1) awareness of, interest in, and shifting attitudes toward the product ; (2) desire and intentions to use the product ; (3) experimentation with the product ; and (4) regular use of the product. These sequences provide a general description of the stages of attachment to a product. When applied to protobacco marketing, these stages can be operationally deÐned. Feighery et al. (1998) described three stages of receptivity to protobacco marketing : (1) seeing tobacco ads, (2) wanting to own tobacco promotional items, and (3) owning tobacco promotional items. To evaluate the e†ects of tobacco advertising and promotion on the susceptibility of smoking on never-smoking adolescents, Evans, Farkas, Gilpin, Berry, and Pierce (1995) devised a 5-point index of an individualÏs receptivity to tobacco advertising. These Ðve measures included (1) recognizing advertising messages, (2) having a favorite tobacco advertisement, (3) naming a brand she or he may buy, (4) owning a tobacco-related promotion item, and (5) being willing to use a tobacco-related promotional item. Those adolescents who scored higher on the receptivity to tobacco advertising index were more likely to be susceptible to smoking. The relationship between receptivity to tobacco advertising and smoking susceptibility was stronger than that of an individualÏs self-reported exposure to family/peer smoking, a second variable under investigation. Measures of a†ective responses . Most of the techniques described above have examined exposure to tobacco-related marketing by assessing explicit cognition. That is, respondents are directly asked to recall their exposure to tobacco-related marketing and provide various details about speciÐed advertisements. Another option (used less frequently in tobacco research) is to use implicit cognition measures, which assess exposure without asking respondents to recollect past experiences. Instead, subjects may be asked to indicate their liking of an advertisement using a Likert-type scale (Aitken, Eadie et al., 1988 ; Aitken, Leathar, & Scott, 1988 ; Unger et al., 1995 ; Wyllie, Zhang, & Casswell, 1998a, 1998b) or to provide openended responses regarding various features of an advertisement (Slater et al., 1996). These preference judgments and cognitive response techniques are based on the ““mere exposureÏÏ e†ect, which states that preferences for stimuli may arise because of repeated exposures, even when individuals fail to recall those exposures (Borstein, 1989 ; Zajonc, 1968). Therefore, preference for tobacco-related advertisements may indicate that the individual previously has been exposed to these advertisements, even if the individual cannot retrieve them from memory. Implicit cognition measures primarily have been used to assess exposure to alcohol advertising (Aitken, Eadie et al., 1988 ; Di Franza et al., 1991 ; Slater et al., 1996 ; Unger et al., 1995 ;


15

Wyllie et al., 1988a, 1998b), although they are applicable to tobacco advertising as well. The few studies utilizing preference judgments in tobacco-related marketing have found signiÐcant positive associations between exposure and ad liking (Arnett & Terhanian, 1998) and between ad liking and smoking behavior (Unger et al., 1995). Although they have not been used widely in tobacco research, implicit cognition measures may be useful because they can assess media exposures that may not be cognitively available to respondents. Additionally, they do not require respondents to provide self-report data, thereby minimizing threats to internal validity. As described above, various methods have been employed to assess di†erent aspects of exposure to tobacco-related marketing. However, it is not known whether these measures are assessing a single underlying construct or multiple constructs. It also is not known which of these measures is associated most strongly with smoking behavior. This study evaluated various measures of pro- and antitobacco marketing exposure, using data from 5,870 eighth-grade students collected in California as part of the Independent Evaluation of the California Tobacco Control, Prevention, and Education Program (Independent Evaluation Consortium, 1998). The Independent Evaluation contained several di†erent questionnaire items designed to assess tobacco-related marketing exposure. The purpose of this study was to determine whether the various measures would cluster into meaningful factors and to determine whether these factors were associated with smoking status. M ethod Data Source

This study used data from a representative sample of eighth-grade California youth (N 5 5870). The data were collected in schools in 18 California counties, and the sample was weighted to represent the population of California youth. The data were collected during the 1996 È 1997 school year as part of the Independent Evaluation of the California Tobacco Control, Prevention, and Education Program (Independent Evaluation Consortium, 1998). The data were collected in the studentsÏ classrooms by trained data collectors during a single class period (approximately 45 minutes), using paper-and-pencil surveys. Because the survey was anonymous, an implied parental consent procedure was used. Consent forms were sent to parents of all students in the selected classrooms. Parents were instructed to sign and return the consent form only if they refused their child permission to participate. Students were allowed to decline participation at any time. Using this method, 96% of all students in the selected classrooms provided both parental and student consent and completed the survey. Measures Perceived pervasiveness of protobacco marketing . The following items were adapted from Schooler, Feighery, and Flora (1996) : ““When you see billboards, how often do you see advertisements for cigarettes or chewing tobacco ?ÏÏ ““When you go to a small market, convenience store, or gas station mini-mart, how often do you see advertisements for cigarettes or chewing tobacco, or items like sports gear, t-shirts, hats, sunglasses, lighters, or ashtrays that have tobacco company names or pictures on them ?ÏÏ ““When you watch TV, how often do you see people smoking ?ÏÏ These questions were rated on a 4-point scale ranging from ““neverÏÏ to ““a lot.ÏÏ

