One of the basic principles of communication planning is to understand audiences better; doing so makes it possible to develop messages most likely to resonate with target audiences and reach them more frequently. 1,2 Audience segmentation is one of the core tools that enables communication and marketing planners to better understand their audiences.3,4 One objective of audience segmentation is to identify the communication channels through which an audience receives information.5,6 Knowledge of communication channels is especially important given the resource constraints typical in public health communication and social marketing efforts and the increasing fragmentation of media channels.3,7 Investing time and resources in delivering messages through channels inappropriate or inefficient for reaching intended audiences is economically wasteful and unlikely to have the intended effect.1,2,8
Surprisingly, to our knowledge no published studies have addressed the communication channels through which cigarette smokers regularly receive information, and whether channels used by smokers differ from those used by nonsmokers. Furthermore, it is not known if communication channels differ among subgroups of smokers (e.g., by demographic characteristics). If smokers pay attention to distinctive communication channels, mass media efforts designed to increase the use of cessation services, such as telephone quit lines can be better planned to target those distinctive channels.
From a communication or marketing planning perspective, it is more important to know the size of the intended audience that could be exposed to messages (reach) than the extent of smoking among specific channel users (prevalence). For example, if a television show had a smoking prevalence among viewers of 15% but attracted 40% of all smokers (reach), the potential effect of a cessation message would be fairly large compared with a show for which 50% of viewers were smokers (prevalence) but who constituted only 1% of all smokers (reach). The cost for placing mass media advertisements, as well as the likelihood that a media company would permit the placement of an unpaid message such as a public service announcement, is directly related to reach: mass communication companies charge higher advertising rates to reach audiences with larger numbers of people.9 The cost for running a 30-second advertisement during the Super Bowl can exceed $2 million because of the overall large audience size10 ; however, this is not likely to be the most cost-efficient approach to expose smokers to cessation ads, especially given the limited budgets for many tobacco control programs.
We used a nationally representative survey to examine measures of mass media channel use among respondents who smoke and respondents who don't smoke cigarettes. The mass media channels included preferred television genres, networks, and programs; preferred radio genres; newspaper use; and magazines read regularly. Additionally, we compared mass media exposure measures by smoking status and stratified by age, gender, and race/ethnicity.
METHODS
Data were obtained from ConsumerStyles and HealthStyles surveys, which were conducted in 2002 and 2003 by the public relations firm Porter Novelli (survey details are provided elsewhere).1,11 Briefly, the Con-sumerStyles survey collects data on many measures including the use of media, consumer products and services, and personal interests. The HealthStyles survey, completed by ConsumerStyles survey respondents, obtains detailed information about a variety of health issues, including smoking.
The surveys were conducted by mail and were designed to be nationally representative. From a panel of 600000 households previously recruited to participate in consumer marketing surveys, a stratified random sample of 10000 respondents aged 18 years and older receive the ConsumerStyles survey each year.12,13 Supplemental efforts were used to ensure adequate representation from low-income and minority populations. The response rate to the ConsumerStyles survey was 61% in 2002 (N=6065) and 59% in 2003 (N=5873). The response rate to the HealthStyles survey was 73% in 2002 (N=4397) and 69% in 2003 (N=4035). For this study, we pooled 2002 and 2003 data to increase sample sizes for subgroup analyses, and respondents who had missing or unknown data on current smoking status were eliminated (n=278). The data set will hereafter be referred to as the Consumer/ HealthStyles survey.
One question was used to assess smoking status: "On average, on the days that you smoke, how many cigarettes a day do you smoke?" Current smokers were those who reported smoking at least 1 cigarette per day (respondents who were not current smokers were instructed to enter "0"). Questions about mass media use included television, radio, newspapers, and magazines. Each of these broad communication channels contained general questions about the extent to which respondents use them. For television and radio, questions addressed the average number of hours per day and usual times of the day that respondents used these media. The question about newspaper use asked about the average number of days per week that respondents read newspapers. For magazines, respondents were asked about specific magazines they read regularly, defined as "almost every issue."
