Article Text

Download PDFPDF

Every document and picture tells a story: using internal corporate document reviews, semiotics, and content analysis to assess tobacco advertising
  1. S J Anderson1,
  2. T Dewhirst2,
  3. P M Ling3
  1. 1Center for Tobacco Control Research and Education, University of California, San Francisco, California, USA
  2. 2College of Commerce, Department of Management and Marketing, University of Saskatchewan, Saskatoon, Saskatchewan, Canada
  3. 3Center for Tobacco Control Research and Education, Department of Medicine, Division of General Internal Medicine, University of California, San Francisco, California, USA
  1. Correspondence to:
 Pamela M Ling MD, MPH
 Department of General Internal Medicine, CTCRE, Box 1390, 530 Parnassue Avenue, Suite 366, University of California, San Francisco, CA 94143-1390, USA; pling{at}


In this article we present communication theory as a conceptual framework for conducting documents research on tobacco advertising strategies, and we discuss two methods for analysing advertisements: semiotics and content analysis. We provide concrete examples of how we have used tobacco industry documents archives and tobacco advertisement collections iteratively in our research to yield a synergistic analysis of these two complementary data sources. Tobacco promotion researchers should consider adopting these theoretical and methodological approaches.

  • B&H, Benson & Hedges
  • B&W, Brown Williamson
  • ITL, Imperial Tobacco Limited
  • LTDL, Legacy Tobacco Documents Library
  • PM, Philip Morris
  • UCSF, University of California, San Francisco
  • corporate documents
  • tobacco
  • advertising
  • research methods
  • communication

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

As of March 2005, roughly a dozen books and over 220 papers were published utilising analyses of tobacco industry documents,1 reflecting that this research approach has rapidly become of great consequence since 1995, when tobacco documents research was first publicly in print.2–6 Nevertheless, there have been recent appeals for greater clarity and rigour in the conduct and reporting of documents research methods,7,8,9,10 including Carter’s call for synthesis (rather than simple description) in documents research. Researchers have also urged for increased efforts to triangulate data found in industry documents with other sources to evaluate the veracity of statements, or to evaluate the results of internal industry activity, such as whether industry plans were carried out.7,9 A parallel source of data about the tobacco industry that has developed alongside tobacco industry documents archives is collections of tobacco advertising. In an effort to respond to these calls, we detail methods for analysing advertising archives and documents databases synergistically.

This article is meant to serve several purposes. First, we discuss a theoretical framework from the discipline of communications that may be useful in guiding documents research on tobacco promotion strategies. Second, we argue that it can be illuminating for tobacco control researchers to analyse advertisements formally as a complement to their review of industry documents (rather than using advertisements as mere illustrations), in which we encourage attempts to integrate research approaches that are commonly used for each separate component of the communication process. Semiotics and content analysis are introduced as two approaches for studying advertisements. Semiotics, also sometimes called semiology, is defined as the science or theory of signs.11,12 Content analysis, meanwhile, is defined as “a systematic technique for analyzing message content and message handling—it is a tool for observing and analyzing the overt communication behavior of selected communicators”.13 Both approaches are well established research methods that emerged from disciplines that are not necessarily familiar to those in public health. Third, we describe methods for analysing tobacco industry documents and advertising collections in an iterative fashion, wherein analysis of one data source assists in conducting and refining analysis of the other, yielding a synergistic analysis of two complementary data sources.

Many influential studies of tobacco marketing and advertising place primary emphasis on either advertisements or planning documents, yet few studies account for both in a formal and synergistic fashion.14–27 A minority of studies assess documents and advertisement content in more detail.28–35 We are aware of one published study that used an analysis of tobacco industry documents to generate hypotheses that were tested using a formal advertising content analysis,36 and another study that explores theoretical and methodological issues in assessing viewers’ responses to smoking depictions in magazines.37 We further this literature by outlining methods for explicitly connecting tobacco documents and advertising archives to avail researchers of the strengths of both. We present cross-disciplinary strategies to increase the richness of data on tobacco promotion strategies found in tobacco industry documents by using tobacco advertising archives; similarly, we discuss how information contained in tobacco industry documents can inform or augment analyses of advertising. We believe this is the first article to treat these methodological issues directly.