16


Perceived pervasiveness of antitobacco marketing . The following items were adapted from Schooler, Feighery, and Flora (1996) : ““When you listen to the radio, how often do you hear advertisements about not smoking or chewing tobacco ?ÏÏ ““When you see billboards, how often do you see advertisements about not smoking or chewing tobacco ?ÏÏ ““When you watch TV, how often do you see stories or advertisements about the dangers of smoking and chewing tobacco ?ÏÏ ““When you go to a small market, convenience store, or gas station mini-mart, how often do you see advertisements about not smoking or chewing tobacco ?ÏÏ These questions were rated on a 4-point scale ranging from ““neverÏÏ to ““a lot.ÏÏ Exposure to protobacco marketing . The following measures were adapted from Pierce et al. (1994) : ““How often do you see other young people wearing clothing or carrying gear, like t-shirts, lighters, gym bags, hats, or sunglasses, that have a tobacco company name or picture on it ?ÏÏ (rated on a 4-point scale ranging from ““neverÏÏ to ““a lotÏÏ). ““How many items do you own that have a tobacco company name or picture on it ?ÏÏ (rated on a 4-point scale ranging from ““noneÏÏ to ““3 or 4 itemsÏÏ). ““Do you think you would ever use or wear something that has a tobacco company name or picture on it, such as a lighter, T-shirt, hat, or sunglasses?ÏÏ (rated on a 4-point scale ranging from ““noÏÏ to ““yes, deÐnitelyÏÏ). ““What is the make of the cigarette brand of your favorite cigarette advertisement ?ÏÏ (coded as 1 if the respondent indicated a favorite, or 0 otherwise). Recognition of speciÐc antitobacco ads . Respondents were asked whether they recalled seeing or hearing, during the past year, speciÐc antitobacco ads that were part of the California Tobacco Control, Prevention, and Education Program. These ads included two television ads, two radio ads, and two billboards that were disseminated throughout California at the time of the survey. If respondents recalled having seen an ad, they were asked a multiple-choice question to validate their comprehension of the ad. The validated recognition score indicates the number of ads the respondent recognized and comprehended correctly (range 5 0 È 6). Because the focus of the current study is on exposure to tobacco-related media in general and not the speciÐc ads in the California media campaign, we used the number of ads recognized correctly as a summary measure of recognition of speciÐc antitobacco ads. This measure was designed to give high scores to respondents who recognized many ads and low scores to respondents who recognized few ads, regardless of which speciÐc California ads they recognized. Smoking status . Respondents were grouped into four stages of smoking initiation : never smokers, susceptible, experimenters, and established smokers. The criteria used to classify respondents into groups are illustrated in Figure 1. This classiÐcation scheme is similar to that described by Pierce (1994), but we created separate categories for the respondents who had ever tried smoking (experimenters and established smokers) and those who had not (never smokers and susceptible). Covariates . Demographic covariates included age, gender, ethnicity, language spoken at home, and school performance. These variables have been associated with smoking in previous studies (U.S. Department of Health and Human Services, 1994). Tobacco-related psychosocial covariates included a seven-item scale of tobacco-related knowledge, a Ðve-item scale of perceived negative consequences of smoking, a Ðve-item scale of perceived positive consequences of smoking, one item assessing best friendsÏ smoking, and one item assessing perceived access to cigarettes (Independent Evaluation Consortium, 1998).


17

F IG U R E 1 Flowchart for determining smoking status.

Data Analysis

Means and standard deviations were calculated for all media variables. The 12 media variables were entered into an exploratory factor analysis with promax rotation. The number of factors was determined by the number of eigenvalues greater than 1. The factor loadings were examined to determine which media variables clustered into common factors. Factor loadings greater than .30 were considered meaningful (Kim & Meuller, 1978). After determining which media variables loaded on each factor, factor scores were created. The variables comprising each factor were standardized to a mean of 0 and a standard deviation of 1 and then summed together to form a factor score. Correlation analyses were used to calculate the correlations among the factor scores. Finally, analysis of variance (ANOVA) was used to determine whether the factor scores di†ered by smoking status. Smoking status was coded as a four-level class variable. To control for confounding, the model was controlled for the demographic and psychosocial covariates described above. Because the respondents were sampled using a complex multistage sampling procedure (Independent Evaluation Consortium, 1998), random e†ects were included for sampling strata, county, and school. The inclusion of these random e†ects controls for the intraclass correlation that may result from the sampling design (Singer, 1998). Because the sampling strata were based on population density and locations of major media markets, this procedure also controls for between-county di†erences in urbanicity and media exposure, which may confound the results. Results Demographic Characteristics of Sample

Table 1 shows the demographic characteristics of the respondents. The mean age of the respondents was 13.2 years (standard deviation 5 0.6 years). The sample was approximately half male and half female. Approximately one-third of the respondents were White (33%), followed by Hispanic/Latino (22%), Asian/PaciÐc Islander (15%), multiethnic (14%), and African American (8%). Slightly over half of the respondents (56%) reported speaking only English at home.


