Article Text

Test of “Light” cigarette counter-advertising using a standard test of advertising effectiveness
1. Saul Shiffmana,
2. Steven L Burtonb,
3. Janine L Pillitteric,
4. Joe G Gitchelld,
5. Michael E Di Marinod,
6. Christine T Sweeneyd,
7. Paul A Wardleb,
8. Gary L Koehlerb
1. aPinney Associates and University of Pittsburgh, Pittsburgh, Pennsylvania, USA, bGlaxoSmithKline Consumer Healthcare, Pittsburgh, Pennsylvania, USA, cPinney Associates, Seattle, Washington, USA, dPinney Associates, Bethesda, Maryland, USA
1. Saul Shiffman, PhD, Pinney Associates, 201 North Craig Street, Suite 320, Pittsburgh, PA 15213, USAshiffman{at}pinneyassociates.com

## Abstract

OBJECTIVE To evaluate systematically the effectiveness of six advertising strategies (two message strategies presented in three different contexts) designed to promote smoking cessation by addressing smokers' misperceptions about Light cigarettes.

DESIGN Smokers viewed one of six, 30 second test television concept advertisements, which varied by message (one emphasising how the sensory effects of Lights can be deceptive, the other describing the effects of vent blocking) and by ad context (non-commercial public service announcement (PSA), promotion of unbranded nicotine replacement therapy (NRT), or promotion of branded NRT). The effectiveness of each advertisement was determined using a validated advertising testing system in which ads were viewed in the context of reviewing a pilot television programme. Response to ads is assessed through shifts in subject choices of products offered as prizes before and after viewing the test advertisements. Included among the possible prizes were cigarettes and various pharmacotherapies for smoking cessation.

SUBJECTS Daily smokers (n = 1890) of Regular (34%), Light (47%), and Ultra Light (19%) cigarettes recruited from eight US cities.

MAIN OUTCOMES MEASURES The primary outcome of interest was the shift away from cigarettes as the selected prize following exposure to the test advertisements. Secondary outcomes of interest included movement away from Light cigarettes and movement towards assisted quitting products.

RESULTS Smokers who saw the advertisement emphasising the sensory characteristics of Light cigarettes were more likely than subjects who saw the advertisement emphasising the effect of vent blocking to move away from cigarettes (OR = 1.97, 95% confidence interval CI 1.25 to 3.09; χ2(1) = 8.69, p = 0.003). Similarly, subjects who saw the advertisement framed as a PSA, rather than as a promotion for either a branded or unbranded NRT product, were also somewhat more likely to move away from cigarettes (OR = 1.51, 95% CI 0.94 to 2.40; χ2(1) = 2.97, p = 0.085). The effect was observed regardless of sex, age, or type of cigarette smoked.

CONCLUSIONS Addressing smokers' sensory perceptions of Light cigarettes and presenting this information in an impartial way is likely to be an effective communication strategy for counter-marketing Light cigarettes.

• Lights
• counter-marketing

## Statistics from Altmetric.com

Despite their apparent failure to reduce health risks,1 cigarettes labelled as “Light” are aggressively promoted by cigarette manufacturers as a reasonable alternative to quitting smoking.2 By convention, Light cigarettes are those with nominal tar delivery between 7–15 mg, and Ultra Lights are those with nominal tar delivery of 6 mg or less. Despite years of press attention to the contrary, many smokers of so-called Light cigarettes continue to believe that they convey a substantial health benefit.3One potential strategy for countering tobacco industry marketing tactics and promoting smoking cessation is to provide smokers with accurate information about the risks of smoking Light and Ultra Light cigarettes. Kozlowski and colleagues developed and tested a 60 second “radio message” informing smokers of the true nature of Light cigarettes.4 The message attempted to convey several important ideas, including: the existence of filter ventilation and the problem of filter vent blocking; the notion that Light cigarettes can provide smokers with just as much tar and nicotine as Regular cigarettes; and that smokers can be easily fooled by the mild taste of Light cigarettes. Following exposure to the message, Light smokers reported it would make them more likely to quit smoking.

In an attempt to extend this line of research further and refine effective message strategies, we previously conducted a randomised trial comparing three different messages, two of which were designed to address myths surrounding the use of Light and Ultra Light cigarettes.5 One message followed an informational strategy, modelled on the work of Kozlowski and colleagues, that explained how blocking filter vents can effectively increase tar and nicotine exposure (“Vents”).4 A second message, based on qualitative research with smokers, and on theoretical and empirical work on somatic perception as an influence on smoking beliefs and behaviour,6 7 addressed smokers' sensory experience that Light and Ultra Light cigarettes felt milder and were thus reasoned to be safer (“Feel”). A third message did not address Lights and Ultra Lights, but emphasised the health risks of smoking in general. Smokers participating in a telephone survey were read one of three brief messages embodying the three message strategies, and their beliefs about Lights and Ultra Lights and interest in quitting were assessed before and after message exposure. The Feel message consistently had the greatest impact on beliefs about these cigarettes and on intentions to quit smoking. These findings highlighted the important role of sensory feedback in supporting misconceptions about Light and Ultra Light cigarettes and the importance of specifically addressing this issue in communications designed to alter attitudes and beliefs about these cigarettes.

