Characteristics of Community Newspaper Coverage of Tobacco Control and Its Relationship to the Passage of Tobacco Ordinances

The Public Health Model of Reporting (PHMR), which promotes a public health perspective on reporting health issues, serves as a useful guide. Our goal is to descriptively assess types of public health information and story characteristics of tobacco news coverage based on this model. PHMR was originally created as a way to spur change in reporting related to crime and violence [13]. However, PHMR has proven useful for other health contexts involving conflict or debate, such as tobacco control [14]. The goal of PHMR is to describe the types of public health facts that would enable readers to gain a broader perspective on issues that may threaten public health [15]. PHMR advocates that part of the solution to public health problems is the provision of critically important information about the causes, costs and consequences of health issues that may prompt individuals to make proactive health choices and, thus, prevent many health problems [16].

Community newspapers provide information on tobacco issues, such as prevention programs and adoption of smoke-free policies [17‐19], and offer a platform for local community members to voice their opinions on the health and economic benefits of smoke-free ordinances [20, 21]. Journalists take a stance on tobacco control ordinances [22], which has been shown to influence their outcome. Of 15 Wisconsin cities with smoke-free ordinance campaigns, all communities that passed ordinances had more news coverage and stronger editorial support from the local newspaper than communities with unsuccessful campaigns [23]. Media coverage of public health issues is a primary way to encourage health-related policy change [24]. Media advocacy, i.e., tactically using mass media to support the public’s ability to organize and advance public health policies [8], has been used to achieve policy success related to tobacco and alcohol control [25].

Frames, which journalists use to communicate complex issues by highlighting some aspects of a particular problem [26], make complex health issues and topics such as tobacco control more salient to readers and can shape the direction of the debate [27]. A national content analysis concluded that stories about tobacco control were framed primarily in terms of political debate rather than public health impact [18]. Tobacco control frames—whether positive/negative or for/against smoke-free ordinances—may influence public opinions or attitudes developed during the formation and discussion of public policy [28]. Smoke-free policies/secondhand smoke and economic issues received the most coverage [29] and provoked the most controversy [29, 30], primarily expressed through editorials and letters to the editor [31, 32].

A number of studies have examined how local and national media cover tobacco control issues, with some consistent emerging trends. Overall, coverage was predominantly positive or neutral/mixed for tobacco control [22, 30, 33‐35], which applied to both news stories and editorials, although some studies have found higher negativity in editorials and letters to the editor [22, 35, 36]. Stories with a positive tobacco control slant had information about enforcement, emphasized the lack of negative economic consequences or the health and economic benefits of policies or worker protection [30, 34]. Negatively slanted articles discussed economic losses and hardships, government intrusion and overreach, inability to enforce policies, and smokers’ rights and individual rights [30, 34]. Other popular frames include non-smokers’ rights, smokers’ rights, business rights and health issues [27, 35].

There were eight main coding categories: (1) topic, (2) story type, (3) locale, (4) slant of event, (5) slant of opinion, (6) story tone, (7) headline tone, and (8) public health facts.

The unit of analysis was the news story. Public health facts were determined by examining individual sentences within news stories. Headline tone was determined by examining individual words in headlines. Tone and slant were determined by examining individual sentences in stories/opinion pieces. Five graduate students were the coders. The codebook was pre-tested by two initial coders. Inter-coder reliabilities were taken at the beginning, middle and end of the project to ensure coding was reliable and had not shifted during the coding process. An overall Scott’s pi [38] index of .82 was reached (range .60–.89) for nominal variables. Observed agreement was calculated at .89 (range .60–1.0) for the ratio-level PH facts.

A new column was added to our content analysis data set, indicating whether cities had passed (1) or not passed (2) a clean indoor air ordinance during the 6-year content analysis period.

Chi square tests of independence were conducted to examine the research questions. Statistical significance was tested at the 95 % confidence level. Statistical analyses were performed using SPSS version 23.

A total of 4671 stories on tobacco were analyzed (see Table 2), 62.7 % of those (n = 2929) were about tobacco control (TC): smoke-free ordinances or cigarette taxes/prices. The remaining 37.3 % of stories were about the health effects of smoking, anti-smoking programs/services, youth smoking issues, and others. TC was covered primarily on a local level, as 48.9 % of stories had a local angle and 41.5 % had a state angle. This differed from non-TC stories, which were primarily written with a state-wide focus (49.3 %) and had fewer local stories (30.4 %). See Table 3.