Providing a precise definition of communication theory is challenging, as this is a highly contested topic among communication theorists and several models have been generated. Many early models of communication are heavily scrutinised because they allegedly fail to: (1) distinguish whether the audience should be studied as individuals or subgroups; (2) specify the social and cultural factors (for example, class) that might impact message meaning; and (3) account for the context of media reception and consumption (for example, someone might interpret a message differently in an educational setting with peers compared to a home setting with family members). According to Lasswell, a leading US political scientist and communications scholar, communication theory is an attempt to illustrate and explain “who says what to whom to what effect”. We present a simple, generic model of communication (similar to those commonly presented in leading marketing textbooks),38–40 which is meant to be a brief introduction to a complex subject. For readers who seek a thorough overview and understanding of these central debates and rival models, we suggest referring to the work of Fiske, Moores, Nightingale, and Wren-Lewis.41–44

Traditionally, mass communication research has epitomised a sender—message—receiver model, with the communication system involving both encoding and decoding of messages (fig 1). The communicative process inherent in tobacco advertising is conceptually comprised of three key components: (1) the “intended” message; (2) the message form, channel, and content; and (3) the “received” message. First, the “intended” message consists of the aims, goals, and objectives of the sender or message source (for example, a tobacco firm), as well as the purpose behind their ad agency’s creation. The “intention” of a message may include linking various lifestyle images or “personalities” with a particular brand, or providing information about a product or service. The source(s) of the message are engaged in a practice known as encoding, in which they determine what to say to consumers and translate the message into words or symbols that they hope will have the same essential meaning to the receiver. For the researcher or analyst, knowledge of the key principles that underlie the production of an effective advertising and promotional campaign is essential when assessing the encoding process, including an understanding of market segmentation, brand management, and perhaps hierarchical models that relate to promotional objectives of informing, persuading, and reminding. Second, the message form represents the character of the promotion, such as spoken word, live action, or written and visual text, as it is transmitted via a selected message channel, such as radio, television, or newsprint, respectively. Message content refers to the actual text, artwork, animation, music, and so forth of the advertisement. Multiple media are typically used for delivering the content of a conventional advertising campaign, including television, radio, magazines, newspapers, outdoor/transit, direct mail, and the internet. In the case of the tobacco industry, the prevailing legal and political environment heavily influences which media are selected for promotional purposes. The decoding component of the communication process refers to the receiver translating the message. The “received” message epitomises the receivers’ interpretations, ideas, and meaning construction.

Figure 1

 The traditional communication process. “Noise” is defined as any distraction that diminishes the effectiveness of the communication process.

The sender and the receiver do not always arrive at common meanings; the encoded message and the decoded message do not necessarily correspond.45,46 Multiple readings of the same advertisement are possible, and readers may also differ in how they respond to the same textual stimuli, which may indicate varying personal and cultural histories or simply how much time they spend reading the advertisement. Nevertheless, advertisements are intentionally encoded with essential meanings such as a brand’s image or personality (for example, advertising for Winston conveys genuineness and blue collar masculinity), which are likely to be best understood by the target market of the brand being promoted. Consumer knowledge, beliefs, attitudes, and opinions are extensively researched by marketers, and if pre-testing of a proposed advertising campaign reveals that the “intended” meaning was not being “received”, the advertisements would be further refined or never put forward. Post-testing is conducted to ensure that consumers recognise and understand fundamental brand imagery and the essence of the message. To assist with creating an effective, relevant, understood message, the producers of advertisements refer to market research findings (that is, feedback), indicating that the production aspects of the communicative process do not occur within a closed system. In this way, consumers act as both the source and the receiver of a message.

An example of a specific documents search

Communication theory provides a framework for searching tobacco industry documents archives and organising and analysing the data. Table 1 displays types of documents and useful search terms for studying the different levels of the communication system. Organising searches and findings in terms of the intended message, the message itself, and the received message guides researchers in appropriately reconstructing and interpreting the processes by which tobacco companies create and refine advertising campaigns. Furthermore, this conceptual framework illuminates the points in the process at which communication proved either effective or ineffective.

