The Relationship Between Editorial and Advertising Content about Tobacco and Alcohol in United States Newspapers

Abstract

Using a nationally representative sample, this study examined the relationship between amount of alcohol and tobacco advertising and related news-editorial content. This study found less tobacco and alcohol advertising in newspapers than did previous research and no relationship between coverage and number of advertisements.

Health threats from alcohol and tobacco use are a global concern among policy makers, medical professionals and researchers. One noted tobacco researcher called smoking’s devastation the paramount behavioral cause of death designed by humans, killing an estimated 5 million people per year. ¹ Furthermore, alcohol abuse was estimated as of 1998 to cost the United States $185 billion per year.²

Investigating how people learn about alcohol and tobacco and their reasons for using, abusing and ceasing to use these substances is necessary before solutions to their associated health problems become viable. There is increasing evidence that tobacco and alcohol advertising is one of the factors influencing cognitions about these products as well as influencing uptake and consumption.³ The present study provides the most authoritative descriptive assessment to date of the extent of such advertising in newspapers, using a nationally representative sample of U.S. newspapers, takes into account newspaper size and regional differences, and investigates whether such advertising influences editorial coverage in these newspapers.

There is growing evidence that advertising of alcohol and tobacco may influence attitudes and behavior, especially among youth, as discussed below. Because alcohol and tobacco are controlled substances, frequently marketed to youth, for whom their consumption is illegal, they are commonly associated with adulthood, which makes them desirable for underaged consumers.⁴ Past research, albeit spotty, has documented substantial alcohol and tobacco advertising in media, possibly lending to the legitimization of their use among adults. Policy makers have argued that consumption of alcohol and tobacco incurs tremendous social costs, leading to bans in terms of advertising and placement and marketing restrictions where youth are targeted.⁵ With increased public policy concern over the dangers associated with consuming these controlled substances, associated changing advertising trends have resulted in fewer alcohol and tobacco newspaper advertisements at the same time coverage of their corresponding health dangers may be on the increase. This beckons the question of the relationship between the commercial and the reportorial sides of these products in the media. Assessing the extent of such advertising in newspapers, then, has obvious relevance from both public policy and public health perspectives.

The question of tobacco and alcohol advertising in newspapers has particular importance to journalists. Work done in the 1990s provided strong evidence that, at least in the case of U.S. magazines, the amount of news and feature content devoted to tobacco risks, smoking cessation, and other such topics was negatively related to the amount of tobacco advertising in the magazine.⁶ Given the central importance of the separation of advertising influence from editorial decisions in the ethos of American journalism, an assessment of such influence in contemporary newspapers is perhaps overdue. Newspaper reporting on controlled substances has been ignored in media research.⁷ This medium, which has historically remained relatively immune from regulation, given First Amendment protection, is important to study. From the standpoint of political economic theory, newspapers have historically kept advertising and news-editorial functions separated as an institutional practice, yet content and audiences are becoming increasingly commoditized.⁸ According to political-economic theories, a relationship exists between the way media are structured economically and the ideology of media content. In media consolidation, a major factor is avoidance of economic risks. Politically, citizens in a democracy must have access to free information in order to participate actively, exercising their democratic rights to vote and maintain or improve their well being. Frequently in the history of U.S. regulation against advertising of controlled substances, the issue of how harmful advertising is for these products has been juxtaposed with the availability of information for media audiences to make sound decisions about using these products. Some research focus has occurred with regard to effects of advertising on children. However, it would seem essential to document whether newspapers’ commercial side is influencing its editorial side.

Little research has analyzed the relationship between newspaper content and advertising. Further, economic studies of alcohol and tobacco advertising’s effects on consumption suggest that focusing on national data, which is most often studied, blurs findings that may be evident locally. Thus, investigating newspaper advertising as it impacts editorial content locally would be useful.⁹

The nature of tobacco and alcohol advertising

Twenty years ago, tobacco was the most widely advertised product in American newspapers.¹⁰ From 1970 to 1997 the tobacco industry increased spending on advertising by about 16 times, to $5.7 billion.¹¹ The Oral Cancer Foundation claimed the tobacco industry was the second largest advertiser in print media, including newspapers and magazines, and the largest billboard advertiser.