Respondents were also asked more specific questions for each mass media category. These included use in both years of each of 17 television genres (e.g., news, soap opera) and 46 television networks (e.g., NBC, The Weather Channel) at least once per week, and 45 specific television shows they watched regularly in 2003 only. Respondents also indicated in both survey years which of 16 radio genres (e.g., classical, talk or call-in radio, hard rock) they routinely listened to, and which of 57 magazines they routinely read.
Definitions of Measures
Reach is defined here as the proportion of all smokers reached by a specific communication channel. Efficiency index scores are a standard approach used in the mass media industry to determine the cost efficiency of advertising expenditures.14 Efficiency index scores are calculated by dividing the reach estimate for a specific target population (e.g., female cigarette smokers) by the reach estimate for the entire population group (e.g., all women) and multiplying the ratio by 100. Index scores higher than 100 indicate better efficiency because proportionately more smokers are potentially exposed; conversely, index scores below 100 represent lower efficiency because proportionately fewer smokers are exposed. For example, if the reach of a television program among all Hispanics was 15%, but the reach among Hispanic smokers was 20%, the index score=(20/15)×100, or 133.3. By definition, if a greater proportion of users of a particular media have a higher smoking prevalence, the corresponding index score would exceed 100, and vice versa.
The index score provides an opportunity to rank communication measures within a specific mass media communication channel, as well as to make standardized comparisons across mass media channels to determine the most cost-efficient communication channels on the basis of available resources. Decisions about where to run advertisements are generally based both on reach estimates and index scores. However, the value of index scores and reach data are not limited to paid advertising. These data can also be used for planning earned media (free media exposure) or entertainment-education activities, such as targeting reporters at certain news media outlets, or entertainment industry gatekeepers, who potentially could reach a large number of smokers.
Statistical Analyses
All analyses were conducted using SAS software version 8.2 (SAS Institute Inc, Cary, NC). Smoking prevalence for general communication measures was calculated for all respondents by demographic subgroups. Demographic groups were age (18-34, 35-54, =55 years), gender, and race/ethnicity (White, Black, or Hispanic). To determine differences in smoking prevalence for index score calculations (e.g., whether they were significantly higher or lower than 100) we assessed statistical significance using 2-tailed P values derived from the χ2 or Fisher exact test. Because of the large number of analyses and the size of the overall pooled (2-year) sample, we used a conservative level (a<.01) to determine significance.
Because of the large amount of data, we included only the 5 highest index scores on the basis of media measures that had the highest reach among the population for television networks, television programs, radio genres, and magazines. We limited the index scores and reach data on television networks to cable channels, because cable television networks have lower advertising costs than the broadcast networks (i.e., ABC, CBS, and NBC) and because the index scores were close to 100 for all the broadcast networks. (Complete data for all media reach estimates and index scores are available from D.N.)
RESULTS
A total of 11660 respondents were included in this study. There were more female than male respondents (57% vs 43%), and slightly more than 50% were aged 35-54 years. Nearly 70% of respondents were White, 13% were Black, 12% were Hispanic, and 5% were of some other race/ethnicity. There were 2532 respondents who were current smokers (21.7%).
More than 79% of smokers watched television for at least 2 hours per day, which was a higher percentage than among nonsmokers (Table 1). Even during times of the day when television viewing was lower (e.g., from 10 AM to 2 PM or from 11 PM to 6 AM), more than 30% of smokers reported regular viewing. Reach was higher for smokers than for nonsmokers for the television genres of weather, made-for-television movies, talk shows, game shows, reality television, soap operas, music channels, and local community programs.
An estimated 42% of smokers listened to the radio for at least 2 hours per day, which was a higher percentage than for nonsmokers (Table 1). Reach estimates for listening during the evening, late-night, and early morning hours were greater among cigarette smokers than among nonsmokers. Of note was that the magnitude of radio's reach among smokers, which ranged from 45.9% to 69.5% between 6 AM and 6 PM, exceeded the reach of television during these hours.