Table 1

 Useful search terms in the Legacy Tobacco Documents Library for conducting research on tobacco advertising as guided by communication theory

With the Legacy Tobacco Documents Library (LTDL) as the documents source, we present a series of searches that were conducted for a previous project that led us to explore a Benson & Hedges (B&H) promotional campaign as an example of targeting women. At first glance, the “For People Who Like to Smoke” campaign from the late 1980s and early 1990s is not a clearly female-oriented campaign; indeed, B&H is considered a unisex brand.47–49 Nevertheless, during our searches of the LTDL using the combination “female”, “marketing”, and “198*” (a decade of interest with respect to the use of psychographics in ad creation), we learned that B&H was a brand commonly listed among market share evaluations of the “female brands”. We subsequently replaced “female” in our search with the term “Benson & Hedges”. This combination returned 4160 documents from the Philip Morris (PM) collection, the producer of B&H in the United States. We then replaced the term “marketing” with “creative” to retrieve 102 documents. One of the returned documents50 was covered with handwritten notes about “Social Acceptability”. This and similar documents pointed to the concern, especially among female smokers, about the erosion of the social acceptability of their tobacco use, and the desire of PM to create an advertising campaign that would address those concerns. Social acceptability became PM’s intended message for the “For People Who Like to Smoke” campaign launched in 1986.

An analysis of both documents and advertisements related to the Capri brand’s promotional campaign from the 1980s demonstrates the usefulness of concurrently examining advertisements in order to gain a more holistic picture of the communication process. Brown Williamson’s (B&W’s) Capri was described as an explicitly female brand sold on the basis of user imagery and emotional appeal (rather than tar characteristics), according to a detailed advertising plan found in the documents.51 The documents reveal that Capri advertising was heavily researched from 1986 through the end of the 1990s. That date range (1986 to 2000) guided our search of the Richard Pollay online archive ( and the Virginia Ernster collection of tobacco advertisements at the University of California, San Francisco (UCSF) for Capri advertising. Organising the images chronologically from earliest to most recent allowed us to mark the beginning and ending dates of several campaigns and slogans through which the brand evolved. The published advertisements also served as outside confirmation that the marketing plans found in the documents were executed and gave insight as to consumer targeting based on the publications in which the ads appeared (in the case of the Ernster collection). It should be noted, however, that the available advertising collections are not complete and cannot answer all questions about advertising or marketing campaigns. Nevertheless, accessing multiple collections improves a researcher’s ability to approximate a complete picture of a brand or campaign.


Semiotics and content analysis are both well-established methods of studying advertising messages and textual elements, with semiotics emerging from the disciplines of linguistics and literary and cultural analysis and content analysis derived from a social science tradition.12,52 We present a brief overview of both methods and refer the interested reader to expository textbooks for greater details on the subject.


Derived from structural linguistics and the work of Ferdinand de Saussure, semiotics is a qualitative approach for studying the content of advertisements. Semiotics, also sometimes called semiology, is defined as the science or theory of signs, where a sign within a system of meaning can be separated into two components: “the signifier” and “the signified”.11,12 The signifier refers to a tangible dimension that is the material vehicle of meaning, whereas the signified is the abstract dimension that actually “is” the meaning. Roland Barthes offered roses as an example: red-coloured roses are concrete dimensions that, within many cultures and contexts, serve as a signifier of romantic or passionate love.12 Like a rose, the cigarette is a sign that is rich in meaning. Malcolm Gladwell calls the cigarette “the ultimate expression of adolescent rebellion, risk-taking, impulsivity, indifference to others, and precocity”.53 Moreover, in brand-sensitive (consumer) cultures, when an individual smokes a particular brand of cigarette, he or she engages in an act of social and cultural distinction. In the US market, for example, Marlboro symbolises rugged, masculine, independent, and heroic overtones, and Virginia Slims has historically conveyed women’s liberation, femininity, and glamour. Such cigarette brands have traditionally been invested with these cultural meanings through the use of symbols and appropriately themed product advertising.

A detailed explanation of the methodology of semiotics can be found from several sources,52,54–56 and there are common steps in semiological analysis that can be taken.57 First, the researcher must determine which advertisements will be analysed and provide a basis for the selection. Criteria must be established such as the historical period of analysis (that is, the sample may consist of a particular advertising campaign and be considered cross-sectional, or conversely, the selected sample may be longitudinal with the advertisements representing a lengthy historical period) and the appropriate sources of the advertisements (for example, advertising archives, the internet, magazines).