18

T A B L E 1 Demographic Characteristics of Respondents

(N 5

5870)

Age 11È 12 13 14 15 È 16 Missing Gender Male Female Missing Ethnicity White Hispanic/Latino Asian/PaciÐc Islander African American Multiethnic Other/missing Language spoken at home Only another language Mostly another language Half English and half another language Mostly English Only English Missing Smoking status Never smoker Susceptible Experimenter Established smoker Missing

N

%

278 4120 1395 43 33

4.8 70.2 23.8 0.7 0.6

2913 2894 63

49.6 49.3 1.1

1955 1285 859 485 837 449

33.3 21.9 14.6 8.3 14.3 7.6

112 261 1457 749 3262 29

1.9 4.4 24.8 12.8 55.6 0.5

2185 941 2227 468 49

37.2 16.0 37.9 8.0 0.8

Descriptive Statistics on Media Exposure Measures

Table 2 shows the percent of respondents who reported exposure to tobacco-related marketing, as measured by each of the questionnaire items. These percentages differed signiÐcantly by smoking status, as shown in the table. In general, established smokers perceived that both protobacco and antitobacco marketing were more pervasive, compared with respondents in the other stages of smoking initiation. The only exception was protobacco billboards, which were perceived to be most pervasive by never smokers. In addition, perceived pervasiveness of protobacco marketing was higher than perceived pervasiveness of antitobacco marketing.

19

* p < .001.

Perceived pervasiveness of antitobacco marketing (% who responded ““a lotÏÏ) Antitobacco media exposure Radio Billboards TV Stores Perceived pervasiveness of pro -tobacco marketing (% who responded ““a lotÏÏ) Protobacco media exposure Billboards Stores See people smoking on TV See young people wearing or using tobacco promotional items Media receptivity Have favorite cigarette ad (% who named a favorite) Ownership of tobacco promotional items (% who own any items) Would wear or use tobacco promotional items (% who responded ““yes, deÐnitelyÏÏ)

Variable

T A B L E 2 Media Exposure and Receptivity by Smoking Status

11 14 20 8

40 40 36 20

39 38 9

50 43 40 19

18 24 4

Susceptible

12 11 21 11

Never smoker

14

59 47

46 42 43 23

11 16 23 12

Experimenter

43

86 62

45 56 52 39

22 19 28 20

Established smoker

795.7*

1062.8* 341.4*

38.9* 51.0* 41.1* 128.1*

46.5* 51.7* 34.7* 42.4*

Chi-square

20 .7338 .7045 .6026 .5972 È È È È È È È È 2.14

Measure

Antitobacco, billboards Antitobacco, stores Antitobacco, radio Antitobacco, TV Protobacco, stores Protobacco, billboards Smoking on TV See youth with promotional items Would use promotional item Own promotional items Have favorite cigarette ad Recognition of speciÐc antitobacco ads

Eigenvalue

1.71

È È È È .7474 .6106 .6818 .5778 È È È È

Factor 2 Perceived pervasiveness of protobacco marketing

Note . The dashes represent factor loadings that were less than .3 after promax rotation.

Factor 1 Perceived pervasiveness of antitobacco marketing

T A B L E 3 Factor Loadings of Media Exposure and Receptivity Measures

1.46

È È È È È È È È .7694 .7400 .6962 È

Factor 3 Media receptivity

1.04

È È È È È È È È È È È .8882

Factor 4 Recognition of speciÐc antitobacco ads


21

Factor Structure of Exposure Measures

Table 3 shows the results of the exploratory factor analysis. The measures clustered into four factors : perceived pervasiveness of antitobacco marketing, perceived pervasiveness of protobacco marketing, media receptivity and recognition of speciÐc antitobacco ads. The items composing each of these factors are shown in Table 3. To create scale scores, the items within each factor were standardized to a mean of 0 and standard deviation of 1 and then summed together. The CronbachÏs alphas for these scales were .57 for pervasiveness of antitobacco marketing, .58 for pervasiveness of protobacco marketing, and .55 for receptivity. CronbachÏs alpha was not computed for recognition of speciÐc antitobacco ads because this scale was the sum of six items that would not necessarily be intercorrelated.

Correlations among Factors

Table 4 shows the intercorrelations among the factors. All of the correlations were in the positive direction and statistically signiÐcant, although they accounted for a small proportion of the variance. The strongest correlation was between perceived pervasiveness of protobacco marketing and perceived pervasiveness of antitobacco marketing.

Associations between Tobacco -Related Marketing Exposure and Smoking Status

Figure 2 shows the associations between the four factor scores and smoking, controlling for the demographic and psychosocial covariates. Perceived pervasiveness of protobacco marketing was highest among established smokers, higher than among susceptibles (p < .005), but not signiÐcantly di†erent from the other three groups. Perceived pervasiveness of protobacco marketing was lowest among susceptibles, signiÐcantly di†erent from never smokers (p < .001), experimenters (p < .005), and established smokers (p < .005).