Although the Master Settlement Agreement in 1998 restricted advertising and promotion, there has been a 66% increase in overall tobacco advertising and promotion dollars since then. Some of the largest increases have occurred in just the past few years.¹² For example, in 2001, U.S. domestic tobacco advertising and promotion costs were $11.2 billion, up 16.9% from 2000.¹³ Despite these increases, newspaper tobacco advertising spending dropped $11.3 million, a decrease of nearly two-thirds in the 1990s.¹⁴ In 2001, newspapers accounted for only 0.3% of the $11.2 billion spent on tobacco ads.¹⁵ This decline continued, and from 2002 to 2005, newspaper advertising expenditures decreased from $25,538,000 (of a total for all advertising of almost $12½ billion) to $1,589,000 (of a total for all advertising of a little over $13 billion).¹⁶

There is evidence that tobacco marketing influences young people..¹⁷ A link has been established between cigarette advertising and smoking among youth,¹⁸ and print advertising has been found to play a significant role in getting adolescents to start smoking.¹⁹ After the legal ban against tobacco advertising on U.S. radio and television in 1971, incidence of taking up smoking by young adults did not begin to show a decline until later in the 1970s, when public campaigns strongly discouraged smoking.

As with tobacco research, alcohol advertising and its relationship to alcohol use, abuse, cessation and other behaviors is equally complex. For alcohol, more so than tobacco, it is difficult to determine advertising expenditures in various media. In fact, a coalition of U.S. alcohol reform and medical organizations has called for a more specific breakdown on alcohol promotion expenditures to be reported by the Federal Trade Commission.²⁰ Nonetheless, some data are available. Alcohol advertising expenditures reported by TNS Media intelligence, a media technology tracking company, showed a decrease in newspaper placement from $29,206,200.00 in 2005 (2.5% of total media spending) to $8,252,300.00 in 2006 (0.8% of total media spending).²¹ Total expenditures for alcohol advertising reported by this group were just a little over $1 billion in 2006. Another report showed almost $4 billion was spent for total alcohol advertising and promotion in 2001, with total advertising expenditures on the increase.²²

Young people are often the target for such advertising.²³ For example, researchers have found that the alcohol advertising in college newspapers surpassed advertising for all other products.²⁴ Furthermore, Martin et al.²⁵ concluded, from a review of the literature and independent research, that young people who have more exposure to alcohol advertising remembered the ads and liked alcohol more than those less exposed, associated drinking with positive outcomes, believed their peers used alcohol at rates higher than they actually did, drank more and intended to drink more.

The alcohol industry has undertaken voluntary self-regulations regarding advertising and marketing to adolescents,²⁶ although advertising for distilled spirits has recently been on the increase.²⁷ The American Medical Association has called for the alcohol industry to discontinue advertisements targeted toward youth, to support warnings against the use of alcohol, and to curb television entertainment content that portrays alcohol as exciting and fun and that does not show adverse consequences of its use.²⁸

Less research has examined the impact of advertising on adult drinking, as this is a legal behavior. Tobacco and alcohol advertising research findings on the effects of consumption have been varied, with economic research on national advertising expenditures showing little effect on consumption.²⁹ However, local published advertising for these products shows some relationships with consumption.³⁰ Coverage of tobacco-related issues has potentially important health outcomes. Pierce and Gilpin,³¹ who found media coverage of smoking and health were related to levels of smoking cessation but not initiation, argued that news media can be critical in changing smoking behavior in the United States and elsewhere.

Relationship between tobacco and alcohol advertising and news-editorial newspaper content

Little research has examined this relationship, perhaps because of institutional separation in journalism of the commercial side, specifically advertising and marketing, from the reportorial side of news production. Further, academic researchers frequently focus on either news or advertising, given their professional affiliations and backgrounds.