A second step in semiological analysis may involve the compilation of a denotation inventory, accounting for both ad copy (linguistics, size, fonts, typography) and images (pictorials). Denotation is the initial meaning that a sign is designed to capture. The analyst may then examine higher levels of signification, taking into consideration connotation, myth, and referent systems. Connotation refers to extended, secondary, or implied meanings of a sign or text (an example of a connotative chain might include dark = night = mystery = fear = evil = forbidden desires), whereas myth is defined as an ancient story dealing with supernatural beings, ancestors, or heroes.11,57 Iconic brands such as Nike have a mythical element, with the brand name making reference to the Greek goddess of victory. With respect to cigarette advertising, the American cowboy and the creation of “Marlboro Country” have a mythical quality. Finally, referent refers to what a sign stands for. The analyst notes the colour, size, prominence, location, relative positioning, repetitiveness, and contrasts of the advertisement elements, and identifies the cultural knowledge that the advertisement makes reference to and is presumably held by the targeted reader.

Moreover, the analyst observes whether different elements may be polysemic, accounting for the fact that signs or texts have multiple, hierarchical meanings with some of them being designated as “dominant/preferred” and others may be classified as secondary, tertiary, negotiated, or opposed. For example, promotions for the popular Canadian cigarette brand, du Maurier, which has a longstanding association with high quality and upscale imagery, once utilised the tagline “For people with a taste for something better” (fig 2). In this example, multiple meanings are apparent for “taste”. One interpretation is that the word refers to the supposedly distinguished flavour of the cigarette brand (referring to a sensory experience with product use); another interpretation is that the brand appeals to those people with the ability to discern what is high quality, aesthetically excellent, and appropriate. Figure 2 also demonstrates the importance of understanding the historical context, such that at the time of the advertisement’s circulation various symbols were seemingly used to communicate cutting-edge technology and quality, yet for the contemporary reader the same symbols appear dated or retro (for example, the stereo system seen in the background).

Figure 2

 du Maurier advertisement, 1986, depicting a reel-to-reel tape deck as cutting-edge technology, yet the technology is now seen as dated or retro.

Like any research method, there are strengths and limitations with a semiotic approach. Semiotics is heavily dependent upon the skill of the analyst; thus a researcher with considerable experience with both the methodology and tobacco advertising will likely generate a more comprehensive and informed analysis than a less experienced researcher. A semiotic approach stresses individual readings of messages. As such, though there is likely to be common agreement about some of the broad themes being communicated in an advertisement, complete agreement is unlikely among analysts when accounting for how detailed the textual analysis may be. The primary strength of semiotic analysis is in the richness of the information conveyed; the primary limitation is that, due to its exploratory nature, reliability or objectivity may be lacking. It is important that the semiotic analysis be performed on advertisements that are both representative of a particular advertising campaign and the brand’s positioning historically, as opposed to handpicking single advertisements to illustrate expedient points. So conducted, semiotic analysis by an experienced analyst who is knowledgeable of advertising and tobacco branding strategies can be a useful technique to include in one’s analytical toolbox for adding a layer of richness to the interpretation of advertising content.

Content analysis

Content analysis is an alternative approach for studying advertisements. Dewhirst and Pollay and chapter 5 of the 1994 US Surgeon General’s report, Preventing tobacco use among young people, provide an overview of existing content analysis studies pertaining to tobacco advertising.58,59 Content analysis is defined as “a research technique for the objective, systematic, and quantitative description of the manifest content of communication”.60 According to Berelson’s definition, objective, systematic, and quantitative are distinguishing characteristics of content analysis.

To enhance objectivity, coders undergo training and are presented with precise and detailed operational definitions, categories, rules, and procedures for analysing communication content. The data are preferably analysed by coders who are unaware of the hypotheses established by the researchers. Content analyses ideally involve multiple coders (including some separate from the authors) that assess communication content independently. Thus, judgments should not entail consultation with the researchers or other coders. Applying the same operational definitions, categories, rules, and procedures to the same dataset, the various coders should secure highly replicable and reproducible results and arrive at similar conclusions. Inter-coder reliability is calculated to indicate the level of agreement among coders.12,60–63 Overall inter-coder reliability should preferably exceed 0.85 (as percent agreement).62