T A B L E 4 Correlations among Media Exposure Scales

Perceived pervasiveness of antitobacco marketing Perceived pervasiveness of antitobacco marketing Perceived pervasiveness of protobacco marketing Media receptivity Recognition of speciÐc antitobacco ads * p < .005.

Perceived pervasiveness of protobacco marketing

Media receptivity

Recognition of speciÐc antitobacco ads

1 .09*

1

1

.23*

1

.04* .04*

.18* .14*

22


F IG U R E 2 Tobacco-related media exposure by smoking status.

A similar pattern was evident for perceived pervasiveness of antitobacco marketing. Perceived pervasiveness of antitobacco marketing was highest among established smokers, higher than among never smokers (p < .05) and susceptibles (p < .0005). It was lowest among susceptibles, signiÐcantly lower than never smokers (p < .05), experimenters (p < .0005), and established smokers (p < .0005). The associations between recognition of speciÐc antitobacco ads and smoking status also followed a similar pattern. Recognition of speciÐc antitobacco ads was highest among established smokers and was signiÐcantly higher than never smokers (p < .05) and susceptibles (p < .005). Recognition of speciÐc antitobacco ads was lowest among susceptibles and was signiÐcantly lower than experimenters (p < .05) and established smokers (p < .05). Receptivity scores were lowest among never smokers and increased steadily through the stages of susceptibility, experimentation, and established smoking. All between-group di†erences were signiÐcant at p < .0005.

Discussion The results of this study indicate that the measures of tobacco-related marketing exposure examined in this study can be grouped into four categories : perceived pervasiveness of protobacco marketing, perceived pervasiveness of antitobacco marketing, recognition of speciÐc antitobacco ads, and receptivity to protobacco marketing. These four factors may represent di†erent aspects or degrees of involvement with tobacco-related marketing exposure, ranging from mere exposure to a deliberate e†ort to obtain tobacco-related products. Although these four factors were modestly intercorrelated, they appear to represent separate dimensions of tobaccorelated marketing exposure. Tobacco-related marketing exposure appears not to be a unidimensional construct, suggesting that multiple measures may be necessary to assess this important risk factor for adolescent smoking. In conducting health communication research, it may be a mistake to assume that the adolescents who per-


23

ceive tobacco-related marketing as most pervasive are also those who will be most a†ected by its messages. It is important for researchers and developers of health communication messages to determine which aspects of tobacco-related marketing exposure they are trying to assess or inÑuence and to select the appropriate measures for their purposes. Because tobacco-related marketing exposure is a multidimensional construct, the choice of marketing exposure measures in a study may inÑuence the results obtained in the study and the conclusions drawn from those results. A study that uses recognition or recall measures can assess the general level of exposure to tobacco-related media in a sample, which is useful for identifying the demographic subgroups that are reached e†ectively and the subgroups that may be targeted heavily by a speciÐc marketing campaign. However, recognition and recall measures cannot assess the respondentsÏ level of involvement or attachment. Receptivity measures may be more useful for assessing the degree to which respondents are personally and actively involved in a tobacco marketing campaign. The positive intercorrelations among the four factors indicate that there may be an underlying factor of overall marketing exposure that inÑuences adolescentsÏ exposure to pro- and antitobacco marketing. Not surprisingly, adolescents who perceive pervasive exposure to antitobacco marketing also perceive that protobacco marketing is pervasive. These may be the adolescents who tend to pay more attention to media and advertising overall, so they are more likely to recall both pro- and antitobacco marketing. Unfortunately, even in states with extensive tobacco control programs such as California, protobacco spending greatly outweighs antitobacco spending (Pierce, Choi, Gilpih, Farkas, & Berry, 1998). Therefore, people who are exposed to antitobacco media campaigns probably are exposed to high levels of protobacco marketing as well. If evaluations of tobacco control media campaigns fail to consider the confounding e†ects of protobacco marketing exposure, they may mistakenly conclude that antitobacco media campaigns are not e†ective, when in reality their e†ects are merely being obscured by the more pervasive protobacco inÑuences. Consistent with other studies (Altman et al., 1996 ; Feighery et al., 1998 ; Pierce et al., 1998), receptivity to protobacco marketing was higher among adolescents who were at more advanced levels of smoking initiation. These results, combined with the results of previous studies, indicate that receptivity to protobacco marketing is a strong correlate of progression along the smoking uptake continuum. Previous studies also have shown this association longitudinally ; adolescents with higher receptivity were more likely to progress to more advanced stages of smoking initiation (Pierce et al., 1998). The results of this study are consistent with the results of previous studies, which have found associations between receptivity and adolescent smoking. Of course, because these data are cross sectional, these results do not prove that receptivity causes smoking. Receptivity to protobacco marketing also may be a consequence of becoming a smoker and identifying oneself as a smoker. The other three factors (perceived pervasiveness of pro- and antitobacco marketing and recognition of speciÐc antitobacco ads) did not show this monotonic association. These measures were highest among established smokers and lowest among susceptibles. One possible explanation for the high pro- and antitobacco marketing exposure among smokers is that smokers may have higher overall media exposure. Many of the risk factors for smoking also are correlates of heavy television viewing, such as depression (Dittmar, 1994), lack of parental supervision (Brooks, Gaines, Mueller & Jenkins, 1998), family conÑict (Brooks et al., 1998), and