Magazines, and women’s magazines in particular, have been a focus of research on tobacco advertising.³² Warner reported that, although some evidence suggested a relationship between editorial content and advertising, the studies conducted thus far were anecdotal or used small samples and weak statistical testing. Warner’s³³ 25-year examination of magazines found a relationship between magazines’ publishing cigarette advertising and less coverage of the risks of smoking cigarettes, with women’s magazines the most extreme in this regard. Studying 99 U.S. magazines from 1959–1986 (except 1970–1972), Warner used a single magazine representing each year. He coded articles, using titles and key words, according to smoking risks, health effects, social implications and politics of smoking. For advertising information, he used information from professional organizations and magazine indices, including total cigarette advertising expenditures and total advertising revenues for the magazines and whether the magazines accepted cigarette advertising. Overall, Warner found that magazines that published cigarette ads were significantly less likely to publish articles on smoking risks, controlling for the size of readership, type of magazine, and publishing following the Surgeon General’s first report on smoking. He reported that those magazines without cigarettes advertising showed a more than 40% likelihood of covering smoking dangers than magazines with cigarette ads. His sub-analysis of women’s magazines revealed an even stronger relationship.

Weis and Burke³⁴ investigated tobacco industry influence on editorial content in media, presenting historical evidence up until the mid 1980s of reporters and writers who got into trouble for criticizing key advertisers. They cited the Reader’s Digest July 1957 report on smoking’s health risks that led to withdrawal of advertising, a series of articles in Mother Jones that linked tobacco with various terminal illnesses, The tobacco industry canceled all its tobacco advertising from Mother Jones after publication of the articles.³⁵ Further, Newsweek had an informal policy of removing anti-tobacco text.³⁶ Ninety percent of daily newspaper editors have reported that advertisers have tried to influence story content, and in some cases keep stories from being published. This research found smaller newspapers to be the most susceptible to advertising influence. The research on tobacco advertising and content in magazines occurred in the 1980s, when about one-third of Americans smoked. Today, about 21% of Americans smoke.³⁷ Given changes in smoking and tobacco use since Warner’s analysis,³⁸ we assume coverage of smoking’s health risks have become more normative in the media.

We believe a parallel investigation of alcohol is useful, given it, too, has had controls in the media. Little research has focused on the relationship between news-editorial content and alcohol advertising. Previous content analyses of alcohol advertising have focused on television, billboards, magazines and some on college newspapers.³⁹ Breed and colleagues content-analyzed student newspapers in two studies, finding in a 1984–1985 study⁴⁰ greater local alcohol advertising than in a 1977–1978 study.⁴¹ Counting the number and size of advertisements, they noted many ads encouraged irresponsible, heavy drinking.

Hypotheses and Research Questions

Following Warner’s lead in documenting a relationship between newspaper advertising and lower newspaper coverage of the dangers of tobacco,⁴² this study looked at the situation 20 years later, with an additional focus on the relationship between newspaper alcohol content and advertising. Examining newspapers is a conservative approach to this issue, as Snyder et al.⁴³ found the lowest expenditures in conventional media to be in newspapers, compared to television, radio, magazines and outdoor advertising. If a relationship is found with newspapers, it would follow that other media might demonstrate an even more serious relationship if they were similarly examined. Going beyond Warner’s research, which linked editorial content with advertising expenditures, this study content analyzed both editorial content and the actual amount of advertising in a nationally representative sample of U.S. newspapers.

Studying newspaper content and advertising will provide useful information for policy makers and health officials. Given some evidence exists of a relationship between amount of advertising and news-editorial content relative to tobacco, we posed the following hypothesis::

H1: The amount of tobacco coverage is negatively related with the amount of tobacco display advertising, controlling for characteristics of newspaper size and geographic region.

The following hypothesis extends this research to examine the same relationship between alcohol advertising and news-editorial content in newspapers:

H2: The amount of alcohol coverage is negatively related with the amount of alcohol display advertising, controlling for newspaper size and geographic region.

In investigating the amount of reporting and advertising relative to each of these controlled substances, we assumed that national averages might obscure differences relating to greater smoking prevalence in some regions of the United States, such as the South, as the South has been found to be the highest region for tobacco use and production.⁴⁴ The Centers for Disease Control report that 45.3 million, or 20.8% of adults smoke, and the states with the highest tobacco use are Kentucky (28.6% of the population using tobacco), West Virginia (25.7%), Oklahoma (25.1%) and Mississippi (25.1%). Newspapers have been found to show economic boosterism regionally⁴⁵ and some differences have been detected across geographical areas.⁴⁶ Therefore, we asked:

RQ1: Is there a difference across U.S. regions --the Northeast, Midwest, South and West--relative to the ratio of tobacco advertisements and total pages?