Systematisation implies that developed definitions, categories, rules, and procedures are applied to the dataset consistently. To ensure a proper basis for comparison, descriptive categories should not be added after the coding process has commenced and definitions are not to be modified. The coders should proceed through the dataset in a different and counter-balanced sequence to help minimise any potential learning and maturation effects. Preferably, the data are pre-selected through recognised randomising procedures and represent an unbiased, representative sample of defined communication content. Researchers should not actively search for advertisements idyllically supportive of their arguments. Content analyses are considered systematic because scientific problems or hypotheses are examined.12,61–63 The description of communications content resulting from content analysis should have general applicability and theoretical relevance.60–62

Quantification is another defining characteristic of content analysis. Content analyses are utilised to measure the extent that analytic categories appear in communication content (that is, relative emphases and omissions are determined) and the data generated are to be amenable to statistical methods. Although researchers often equate quantification with strict frequency counts and the assignment of numerical values, quantitative words such as “more”, “always”, “increases”, and “often” may also be used.60,62

A key strength of content analysis is that it allows researchers to seek objective answers to research questions; by objective, we acknowledge that this merely reflects acceptable reliability measures among coders so that the analysis is seen as more than an individual and impressionistic interpretation. Additional strengths of content analysis are that qualitative data can be dealt with in quantitative terms, and a large number of advertisements can be assessed in which patterns and trends may be observed that might otherwise be difficult to detect.12 Nevertheless, content analysis has been dismissed by some scholars, including Colin Sumner, who argue that this approach places emphasis on the “repeatability” of signs rather than their signification.64 Content analysis should not be used to make deductions about the reader’s interpretation of an advertisement. The definitions put forward of content analysis reveal that, in researchers’ attempts to maximise overall inter-coder reliability, the focus of assessment is often the manifest content of communication (that is, the denotative level of an advertisement) rather than the latent or connotative levels.12 For manifest content analysis, coders limit their analysis to explicit advertising content rather than also interpreting the likely meaning conveyed by advertising messages. Reliability measures may be misleading because data that generate high percentage agreement among coders may reflect a simplistic data-coding instrument and mundane information collected.

Although semiotics and content analysis are in many respects relatively different methodologies—and researchers typically select one approach at the expense of the other—we encourage researchers to explore a multi-method approach as research questions deem necessary (with the strengths of one approach in part answering the weaknesses of the other). Leiss, Kline, and Jhally have proposed a combined semiotics—content analysis approach.12


An example of semiotic analysis and the use of internal documents

Imperial Tobacco Limited (ITL) is Canada’s largest tobacco manufacturer, accounting for more than two out of every three cigarettes sold domestically, and Player’s is one of their flagship trademarks. According to ITL’s Player’s 1988 document, “In order to move Player’s Light up on the masculinity dimension, we will continue throughout [Fiscal] F’89 to feature creative which reflects freedom, independence and self-reliance in a relevant fashion for young males”.65 The advertisement in fig 3 was created for the campaign discussed in the planning document, and it was representative of many preceding and concurrent ads for the brand.30,31,66 For demonstration purposes, we conducted a brief semiotic analysis of this ad. We offer the following as our observations rather than as a definitive analysis of the advertisement. Although semiotics inevitably involves the assessment of a relatively small number of ads (typically around five in a manuscript), it is expected that the ads in the dataset are representative of other promotions from the same campaign and that the ads have an inter-textual relationship (that is, meaning is not likely to be isolated to the single advertisement). It is with this expectation, and due to space considerations, that we offer this demonstration analysis of one advertisement only.

Figure 3

 Player’s “hiker” advertisement, 1988, showing a solitary man in the wide-open, rugged outdoors to communicate freedom, independence, masculinity, and self-reliance. The tagline also communicates boldness and autonomy.