24


low involvement in education and reading (Clarke & Kutz-Costes, 1997 ; Koolstra, van der Voort, & van der Kamp, 1997). Therefore, adolescent smokers may watch more television than nonsmokers do, thereby exposing themselves to more antitobacco television ads. In addition, adolescent smokers who tend to have less parental supervision and school involvement may spend more time at convenience stores and sporting events where tobacco is advertised. Adolescent smokers may perceive tobacco-related marketing to be more pervasive because they may tend to place themselves in situations where they will be exposed to higher levels of marketing overall. However, it is likely that the associations between tobacco-related marketing exposure and smoking status reÑect more than just adolescent smokersÏ high levels of media usage. The high perceived pervasiveness of protobacco marketing among established smokers suggests that adolescents who notice and attend to protobacco marketing may be more likely to become smokers. Alternatively, this may indicate that adolescents who are smokers are more likely to attend to and recall protobacco advertising. Tobacco-related ads, whether pro- or antitobacco, may be more salient to smokers and may elicit pleasant thoughts about smoking. Therefore, smokers may be more likely to recall the tobacco-related imagery in ads. This may create a vicious circle, in which protobacco marketing increases the probability of smoking, and smoking increases the future salience of protobacco marketing. The high perceived pervasiveness of antitobacco marketing among established smokers is counterintuitive. Antitobacco media campaigns are designed to prevent smoking ; therefore, adolescents who recall more antitobacco advertising would be expected to show lower levels of smoking. However, this prediction is based on the assumption that antitobacco media campaigns are e†ective in preventing smoking. Unfortunately, some antitobacco media campaigns, such as those sponsored by the tobacco industry, may send mixed messages to youth. For example, a recent tobacco-industry-sponsored campaign implies that smoking is acceptable for adults but not for adolescents (i.e., that smoking is a way to appear grown-up) (Americans for NonsmokersÏ Rights, 1998). Other antitobacco advertisements may use strategies that are not based on theories of e†ective health communication. For example, they may rely heavily upon fear appeals or emphasize the long-term consequences of smoking rather than the immediate consequences. Ine†ective antitobacco advertisements may have no e†ect on adolescent smoking prevention, or they may even have the unintended e†ect of making smoking appear more rebellious, normative, and salient. The low perceived pervasiveness of both pro- and antitobacco marketing among susceptible nonsmokers is worthy of note. The susceptibility stage is the stage of smoking initiation that precedes actual experimentation. Adolescents in this stage have begun to entertain the possibility that they may try smoking in the future, but they have not yet tried smoking. Previous studies (Unger et al., 1995) have found that susceptible nonsmokers gave especially high liking ratings to tobacco advertisements, higher than the ratings given by nonsusceptible nonsmokers and current smokers. This suggests that susceptible nonsmokers may be especially vulnerable to protobacco media inÑuences. They may look to tobacco advertisements for information about the social beneÐts of smoking, and these advertisements may lead them to try smoking. However, in the Unger et al. (1995) study, recognition of tobacco advertisements was not higher among susceptible nonsmokers than among other adolescents. This suggests that although susceptible nonsmokers may respond favorably to tobacco advertisements when they see them, they


25

are not particularly likely to remember having seen them. In the current study, susceptible nonsmokers were less likely than other adolescents to recall seeing both pro- and antitobacco advertisements. It is not clear why susceptible nonsmokers would be particularly unlikely to recall tobacco-related advertising. Perhaps their uncertainty about whether to try smoking generalizes to uncertainty about tobaccorelated constructs in general, including the amount of tobacco-related advertising they recall. Further studies are necessary to understand the cognitive processes associated with susceptibility to smoking. Of all the measures examined in this study, the receptivity measures were associated most strongly with smoking status, and they were the only measures to show a monotonically increasing relationship. One explanation for this is that the receptivity measures involve cognitive and a†ective judgments about tobacco advertising, in addition to mere exposure and comprehension. The other measures assessed in this study measure merely whether adolescents recalled and comprehended tobaccorelated ads. The receptivity measures include more cognitive and emotional involvement (e.g., selecting a favorite ad, deciding to use tobacco promotional items). According to McGuire (1985), this indicates a more advanced level of cognitive processing. Therefore, it seems reasonable that these measures would be more strongly correlated with smoking status. The Ðnding that receptivity measures were most strongly associated with smoking status suggests that receptivity may be the most critical factor to examine in future studies exploring the impact of cigarette advertising on youth smoking behavior.