Although an estimated 39% of smokers read the newspaper at least 5 days per week, smokers were more likely than nonsmokers to not read the paper at all or read it less than 5 days per week. Analyses stratified by demographics demonstrated particularly large differences in newspaper readership by age: only 21% of smokers aged 18-34 years read newspapers at least 5 days per week, compared with 58% of smokers aged 55 years or older. More than 28% of smokers reported not reading any of the magazines included in the surveys, compared with nearly 23% of the total population.
The cable television networks that had the 5 highest reach estimates for smokers and their respective index scores are listed in Table 2. With the exception of people aged 55 years or older, several networks with the highest reach estimates for smokers (USA, The Discovery Channel, Lifetime, TNT, and TBS) had index scores significantly greater than 100. The television programs CSI, Law & Order, Everybody Loves Raymond, and ER generally had the highest reach estimates overall and among most subgroups of smokers (Table 2). Compared with the overall adult population, significantly higher index scores for smokers were seen for Law & Order overall and among several subpopulations; statistically significant index scores exceeding 125 were seen for The Simpsons, King of the Hill, and That '70s Show among respondents aged 18-34 years, and for The Parkers among Black respondents.
Radio genre data for reach among smokers and for index scores are listed in Table 3. The magnitude of many reach estimates for radio was comparable to that for television networks and programs. Except for Black respondents and adults aged 18-34 years, there was little variation overall or by subgroups for genres with the highest reach estimates among smokers; oldies, classic rock, country, and top 40 generally dominated. Classic rock and country genres had the highest index scores for smokers both overall and among most subgroups.
Magazine reach estimates, with the exception of those for people aged 55 years or older and Black respondents, were much lower than for television and radio. TV Guide was the only magazine that was consistently in the top 5 reach estimates for smokers in most groups and consistently had index scores higher than 100. Playboy among men, Playboy and Glamour for adults aged 18-34 years, and Jet and Reader's Digest for Black respondents, were the only other magazines that had significantly elevated index scores. Several magazines that had the highest reach estimates do not accept advertising, which demonstrates the need to attempt reaching readers through other means.
DISCUSSION
We demonstrated that differences exist between smokers and nonsmokers for most mass media exposure measures. Smokers were more likely to be more frequent users of television and radio than were nonsmokers but less likely to be magazine or newspaper readers. Almost one third of smokers were daytime or late-night television viewers, although reach was lower than at other times of the day. Among the most popular cable television networks, television programs, radio genres, and magazines, certain ones had relatively high reach estimates for smokers and elevated index scores. Examples included USA, Lifetime, and The Discovery Channel cable television networks; oldies, classic rock, and country radio genres; and TV Guide magazine. Radio, in particular, could be a valuable medium for promoting smoking cessation messages to a broader audience, and the cost to produce and air ads is relatively low.15,16
There are at least 3 likely explanations for the observed differences in mass media channel use measures between smokers and nonsmokers. The first explanation is socioeconomic status. Smoking is more common among people of lower socioeconomic status, 17,18 who tend to be more frequent users of television and radio (and less likely to read newspapers or magazines) than are nonsmokers. 15,18,19 The second explanation is that there are differences in mass media use among certain population groups. Newspaper and magazine readers (especially newspaper readers) tend to be older20 ; because smoking prevalence declines with increasing age, this would help explain lower smoking prevalence among users of these mass media channels.17 Younger adults tend to be heavier radio users than older adults,20,21 and Blacks tend to be heavier users of television than other subgroups. 19 Of note was that Hispanics and Whites had similar patterns of mass media use and cigarette smoking. The third likely explanation is that most media organizations develop content for specific audience segments (e.g., young adults or women)22 and smoking prevalence varies among population groups.17
The information from this study could be especially valuable for media planning by tobacco control programs or advocacy organizations. There is strong evidence that mass media campaigns are effective at increasing smoking cessation,23 because paid advertising or public service announcement messages can increase the use of cessation services, such as telephone quit lines or pharmacotherapies (e.g., nicotine replacement products). By using demographics and smoking-status data, coupled with media measures of exposure, distinct messages can reach target smoking population groups more effectively and efficiently.