To convey masculinity, a male model is depicted, and the model’s rugged appearance is evident with his physical features, clothing worn, and the geographical setting. Freedom—defined as the condition of being free of restraints—is represented by a landscape of wide-open space that includes a predominant skyline, which is blue-coloured to match what is seen on Player’s packaging. The physical setting suggests a lack of restrictions and an opportunity for free will relative to an urban environment dense with people, traffic lights, regulations, and police surveillance. A secondary meaning of freedom is bluntness, outspokenness, and boldness, which is suggested by the tone of the tagline, the reputation of Player’s as a strong tasting cigarette, and the brand’s longstanding association with risk-taking, adventurous activities (for example, windsurfing, kayaking, hang-gliding, and auto racing).30,34,66 Both the tagline (the use of “you” makes it more personalised) and showing one person in the advertisement communicates independence, autonomy, and self-reliance. Player’s advertisements have traditionally depicted individual sports rather than team sports, and it is commonplace that activities are shown during non-strenuous moments. The hiker, for example, is shown taking a break rather than during an arduous instant that might link the model with heavy breathing and lead readers to counter-argue that such an athlete is unlikely to be a smoker or a smoker is unlikely able to competently pursue such an activity. Branding refers to the use of a name and logo to identify a product,38,67,68 and in this case, the Player’s brand name conveys tradition by paying homage to John Player, a main founder of Imperial Tobacco, and the sailor logo, complemented by “Player’s Navy Cut”, embodies masculinity, independence, and freedom. Finally, the advertisement is pictorial (the sky encompasses the most text space), with the ad copy limited to the tagline statement and a “branding” orientation.

An example of content analysis and the use of internal documents

A review of internal tobacco industry documents can help inform hypotheses that are formulated for a content analysis study. For example, with “masculinity” being a central theme communicated in the promotion of Player’s, as stated in the 1988 Player’s planning document,65 it is anticipated that depicted models will tend to be exclusively males. Thus, a hypothesis might include: “When models are portrayed, males exclusively will be the most common depiction for Player’s”. This hypothesis could be tested by the following item of a data coding instrument:

Sex of People Featured

  • 01-Male, alone

  • 02-Multiple males

  • 03-Female, alone

  • 04-Multiple females

  • 05-Mixed

  • 09-N/A

Referring to fig 3, the coder would likely select “01”. Our hypothesis would be supported if the combined frequency of categories 01 and 02 exceeded the combined frequencies of categories 03, 04, and 05 for the Player’s advertisements assessed in the dataset. Other items that might be coded when testing hypotheses related to cigarette advertising include copy volume (that is, number of words), demonstrations of product use, information categories (for example, claims pertaining to the product’s price or availability), and the presence or absence of lifestyle dimensions such as adventure or nationalism. A data-coding instrument, consisting of 65 items specific to the assessment of cigarette advertising, has been developed for content analysis purposes.69

Using the advertising collections and documents archives together

An iterative method allows the researcher to build upon knowledge obtained through both advertising and documents archives, thereby gaining insight into the complexities, breadth, and depth of advertising strategies employed by the tobacco industry in their attempts to acquire or retain consumers. The researcher may, for example, begin by analysing a collection of advertisements that suggest an intended message in a specific campaign. One then turns to the documents archives to explore the veracity of these suppositions and may discover a competitive motivation behind the creation of a brand’s image. The researcher may return to the advertising collections to compare images in one brand’s campaign with those of a competing brand, or to assess changes in the promotional campaign over time. A further search of the documents may reveal the superior performance of one campaign over another.

Again, using Capri advertising and the LTDL as a document source example, we observed a change in Capri advertising in 1995 as the “She’s Gone to Capri” promotional campaign was introduced, which motivated a targeted search for documents on this campaign. We used the search terms “Capri”, “market*”, and “1995*” to begin uncovering the “She’s Gone to Capri” story. One document in particular, “B&W Capri Objectives and Strategies”,63 provided us with the brand’s sales history to date (September 1995), the demographics (women 35–59 years of age) and psychographics (women who value femininity, elegance, and gracefulness) of the brand’s users, and the company’s expressed intent to position the brand as an “escape from the everyday” through the use of “place (architecture, mood)”.70 This description fit with our observation from the Ernster and Pollay advertising collections of the inclusion of gardens and architecture in the newly launched Capri campaign (fig 4). The setting of distant places and times, as well as the shift from boldness to elegance in the models’ wardrobes and body language, communicated the desired escapist message.

Figure 4

 Capri advertisement, 1995, using architecture, gardens, lighting effects, and model’s wardrobe and body language to communicate relaxation, elegance, private time, and escape in a romantic, Mediterranean locale.

Another review of the advertising collections revealed that “She’s Gone to Capri” images appeared in 1995, so we searched for marketing plans, creative briefs, and top-line reports for the five or so years preceding this date. These searches led us to our earlier discovery of Project Delta, the first rumblings in B&W’s marketing department about a relaxing, escapist, fantasy-evoking brand called “Capri”, and artists’ renderings or early mock-ups of images that eventually became printed Capri advertisements. Thus, we had uncovered both the intended message and several variations of message content.