Implications and Recommendations

The results of this study suggest several strategies for developing health communication messages to prevent adolescent smoking. First, because susceptible nonsmokers represent a high-risk group that may not be reached e†ectively by existing antitobacco messages, perhaps messages could be developed to target this group directly. Susceptible nonsmokers are adolescents who have not yet tried smoking but are considering experimenting in the future if presented with the opportunity. These adolescents may believe that experimenting with cigarettes is not harmful as long as they do not become addicted, or they may have an intense curiosity about the physical sensations involved in smoking. Perhaps health communication messages could be developed that emphasize the immediate and long-term e†ects of smoking only a few cigarettes, as well as the fact that most people do not initially Ðnd smoking to be a pleasurable experience. In addition, because receptivity to tobacco marketing is associated so strongly with smoking behavior, perhaps media literacy education programs could be developed to inoculate adolescents against becoming receptive to tobacco marketing. For example, adolescents could be taught about the ways that speciÐc cigarette marketing campaigns are designed to sell speciÐc cigarette brands to speciÐc demographic groups. This would inform adolescents that their choice of a favorite brand is largely the result of successful manipulation and advertising tactics of the tobacco industry. Adolescents also could be reminded that they become free advertising vehicles for the tobacco industry when they wear, use, or display tobacco promotional items. If adolescents become aware of the ways that the tobacco industry manipulates and uses them, they may become more resistant to these marketing strategies and therefore less receptive to tobacco marketing.

26


Because tobacco-related marketing exposure involves multiple dimensions, the most persuasive health communication messages may be those that involve multiple dimensions. For example, the tobacco industry may attempt to create messages that are noticeable, memorable, ubiquitous, persuasive, and emotional. Antitobacco health communication messages also may be most e†ective if they have all or most of these qualities.

Limitations

The design of this study creates several limitations. These results are based on crosssectional data, so causal relationships cannot be determined. It is not clear whether receptivity to tobacco-related marketing causes adolescents to progress from nonsusceptible never smokers to established smokers. It is possible that the causal sequence of events operates in the opposite direction. Adolescents who are beginning to experiment with smoking may begin to pay more attention to tobaccorelated marketing, or they may recall tobacco-related marketing more strongly because tobacco advertisements elicit cognitive associations with the reinforcing e†ects of smoking. Longitudinal studies of tobago-related marketing exposure and smoking initiation are necessary to determine the direction of causality. Until longitudinal studies have clariÐed the direction of causality, the results of this study should be interpreted with caution. These results do not indicate that tobaccorelated marketing causes adolescent smoking. This study found statistically signiÐcant associations in a very large sample. However, most of the e†ect sizes in this study were small to moderate. Many other factors contribute to adolescent smoking, including peer inÑuences, personality factors, biological factors, and environmental factors. Compared with these other powerful inÑuences, exposure to tobacco-related marketing accounts for a small proportion of the variance in adolescent smoking. Therefore, a large sample is necessary to detect these small e†ects. Nevertheless, although these e†ects are small compared with the magnitude of other inÑuences, these results indicate that exposure to tobacco-related marketing does account for a small proportion of the variance in adolescent smoking. Because smoking has such serious health consequences, it is important not to dismiss variables found to be related to adolescent smoking, even if their contribution to the variance in smoking is relatively small. Altering even the more subtle inÑuences on smoking behavior could be instrumental in preventing tobacco-related morbidity and mortality. Some of the scales used in this study had borderline internal consistency reliability, indicating that they may represent two or more distinct constructs. Further research is needed to determine exactly the best way to word and combine questionnaire items to obtain the most valid and reliable measures of tobacco-related marketing exposure. These results are based on adolescentsÏ self-reports of their marketing exposure, smoking status, and other psychosocial variables. Respondents may have biased their responses to avoid punishment or to present a favorable impression to the experimenters. However, because all surveys were anonymous, it is unlikely that the respondents would have been motivated to alter their responses signiÐcantly. The positive correlations among the factors may reÑect an underlying tendency to pay attention to media. For example, adolescents who read billboards more frequently will be more likely to see both protobacco and antitobacco billboards, more


27

likely to recall antitobacco ads on billboards, and possibly more likely to have a favorite cigarette ad or brand. Unfortunately, the Independent Evaluation did not include questions about the respondentsÏ general use of media (e.g., frequency of watching television, frequency of reading magazines, frequency of looking at billboards, attendance at sporting events, etc.), so we were unable to control for this possible confound. The Independent Evaluation was designed to measure California residentsÏ exposure to a statewide tobacco control program, including a statewide media campaign. The messages developed for the media campaign were targeted to the audiences most at risk for smoking. Therefore, some di†erences in exposure to antitobacco media may be because of the fact that the California antitobacco media messages were intended for speciÐc groups. If a di†erent antitobacco media campaign had been evaluated, a di†erent pattern of results may have been obtained. Conclusions