TC stories were primarily events/issues (57.6 %) followed by editorials/letters to the editor (41.7 %) and very few features (.7 %). Many more editorials were written on TC topics (41.7 %) versus non-tobacco control ones (12.1 %).

Of the event/issue/feature stories about TC, 52.9 % had a positive slant for TC and 37.2 % had mixed slant. Only 9.9 % upheld the position of the tobacco industry. Of editorials on TC issues, 53.6 % had a positive slant for TC, 11.3 % had a mixed impact on TC, and 35.2 % upheld the position of the tobacco industry.

In terms of tone of TC coverage, the vast majority of stories had a neutral tone (99.2 %) and a slightly lower number presented neutral headlines (93 %). However, 5.5 % of stories did run with negative headlines, compared to only 1.1 % with positive ones. For editorial content on TC, the tone of the text and headline continued to be predominantly neutral (86.3 % for stories and 84.9 % for headlines). However, a larger presence of negativity was noticed in both copy and headlines (12.2 % for copy, 11.4 % for headlines).

The ratio variable PH facts had high skewness and kurtosis and lacked normal distribution, which necessitated its transformation into a categorical variable. Four categories of facts were created: 0 facts, 1 fact, 2 facts, ≥3 facts for each type of PH fact. Chi square tests were applied to test for relationships with other variables.

TC stories used fewer public health facts compared to non-TC stories. For TC stories, there were .58 context facts on average, compared to .68 for non-TC stories; .22 costs facts on average, compared to .46 for non-TC stories; and .35 consequence facts on average compared to .81 in non-TC stories. Differences were significant for costs (χ ²  = 153.75, df = 3, p < .001) and consequences (χ ²  = 140.21, df = 3, p < .001).

When TC coverage was examined on its own, several relationships emerged between the use of public health facts and the stories’ locale, type, and slant. Chi square tests were used to check for differences between expected and observed frequencies of public health facts. Standardized residuals helped examine for differences in each cell of the test. When those residuals were greater than ±1.96, differences in the cell were deemed significant.

A significant relationship existed between locale and number of facts (\(\chi_{\text{context}}^{2}\) = 41.15, df = 6, p < .001; \(\chi_{\text{costs}}^{2}\) = 78.82, df = 6, p < .001; and \(\chi_{\text{conseq}}^{2}\) = 18.75, df = 6, p = .005). Standardized residuals showed that more state stories contained 1 and 2 context facts, while fewer local stories had 2 facts than expected. On costs, local stories had fewer cost facts than expected, while state stories had more than expected. On consequences, only national stories had more 2 and 3 facts than expected.

A significant relationship existed between story type and number of facts for context and costs (\(\chi_{\text{context}}^{2}\) = 26.69, df = 3, p < .001; \(\chi_{\text{costs}}^{2}\) = 21.88, df = 3, p < .001). Standardized residuals showed that there were more news/features with 2 context facts than expected and fewer editorials with 2, 3 or more facts than expected. There were more news/features with 1 cost fact than expected and fewer editorials with 1 cost fact than expected.

There was a significant relationship between the slant of events towards TC and the number of PH facts used (\(\chi_{\text{context}}^{2}\) = 48.71, df = 9, p < .001; \(\chi_{\text{costs}}^{2}\) = 36.35, df = 9, p < .001; and \(\chi_{\text{conseq}}^{2}\) = 18.04, df = 9, p = .035). Standardized residuals showed that events that were positive for TC were covered with more context facts than expected, while events that were negative for TC had fewer stories with 1 fact than expected. Events with mixed slant on TC were covered with more stories with 1 and 2 cost facts than expected and more stories with 3 or more consequence facts than expected.

There was also a significant relationship between the slant of opinion towards TC and the number of PH facts used (\(\chi_{\text{context}}^{2}\) = 107.12, df = 9, p < .001; \(\chi_{\text{costs}}^{2}\) = 35.01, df = 9, p < .001; and \(\chi_{\text{conseq}}^{2}\) = 51.99, df = 9, p ≤ .001). Editorials with a positive slant for TC had more stories than expected with 1 context fact and those that withheld the position of the tobacco industry had fewer stories than expected with any number of context facts. For costs facts, fewer editorials than expected with both positive and negative slant for TC contained 1 fact. For consequences, more opinion pieces with positive slant than expected contained 1 and 2 facts, while fewer opinion pieces with negative slant contained 1 and 3 or more facts.