This prior work suggested that messages addressing sensory experiences would be most effective at changing beliefs and behaviour. However, the studies were limited in the methods used to deliver and test the messages, which were read to subjects over the phone by an interviewer. On the one hand, this mode of delivery weakens the impact of the messages: the messages were presented only once, they lacked the visual appeal or persuasion that would characterise a television commercial, and they were delivered by an anonymous source with no authority. On the other hand, the effect of the messages might have been heightened because subjects were asked to focus on the message, and the message exposure was embedded in a long interview about smoking, which would have caused respondents to focus on their smoking behaviour. In contrast, real television commercials are usually seen in a cluttered media environment, competing for attention with other commercials and the television shows themselves, and without a prior priming focus on smoking. Finally, the survey measure of intention to quit was hypothetical, without connection to real world choices.

The effectiveness of an advertisement may also be affected by the context—for example, whether the ad is seen as trying to sell something, which affects the presumed source of the message and its motivation. Accordingly, this study also examined the role of message context in influencing behaviour. Specifically, both the Feel message and the Vents message were presented in three different contexts: as a public service announcement (PSA) in which no reference is made to any stop smoking products and no advertising sponsor is identified; as a non-branded stop smoking product advertisement in which smokers are directed to use the nicotine patch and gum; and as a branded NicoDerm CQ nicotine patch advertisement, in which the NicoDerm CQ patch is specifically promoted as an alternative to smoking.

## Method

### SUBJECTS

#### Smoking status

Since every effort was made to have the subjects blinded to the true purpose of the study, information related to cigarette smoking was collected after the test was completed. (To ensure that exposure to the test procedures did not bias reports of smoking history collected afterward, we ran a separate control group (n = 263) that only completed the post-test questionnaire, without having seen any smoking relevant advertisement or selected smoking related prizes. Their reports of smoking history (see variables in table 1) did not differ significantly from those in the other groups), validating the post-test smoking assessment. This group did not figure in any other analyses.) Participants were asked if they currently smoke cigarettes on a regular basis; how many cigarettes they smoke per day; the type of cigarettes they smoke most often (Regular, Light, or Ultra Light); and, for those reporting that they smoke Light or Ultra Lights, if they have always smoked these brands, or switched from Regular cigarettes. (In one session, experimenter error led to these questions being omitted. In this instance, the smoking data were collected in the course of a follow up interview 72 hours later.)

Table 1

Background characteristics and smoking history of study participants overall and by experimental condition

#### Control group

In addition to the six experimental groups, a “no message” control group was also run. Subjects in this group viewed the pilot programme and other commercials and completed assessment procedures, but were not exposed to any smoking relevant advertisement. This controls for the effects of this repeated assessment in the absence of any effect of smoking related advertisements.

### DATA ANALYSIS

The logistic regression compares post-exposure selections by group. To assess directly whether changes within any given condition were systematic and significant, we used McNemar's test, applied to the pre-post transition matrix for product selection; this test assesses the significance of pre- to post-shifts in selections.10Subjects who were missing either pre- or post-choice selection data were excluded from outcome analyses, leaving a final sample of 1566.

## Results

### DEMOGRAPHICS AND SMOKING HISTORY

Table 1 shows the demographic and smoking characteristics by treatment group for all enrolled smokers. No significant differences were observed among any of the experimental or control groups on any of the characteristics of interest, whether tested on the whole sample (n = 1890) or on subjects with complete data (n = 1566).

### MOVEMENT AWAY FROM CIGARETTES

Table 2 displays the pre- and post-test selections by experimental group, and fig 1 shows the average ARS Persuasion® scores by experimental group.

Table 2

Pre- and post-test selection by experimental group

Figure 1

Average ARS Persuasion® score by experimental group for the change in the number of subjects selecting cigarettes.

Positive scores indicate elicited movement towards the selection of cigarettes after viewing the test advertisement. Negative scores represent movement away from the selection of cigarettes after viewing the test advertisement. Because of small cell counts which would violate analysis assumptions, the unbranded and branded NRT conditions were collapsed into a general “NRT ad” condition.

#### No message control group

Among those control subjects who did not view a smoking related commercial, but who did select a prize before and after viewing the pilot programme, the test procedures did not impact product choice: no significant movement either toward or away from cigarettes was observed (χ2 = 0.06, p = 0.808).

#### Main effects of message and context

Controlling for pre-test selections, subjects who saw the Feel advertisement were significantly more likely than subjects who saw the Vents advertisement to move away from cigarettes after seeing the advertisement (odds ratio (OR) = 1.97, 95% confidence interval for the odds ratio (95% CI) 1.25 to 3.09; χ2(1) = 8.69, p = 0.003). On univariate McNemar's tests within each group, only the Feel advertisement resulted in a significant change in cigarette selection (χ2 = 10.37, p = 0.001); the Vents advertisement did not (χ2 = 0.18, p = 0.668).