Despite these limitations, the results of this study provide important new information about measurement of tobacco-related marketing exposure among adolescents. These results suggest that tobacco-related marketing exposure is a multidimensional construct that involves multiple cognitive processes, including recognition, recall, comprehension, and attitude change. Studies of the e†ects of tobacco-related marketing may yield more useful information if they include multiple measures of this complex construct. This study also suggests that susceptible nonsmokers may not be receiving the full beneÐt of antitobacco media messages. These are the adolescents who are at highest risk of beginning to smoke in the near future. Because media is an e†ective tool for reaching large numbers of people, especially adolescents at risk for smoking, continuing research is needed to develop the most e†ective antitobacco media campaigns and to deliver them e†ectively to adolescents. If e†ective antitobacco media campaigns can be developed to counter the e†ects of protobacco marketing among adolescents, these media campaigns may contribute to the prevention of smoking and tobacco-related disease among todayÏs adolescents. References Aitken, P. P., Eadie, D. R., Leathar, D. S., McNeill, R. E. J., & Scott, A. C. (1988). Television advertisements for alcoholic drinks do reinforce under-age drinking. British Journal of Addiction , 83 , 1399 È1419. Aitken, P. P., Leathar, D. S., & Scott, A. C. (1988). Ten-to-sixteen year oldsÏ perceptions of advertisements for alcoholic drinks. Alcohol and Alcoholism, 23, 491È500. Altman, D. G., Levine, D. W., Coeytaux, R., Slade, J., & Ja†e, R. (1996). Tobacco promotion and susceptibility to tobacco use among adolescents aged 12 through 17 years in a nationally representative sample. American Journal of Public Health , 86, 1590 È 1593. Americans for NonsmokersÏ Rights. (1998, September). Tobacco Industry ““PreventionÏÏ Programs. Position paper. Ammerman, S. D., & Nolden, M. (1995). Neighborhood-based tobacco advertising targeting adolescents. W estern Journal of Medicine, 162 (6), 514 È518. Arnett, J. J., & Terhanian, G. (1998). AdolescentsÏ responses to cigarette advertisements : Links between exposure, liking and the appeal of smoking. T obacco Control , 7, 129 È133. Bauman, K. E., La Prelle, J., Brown, J. D., Koch, G., & Padgett, B. (1991). The inÑuence of three mass media campaigns on variables related to adolescent cigarette smoking : Results of a Ðeld experiment. American Journal of Public Health, 81 , 597È 604.

28


Bechtel, R., Achelpohl, C., & Akers. (1971). Correlations between observed behavior and questionnaire responses on television viewing. In E. Rubstein, G. Comstock, & J. Murray (Eds). T elevision and Social Behavior : V ol. 4. T elevision in Day -to -Day L ife : Patterns of Use. Washington, DC : U.S. Government Printing Office. Borstein, R. F. (1989). Exposure and a†ect : Overview and meta-analysis of research, 1968È 1987. Psychological Bulletin, 106 , 265 È 289. Botvin, G. J., Goldberg, C. J., Botvin, E. M., & Dusenbury, L. (1993). Smoking behavior of adolescents exposed to cigarette advertising. Public Health Reports, 108 , 217È 224. Brooks, P. H., Gaines, L. S., Mueller, R., & Jenkins, S. (1998). ChildrenÏs television watching and their fathersÏ drinking practices. Addiction Research, 6, 27È 34. Clarke, A. T., & Kutz-Costes, B. (1997). Television viewing, educational quality of the home environment, and school readiness. Journal of Educational Research, 90 , 279 È 285. DiFranza, J. R., Richard, J. W., Paulman, P. M., Wolf-Gillespie, N., Fletcher, C., Ja†e, R. D., & Dittmar, M. L. (1994). Relations among depression, gender, and television viewing of college students. Journal of Social Behavior & Personality, 9, 317È328. Evans, N., Farkas, A., Gilpin, E., Berry, C., & Pierce, J. P. (1995). InÑuence of tobacco marketing and exposure to smokers on adolescent susceptibility to smoking. Journal of the National Cancer Institute , 87 , 1538 È1545. Feighery, E., Borzekowski, D. L. G., Schooler, C., & Flora, J. (1998). Seeing, wanting, owning : The relationship between receptivity to tobacco marketing and smoking susceptibility in young people. T obacco Control , 7 , 123 È 128. Hafstad, A., & Aaro, L. E. (1997). Activating interpersonal inÑuence through provocative appeals : Evaluation of a mass media-based antismoking campaign targeting adolescents. Health Communication, 9 , 253 È 272. Hafstad, A., Aaro, L. E., Engeland, A., Anderson, A., Langmark, F., & Stray-Pedersen, B. (1997). Provocative appeals in antismoking mass media campaigns targeting adolescents : The accumulative e†ect of multiple exposures. Health Education Research, 12, 227È236. Hafstad, A., Aaro, L. E., & Langmark, F. (1996). Evaluation of an antismoking mass media campaign targeting adolescents : The role of a†ective responses and interpersonal communication. Health Education Research, 11 , 29 È 38. Independent Evaluation Consortium. (1998). Final Report of the Independent Evaluation of the California T obacco Control Prevention and Education Program : W ave 1 Data , 1996È 1997 . Rockville, MD : The Gallup Organization. Johnston, L. D., OÏMalley, P. M., & Bachman, J. G. (1998). National survey results on drug use from the Monitoring the Future study , 1975È1997 . V olume I : Secondary school students . (NIH Publication No. 98-4345). Rockville, MD : National Institute on Drug Abuse. Kim, J., & Mueller, C. W. (1978). Introduction to Factor Analysis . Newbury Park, CA : Sage. Koolstra, C. M., van der Voort, T. H. A., & van der Kamp, L. J. (1997). TelevisionÏs impact on childrenÏs reading comprehension and decoding skills : A 3-year panel study. Reading Research Quarterly , 32 , 128È 152. McGuire, W. J. (1985). Attitudes and attitude change. In G. Lindzey & E. Aronson (Eds.), Handbook of Social Psychology , V ol. II pp. 233 È 346 New York : Random House. Murray, D. M., Prokhorov, A. V., & Harty, K. C. (1994). E†ects of statewide smoking campaign on mass media messages and smoking beliefs. Preventive Medicine, 23 , 54 È 60. Peter, J. P., & Olson, J. C. (1990). Consumer Behavior and Marketing Strategy , 2nd ed . Boston Irwin. Pierce, J. P., Choi, W. S., Gilpin, E. A., Farkas, A. J., & Berry, C. C. (1998). Tobacco industry promotion of cigarettes and adolescent smoking. JAMA , 297 , 511È515. Pierce, J. P. Evans, N., Farkas, A. J. Cavin, S. W., Berry, C., Kramer, M., Kealey, S., Rosbrook, B., Chen, W., & Kaplan, R. M. (1994). T obacco Use in California. An Evaluation of the T obacco Control Program 1990È1993 . La Jolla, CA : University of California, San Diego.