Of the overall number of stories on tobacco in Missouri (N = 4671), 51.2 % was published in towns/cities with smoke-free ordinances. There was a significant difference in the number of TC stories published in towns with a smoke-free ordinance and without one (χ² = 412.9, df = 1, p < .001). Towns with an ordinance had more stories on TC in their newspapers and fewer stories on other tobacco topics than expected. Towns without a smoking ordinance had fewer stories on TC but more stories on other tobacco topics.

There was a significant difference in the types of stories published in towns with smoke-free ordinance versus towns without (χ² = 159.59, df = 1, p < .001). In towns with ordinances, more editorials and fewer news stories were published than expected. On the contrary, in towns without ordinances, there were more news stories and fewer editorials. The tobacco-related topics that dominated the newspaper coverage also differed (χ² = 506.55, df = 5, p < .001). In towns with ordinances, stories about smoke-free ordinances and cigarette taxes (i.e., stories on TC) were more common than expected, while the remaining topics appeared less often than expected. In towns without ordinances, the opposite was observed: more stories than expected on smoking effects, anti-smoking programs and youth issues and fewer stories than expected on tobacco control topics.

Last, the relationship between presence of ordinances and PH facts was examined and found significant for costs (χ² = 36.69, df = 3, p < .001) and consequences (χ² = 29.14, df = 3, p < .001). In towns with ordinances, fewer stories than expected contained 1 cost fact and 2, 3 or more consequence facts. In towns without ordinances, more stories than expected contained 1 cost fact and more stories than expected contained 2, 3 or more consequence facts.

Much of public knowledge about health problems associated with secondhand smoke comes from the news media. The PHMR provides a theoretical context in which to understand frames and public health facts associated with tobacco control coverage. Our results showed that coverage may not be living up to the standards set forth by the PHMR. Of the three types of public health facts examined, context facts were most prevalent. Additionally, while we might expect that tobacco control stories would highlight the positive health benefits of enacting smoke-free policies, our results revealed the opposite—there were fewer public health facts present in tobacco control as compared to non-tobacco control stories. Towns that already had passed a smoke-free ordinance had fewer public health facts overall. Reporters were also more likely to use episodic coverage and avoid public health facts that would likely provide context and depth to help readers understand the health problems associated with secondhand smoke.

These findings are useful for media advocates who seek to inform communities through exposure to targeted messages, and mobilize them to influence policy makers and voters to reform policy. The findings also offer practical suggestions for health and strategic communicators and those involved in public policy, as news stories and editorials can be used to bring about health related and public policy changes to communities. Specifically, health communicators should pay more attention to editorials, where much of the debate about policy changes occurred. Editorials contained higher levels of negativity than news stories and were used actively by proponents of the tobacco industry to oppose restrictive measures. Therefore, health advocates need specific strategies addressing editorial content in local newspapers when planning tobacco control campaigns. Health advocates should also provide relevant local information to journalists, as findings showed that coverage of tobacco control was predominantly local. Further, a higher use of public health facts about the consequences and costs of secondhand smoke could provide more persuasive power to news stories and opinions. The reasons for introducing smoke-free ordinances are the harmful effects of secondhand smoke and the associated healthcare and societal costs. Yet, these reasons are rarely mentioned in stories about ordinances, and using more public health facts on costs and consequences could address that. Finally, towns without smoke-free ordinances which are considering policy changes would benefit from an early visibility of the issue in the local print media, as people in these towns rely on local newspapers for their health news [39]. That would set the agenda and drive the issue forward, as well as could help health advocates outline the key benefits of such policies from the start. Local newspapers have been shown to be important for informing small communities (and city leaders) about the pulse and voice of where residents stand on a given issue.

One of the strengths of this study was the ability to relate content analysis results back to the environmental/societal context in which they appear, i.e., the cities in which the newspapers reside. However, causality cannot be established using content analysis data. Additionally, a content analysis cannot tell us about the intentional or unintentional (i.e., gatekeeping) selections of journalists, editors or publishers of the newspapers. Although the total number of public health facts was counted in each story, we limited the number of public health facts to three main types, which may not have captured other types that may be present. Last, while we assume that public health facts have an impact on public understanding of health issues and on individuals’ health behaviour, our study cannot determine whether this is the case. Future studies can remedy these limitations.