The region with the highest alcohol use is the Midwest, followed by the West, the Northeast and the South, although no particular region is markedly higher than another.⁴⁷ Just as for tobacco advertising, it is possible that looking at national averages may obscure regional differences in alcohol advertising, perhaps related to differences in alcohol consumption patterns. Therefore, we asked:

RQ2: Is there a difference across U. S. regions --the Northeast, Midwest, South and West--relative to the ratio of alcohol advertisements and total pages?

More speculatively, the extent of tobacco advertising may differ by newspaper size because of differences in the nature of the advertising base for large versus smaller newspapers. Therefore, we ask:

RQ3: Is there a difference among different sizes of newspapers relative to the ratio of tobacco advertisements and total pages? Likewise, there may be differences in the amount of alcohol advertising by newspaper size because of differences in the nature of the advertising base for large versus smaller newspapers:

RQ4: Is there a difference among different sizes of newspapers relative to the ratio of alcohol advertisements and total pages?

Method

Data were from a grant funded by the National Institute on Alcoholism and Alcohol Abuse (AA10377) that studied newspaper coverage of alcohol and tobacco risks and media advertising of alcohol and tobacco. This study used a nationally representative sample of daily newspapers. Two 28-day constructed months were created, one each from 2002 and 2003. Each constructed month was balanced with respect to day of the week to address variations in daily news holes.⁴⁸ Furthermore, to address seasonal variations in media coverage, each year’s constructed month consisted of one constructed week from each season.⁴⁹

To create the sample, media outlets across the country were stratified based on market size. The designated market area (DMA^®)is a particularly useful sampling unit in this case because it provides information on market size and defines markets at the local level.

The country’s 210 DMAs were rank ordered and then divided into six strata, with each stratum consisting of approximately 1/6^th of all U.S. households. The number of markets per stratum varied from three in the top stratum to 119 in the bottom stratum to achieve approximate equivalence in population per stratum. Using six strata permitted reasonable regional representation in all strata and reasonable homogeneity of market size within each stratum.⁵⁰ On each day sampled, one DMA was randomly selected from each of the six strata for a total of six DMAs per day of the sample.

For each DMA chosen on a given sample date, three daily newspapers, representing different circulation categories, were selected. After a DMA was chosen, three daily newspapers were selected from that DMA: the largest newspaper,⁵¹ one randomly selected from above the median circulation split for that day, and one randomly selected from below the split.⁵² The resulting sample was representative of all newspapers within the sampled DMAs.⁵³ The total number of newspapers sampled was 955. This study coded news editorial stories that mentioned alcohol and/or tobacco. Also, it determined the average number of alcohol and tobacco ads per page.

Alcohol and tobacco story identification

As part of a larger study of media content, coders selected newspaper stories about the following topics: alcohol, illegal drug use, tobacco, violent crime and accidents. Stories qualified to be in the larger study if they were news and editorial content, specifically news stories, features, sidebars, editorials, letters to the editor, columns and commentaries. Sections that were not coded for the news editorial part of this study included sports box scores, entertainment reviews, calendars of events, obituaries and advertisements, including classified and personal ads.

Only the alcohol and tobacco stories are used in the current analysis. To be coded as an alcohol or tobacco item, stories had to mention the growing, manufacturing, distribution, sale or use of either of these substances in the beginning of the story (i.e., the story teasers, headline, subheadline, and first two paragraphs of the story). Cohen’s kappa and Scott’s pi reliabilities for determining the presence or absence of alcohol and tobacco in a story ranged from .85 to .94 for alcohol, and from .90 to .96 for tobacco (a range is provided because multiple pairs of coders tested coding scheme reliability). Because of the lengthy time involved in coding stories over the two-year period, we tested twice for intercoder drift; reliabilities remained good, ranging between .88 and .95 for alcohol, and between .67 and .89 for tobacco. Kappas in the .61 to .80 range indicate substantial agreement, and kappas above .80 indicate almost perfect agreement.⁵⁴ Because pi is based on the same principles as kappa, these ranges can apply to both statistics.

Alcohol and tobacco advertisement identification

Two trained coders used a coding scheme to determine the number and nature of alcohol and tobacco advertisements. For this study, we defined alcohol and tobacco ads as follows: display ads that promote the use or nonuse of alcohol and/or tobacco. Classified advertisements, advertising inserts, in-house advertisements, advertorials and advertisements for books, films, plays and recorded music were not coded, as they might only include minor reference to alcohol or tobacco. An ad that continued across pages but did not stand alone on any page was considered just one advertisement.