29

Pierce, J. P., & Gilpin, E. A. (1995). A historical analysis of tobacco marketing and the uptake of smoking by youth in the United States : 1890-1977. Health Psychology , 14, 500 È508. Pierce, J. P. Gilpin, E., Burns, D. M., Whalen, E., Rosbrook, B., Shopland, D., & Johnson, M. (1991). Does tobacco advertising target young people to start smoking ? Evidence from California. JAMA , 266 , 3154 È 3158. Pierce, J. P., Gilpin, E. A., Emery, S. L. Farkas, A. J., Zhu, S. H., Choi, W. S., Berry, C. C. Distefan, J. M., White, M. M., Soroko, S., & Navarro, A. (1998). T obacco Control in California : W hoÏs W inning the W ar ? An Evaluation of the T obacco Control Program 1989È 1996 . La Jolla, CA : University of California, San Diego. Popham, W. J., Potter, L. D., Hetrick, M. A., & Muthen, L. K. (1991). Evaluation the California 1990È 91 tobacco education media campaign : Executive report number 4 wave four survey results. Report submitted to California Department of Health Services. Schooler, C., Feighery, E., & Flora, J. A. (1996). Seventh gradersÏ self-reported exposure to cigarette marketing and its relationship to their smoking behavior. American Journal of Public Health, 86 , 1216 È 1221. Singer, J. D. (1998). Using SAS PROC MIXED to Ðt multilevel models, hierarchical models, and individual growth models. Journal of Educational and Behavioral Statistics, 24 , 323 È 355. Slater, M. D., Rouner, D., Murphy, K., Beauvais, F., Van Leuven, J., & DomenechRodriquez, M. M. (1996). Male adolescentsÏ reactions to TV beer commercials : The e†ects of sports content and programming context. Journal of Alcohol Studies , 57 , 425 È 433. Unger, J. B., Johnson, C. A., & Rohrbach, L. A. (1995). Recognition and liking of tobacco and alcohol advertisements among adolescents : Relationships with susceptibility to substance use. Preventive Medicine, 24 , 461È 466. U.S. Department of Health and Human Services. (1994). Preventing tobacco use among young people : A report of the Surgeon General . Atlanta, GA : U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health. Worden, J. K., Flynn, B. S., Solomon, L. J., Seckerwalker, R. H., Badger, G. J., & Carpenter, J. H. (1996). Using mass media to prevent cigarette smoking among adolescent girls. Health Education Quarterly, 23 , 453 È 463. Wyllie, A., Zhang, J. F., & Casswell, S. (1998a). Responses to televised alcohol advertisements associated with drinking behavior of 10 È17 year olds. Addiction , 93 , 361È 371. Wyllie, A., Zhang, J. F., & Casswell, S. (1998b). Positive responses to televised beer alcohol advertisements associated with drinking and problems reported by 18 to 29-year olds. Addiction , 93 , 749 È 760. Zajonc, R. (1968). Attitudinal e†ects of mere exposure. Journal of Personality & Social Psychology Monographs , 9, 1È 27.