When reviewing the “She’s Gone to Capri” images, we also discovered a noteworthy change: models were eventually eliminated from the advertisements. The model-absent images first appeared in 1998, so we searched documents just before then for clues about the change. Indeed, B&W executives noted two general reasons: (1) the effect the change had on making the scenes more “ownable” and personalised for the viewing audience or reader; and (2) the possibility of future advertising stipulations restricting the use of certain imagery (such as the depiction of human models). In this case, consumer feedback showed B&W that the intended message was being correctly received by a sampling of the target market.


Drawing from the discipline of communications, we have presented a theoretical framework for conducting tobacco documents research on advertising. By parcelling the mechanisms of tobacco advertising communication into separate elements—that is, the sender (the tobacco industry or advertisers), the message (as encoded by tobacco firms and their hired communications firms), and the receiver (the potential consumer)—we have suggested documents search strategies that help reconstruct the process of tobacco advertising from early brand planning to final ad executions.

We have introduced semiotics and content analysis and described their utility for researchers interested in tobacco advertising. We have also presented a methodology for making use of documents and advertising databases to systematise research on tobacco advertising. Further, we have offered examples of analysing both tobacco documents and advertising images that allow researchers to take advantage of the strengths of both data sources simultaneously.

Analysis of both documents archives and advertising collections can be organised by the different levels of the communication system. What the “sender” intends can be obtained through marketing plan documents and creative briefs. Collection and analysis of advertisements will shed light on message content. Prototype advertisements and drafts of ad copy found in the documents archives are useful for examining the development of message content, especially when commentary and discussion between executives is evident in those types of documents. Analysis of the advertisements themselves, qualitatively through semiotic analysis or quantitatively through content analysis, augments study of this level of the communication system. Finally, focus group transcripts, company presentations of market research results, and quantitative and qualitative test reports provide insight about how “receivers” may have decoded the message and allow researchers to track the feedback process.

Ideally, complete and comprehensive datasets would exist that all researchers could easily access, yet in reality gaps in the available data are inevitable; thus, trying to capture as much information as possible is important. Through iterative analysis of both the documents archives and advertising collections, researchers can attempt to determine the correspondence of message meanings across differing levels of the communication system. Researchers may avail themselves of both data sources to converge on a viable representation of the total process.

Virtually all tobacco documents research focuses on the past and should be treated with caution when predicting where the tobacco industry is headed next. Though we certainly see repeated patterns of activity within the tobacco industry, both the historical nature of findings and the dynamic nature of the industry itself must be acknowledged. Nevertheless, the concurrent use of documents and advertisements can facilitate a wide range of tobacco control activities regarding marketing. First, exposing cigarette advertisers’ actual intent in creating (encoding) persuasive messages can counter the tobacco industry’s public relations statements about responsible cigarette marketing. Second, understanding what message channels and content are most effective for the tobacco industry can guide policymakers in devising both adequately responsive and knowledgeably proactive measures for minimising the tobacco industry’s marketing effectiveness. Third, studying the responses of “receivers” to tobacco advertising can both educate tobacco control advocates about what carries meaning for different groups of consumers and inform the creation of counter-advertising campaigns that will resonate with them.

What this paper adds

Since the first research papers using tobacco industry documents were published in 1995, the field of tobacco documents research has grown dramatically. The number of tobacco advertising analyses based upon industry documents is increasing, though the majority of these papers do not offer theory-based, formal analyses of advertising images or campaigns.

This article presents a theoretical framework for researching tobacco advertising, provides examples for using tobacco industry documents archives and the available collections of tobacco advertisements in a synergistic manner, and offers two methods for formally analysing tobacco advertising images. These theoretical and methodological advances will increase the rigour of research on tobacco advertising and may facilitate both counter-advertising efforts and the refinement of tobacco control policies related to tobacco industry market research and promotion practices.


This work was supported by the American Legacy Foundation, the Canada-US Fulbright Program, the Canadian Tobacco Control Research Initiative Idea Grant, the Flight Attendant Medical Research Institute, and National Cancer Institute Grant number CA-87472. We would like to thank Edith Balbach, Stacy Carter, Gerard Hastings, and Rick Pollay for their extensive comments on a previous draft of this work.


View Abstract


  • Competing interest: none declared