Coders read up to the three largest sizes of advertising type to see whether any words in the ad related to alcohol or tobacco. If the second or third largest size of type within an advertisement was the body copy, defined as two or more lines of solid text within an advertisement, then coders stopped reading at the preceding level. Sample terms used to determine whether an advertisement mentioned alcohol were: drinking, drafts, happy hour, beer, ale, whiskey, brewery, shots and alcohol. Sample terms used to establish tobacco content were any brand of tobacco, nicotine, cigarettes, menthols and smoking.

To check the reliability of the coding scheme, coders coded a randomly selected sample of newspapers that represented all six DMAs. Coders independently counted the number of alcohol advertisements, the number of tobacco advertisements and the total number of display advertisements. This yielded Pearson product moment correlations coefficients of .93 for coder agreement on the number of display advertisements per page and .92 for coder agreement on the number of alcohol and tobacco display advertisements per page.

Results

A total of 955 newspapers were coded for this study, with 97 tobacco ads and 837 alcohol ads analyzed. The amount of tobacco advertising in newspapers in the post-MSA era results showed that 88 newspapers contained tobacco display advertisements and 867 (90.8%) newspapers did not contain tobacco ads. Of the newspapers with tobacco ads, 80 (8.4%) included one tobacco ad, 7 (.7%) included 2 and 1 (.1%) included 3 tobacco ads, for a mean of .10 across all of the newspapers.

Turning to alcohol findings, the total number of newspapers with alcohol display advertisements was 386, with 569 (59.56%) of the newspapers containing no alcohol advertisements. The number of alcohol ads appearing in the newspapers were as follows: 193 (20.2%) of the newspapers had 1 ad, 85 (8.9%) included 2 ads, 46(4.8%) included 3 ads, 25 (2.6%) included 4 ads, and the number of papers with more than 4 dropped considerably under 1% with the most ads included in 1 newspaper at 19. The mean number of alcohol advertisements per issue across all the newspapers was .94.

All told, 1,471 alcohol stories and 466 tobacco stories were analyzed from the newspapers. Of the 955 newspapers, 269 (28.2%) contained one story about alcohol, with 182 (19.1%) containing 2 stories, 107 (11.2%) containing 3 stories, 61 (6.4%) containing 4 stories, 26 (2.7%) containing 5 stories, 14 (1.5%) containing 6 stories and the number of newspapers containing more than 6 stories under 1%, with one newspaper publishing 12 stories. The mean number of stories published was 1.57.

As for tobacco stories, 256 (26.8%) of the newspapers contained 1 story about tobacco, 71 (7.4%) contained 2 stories, 13(1.4%) contained 3 stories and the number of papers publishing more than 4 stories was under 1% of the total newspapers. The mean was .50 stories across the newspapers.

Hypothesis testing used bivariate correlation analyses, with subgroup comparison on the control variables. Testing Hypothesis 1, which predicted a negative relationship between the number of tobacco advertisements and number of articles published on tobacco, yielded no significant finding (r= −.05, p <.41, N=334), suggesting no relationship between the amount of news coverage of tobacco and tobacco display advertising. Looking with subgroupings of the two control variables, the nonsignificant result holds up. Subgrouping according to above and below the median split on size of paper yielded the following partial correlations: Relationship between number of tobacco advertisements and number of tobacco stories in larger newspapers r= −.02, p<.76, N=318; relationships between number of tobacco advertisements and number of tobacco stories in smaller newspapers r= −.01, p<.83, N=286. Subgrouping according to geographical regional showed the relationship between number of tobacco ads and tobacco stories in the Northeast at r=.03, p<.68, N=159; in the Midwest at r= −.04, p<.55, N=258; in the South at r= −.03, p<.60, N=328; in the West at r= −.07, p<.36, N=193.