Cost is especially important in the current fiscal climate because of declines in funding for tobacco control programs and media efforts at both national and state levels.24 Organizations usually have limited budgets for tobacco control programs. Therefore, it is important to maximize the budget and reach the largest number of smokers at the lowest possible cost. Paid anti-tobacco ads that are placed in mass media are often the first to be cut by programs because of the expense.25 Public health program staff can use reach estimates among smokers, in conjunction with higher index scores, to better allocate media dollars to reach the largest number of smokers with the greatest cost efficiency. Paid advertisement rates for broadcast television networks and radio, for instance, are lower than advertising rates for broadcast networks.15,16
To provide some specific examples, program planners could use these data to maximize exposure of adult smokers who are Black by placing paid smoking cessation messages on USA, TNT, or TBS television net-works; radio stations that play rhythm and blues or oldies genres; or in Jet magazine. Each of these means of communication ranked within the top 5 for reach estimates among Black smokers and had index scores significantly higher than 100. Similarly, to more efficiently reach smokers aged 18-34 years, planners should consider placing smoking cessation ads on The Discovery Channel, Lifetime, or USA cable networks; The Simpsons or King of the Hill television programs; or radio stations that play classic rock, alternative/progressive rock, or hard rock music.
These findings can be used in earned media or entertainment-education efforts. For example, media gatekeepers or creative artists could be encouraged to include messages about cessation in daytime television shows (e.g., local or national news, talk shows, or soap operas). Smoking cessation story lines could be developed for specific television shows that are most likely to reach the larger audiences of smokers.26,27 Although the use of public service announcements has been criticized because they often run at times of the day when there are smaller audiences (e.g., late at night),3 our findings suggest that late-night and daytime viewing hours attract a disproportionate share of viewers who smoke; thus, public service messages airing during these times may still be of value.
Our study has several limitations. We used a national mail survey that relied on self-reports of tobacco use and mass media use without independent verification. Previous research suggests that self-reports of cigarette smoking slightly underestimate actual use on the basis of biological measures.28 The question used to estimate smoking prevalence differed from questions that are used in major national health surveys; however, our estimate differed by less than 1 percentage point from the national prevalence estimates of adult smoking from the 2002 and 2003 National Health Interview Surveys.17,18 Previous research has shown that national prevalence estimates for smoking, obesity, and other health measures from ConsumerStyles and HealthStyles surveys are comparable to those obtained from the Centers for Disease Control and Prevention's Behavioral Risk Factor Surveillance System.11
Our self-reported estimates of reach among the general population for television, radio, and magazine measures are consistent with Nielsen,29 Arbitron,21 and magazine circulation data.30 Several of the television shows included in the 2003 Consumer Styles and HealthStyles surveys (e.g., Friends) are no longer produced. We analyzed a large amount of data; thus, despite using P < .01 as the cutoff level for our comparisons, some of our findings of statistical significance could be based on chance alone. Last, smoking cessation messages need not be targeted solely toward smokers, because family members and friends can encourage smokers to quit or support them during attempts to quit.31
Our study used only top-line data that would be most helpful for planning to deliver smoking cessation messages through specific mass media communication channels, which is why we emphasized reach measures. However, this study merely scratched the surface of the potential uses for this data set. More extensive analytic research could be done on the independent correlates or predictors of communication measures with smoking prevalence or intention to quit. In addition, cluster analyses or similar approaches could be used to more fully develop communication-related taxonomies for smokers on the basis of the full range of measures included in this study, and similar to what has been done previously with psychographic data.1,2
Finally, the approach we used in this study is similar to that used by private organizations and businesses to better communicate messages to their target audiences. We believe that this methodology could be readily applied to examine mass media communication measures and improve planning and resource allocation for other public health issues, such as diabetes, cancer screening, and alcohol use.
(C) 2008 American Journal of Public Health. via ProQuest Information and Learning Company; All Rights Reserved