Hypothesis 2, which predicted a negative relationship between the amount of coverage of alcohol and the amount of alcohol display advertising, controlling for newspaper size and geographic region, was not supported (r=.08, p<.17, N=334). Looking at the patterns of relationships within the control variable subgroups showed nonsignificant findings, as well. Subgrouping according to size showed the following relationships between number of alcohol ads and number of stories about alcohol: For newspapers above the median split, r= −.04, p<.54, N=318; for newspapers below the median split, r=.04, p<.46, N=286. Subgrouping according to geographical regions for the same relationship results in the following: The Northeast r=.06, p<.47, N=159; the Midwest r=.03, p<.62, N= 258; the South r= −.02, p<.73, N=328; the West r=.01, p<.91, N=193.

Turning to the study’s research questions, RQ1 tested whether there were differences across U.S. regions (the Northeast, Midwest, South and West) regarding the ratio of tobacco advertisements to total pages. This did not yield a significant result (X²=1.517, p <.65). RQ2, which examined the same differences for alcohol advertisements, also did not show significant differences across geographic regions (.X²=3.64, p <.30).

Research Question 3, which questioned the relationship between the ratio of tobacco ads to total pages by newspaper size, also proved non-significant (X²=8.43, p <.24). Research Question 4, which examined the same question as RQ3, but for alcohol advertising, also yielded a non significant difference (X²=3.961, p <.80).

Discussion

The infrequency with which tobacco advertisements were found in newspapers is encouraging from a public health standpoint. It appears that the Master Settlement Agreement and changes in public norms, including social stigma against consumption of tobacco, have led to the near disappearance of tobacco advertising from newspapers, where once it was extremely prominent. While the frequency of alcohol advertising in newspapers was higher than for tobacco, it is still in absolute terms low, averaging less than one ad per newspaper edition.

It is important to note that these advertisements have likely moved to other media where youth may be still be a principle target. Although these findings suggest modest alcohol and tobacco advertising levels in general interest U.S. newspapers does highlight another issue of considerable public health concern. Although college newspapers were not part of this study they have been the subject of recent content analyses which suggest they are a target for advertising alcohol and tobacco.⁵⁵ Such findings are particularly disturbing when placed in the context of the present study: Heavy substance advertising in college newspapers, which reach an audience that is largely underage at least with respect to alcohol use, is not a reflection of norms in newspaper advertising, but is in fact anomalous and therefore perhaps worthy of greater attention.

The inability to replicate previous research that found a negative relationship between amount of tobacco advertising and editorial coverage (albeit in magazines, not newspapers) is unsurprising. Even if the same had been true of newspapers two decades ago, tobacco advertising has decreased to such a low level in U.S. newspapers that its presence or absence is unlikely to be a source of economic concern to publishers.

The lack of findings on the relationships between tobacco and alcohol advertisements and newspaper size suggests that information about health risks may be spreading effectively across the United States. This might also be evident in reductions nationally in consumption of alcohol and tobacco within certain population subgroups, but that is beyond the scope of this study.

Cigarette smoking has been decreasing for most age groups, with some rises over the past 15 years among youth.⁵⁶ Youth 18–24 generally decreased from a high in 1983 of 35% smoking to around 30% smoking in 2003, with occasional small increases and decreases over the last few years. Adults in all other categories showed a decrease in cigarette smoking: 24 to 44-year-olds moved from 35% to 30% smoking; 45 to 65-year-olds moved from 35% to 28% smoking; people 65 and older moved from 18% to 10% smoking.

Although excessive drinking behaviors have been on the increase for youth, alcohol consumption generally has been steady, if not declining for the adult population, gauged by having at least one drink a day or having a drink in the last 30 days.⁵⁷

It is possible that tobacco and alcohol newspaper advertising is decreasing simply because the age of newspaper readers is increasing, and, consequently, newspapers may no longer be a good outlet for advertising to youth and young adults. However, people across all age groups, as well as people across education groups, are reading newspapers less.⁵⁸ It is possible that, decreases in newspaper advertising are at least partially caused by advertisers using media outlets that target consumers more effectively than do newspapers.

Although the prediction was not supported the overall finding regarding the lack of impact of tobacco and alcohol advertising on newspaper editorial coverage may be viewed as optimistic from the standpoint of public policy. Such results are encouraging, given the dismal forecasting by many media economic theorists with regard to market driven journalism.⁵⁹ However, this finding may be less attributable to virtue amongst editors and publishers than to our other major, descriptive finding regarding the relatively small role of alcohol and tobacco advertising in the support of most newspapers.