Controlling for pre-test selections, subjects viewing the advertisement in the PSA context tended to be more likely than subjects who saw the NRT advertisements to move away from cigarettes (OR = 1.51, 95% CI 0.94 to 2.40; χ2(1) = 2.97, p = 0.085). On univariate tests within each group, only the PSA advertisement resulted in significant change in cigarette selection (χ2 = 6.10, p = 0.014); the NRT advertisements did not (χ2 = 0.28, p = 0.599).

Because the Feel message was shown to be more effective, we also evaluated the context specifically among subjects who saw the Feel message. Within the Feel group, the PSA framing was significantly more effective in moving smokers away from cigarettes (OR = 2.08, 95% CI 1.03 to 4.20; χ2(1) = 4.18, p = 0.041).

#### Interaction effect

Controlling for pre-test selections, analyses of the interaction of message content and message context were conducted. No significant interaction was observed between message content and message context for movement towards or away from cigarettes (OR = 1.85, 95% CI 0.73 to 4.71; χ2(1) = 1.66, p = 0.197).

### MOVEMENT AWAY FROM LIGHT CIGARETTES

Figure 2 shows the average ARS Persuasion® scores by experimental group. Positive scores indicate elicited movement towards the selection of Light cigarettes after viewing the test advertisement. Negative scores represent movement away from the selection of Light cigarettes after viewing the test advertisement.

Figure 2

Average ARS Persuasion® score by experimental group for the change in the number of subjects selecting Light cigarettes.

#### No message control group.

There was no significant change in the selection of Light cigarettes among those subjects who completed two successive product selections without seeing any smoking related advertisement (χ2 = 0.11, p = 0.739).

#### Main effects of message and context

Controlling for pre-test selection, there was no significant effect of advertisement content on moving smokers away from Light cigarettes (OR = 1.40, 95% CI 0.79 to 2.50; χ2(1) = 1.33, p = 0.249). However, on univariate tests within each group, only the Feel message resulted in a trend of movement away from Light cigarettes (χ2 = 3.12, p = 0.077); the Vents message did not (χ2 = 0.15, p = 0.695).

Controlling for pre-test selections, subjects who saw the PSA advertisements were significantly more likely to move away from Light cigarettes (OR = 2.45, 95% CI 1.28 to 4.69; χ2(1) = 7.30, p = 0.007). On univariate tests within each group, only the PSA advertisement resulted in significant change in Light cigarette selection (χ2 = 8.33, p = 0.004); the NRT advertisement did not (χ2 = 0.29, p = 0.590).

Because the Feel message was most effective overall, we tested the effect of context within that message. The analysis demonstrated that the PSA framing was the most effective context for the Feel message in terms of moving smokers away from Light cigarettes (OR = 2.91, 95% CI 1.15 to 7.33; χ2(1) = 5.13, p =0 .023).

#### Interaction effect

No significant interaction was observed between message content and message context for movement towards or away from Light cigarettes (OR = 1.65, 95% CI 0.48 to 5.73; χ2(1) = 0.63, p = 0.428).

### MOVEMENT TOWARD TREATMENT

Figure 3 shows the average ARS Persuasion® scores by experimental group. Positive scores indicate elicited movement towards the selection of assisted quitting products after viewing the test advertisement. Negative scores represent movement away from the selection of assisted quitting products after viewing the test advertisement.

Figure 3

Average ARS Persuasion® score by experimental group for the change in the number of subjects selecting treatment items.

#### No message control group

Subjects who completed two successive product selections without seeing any smoking related advertisement showed significant movement away from selection of treatment products (χ2 = 26.47, p < 0.001), with an ARS Persuasion® score of −12.82%. This suggests that the test procedures themselves (that is, viewing the show and ads and repeated product selection) yielded artifactual changes in selection of smoking cessation aids. This substantially compromises the validity of this measure as an index of advertisement effects on selection of treatment products.

### MODERATORS

We tested for interactions between the message content and demographic characteristics, to examine whether the feel message was more or less effective with particular subgroups. We found no interaction between message content and sex, smoking rate, or type of cigarette smoked (that is, Regular, Light, or Ultra Light). We also tested whether the two messages had differential effects on young adult smokers (18–25 years old) versus older smokers (> 25 years); there were no differences.

## Discussion

These results, along with those of the previous test of these message strategies,5 validate the importance of addressing smokers' somatic perceptions of “lightness” when smoking reduced yield cigarettes. Because such sensory experiences fit into smokers' naive models of smoking related harm, addressing these experiences may be important in changing beliefs and behaviour.6 7

It was striking that even though the advertisements reduced interest in cigarettes, they appeared to reduce the number of smokers who chose treatment related products. Unfortunately, analysis of a control group that saw no smoking relevant advertisements showed that even they significantly reduced their selection of treatment products. It appears that simply being presented with the product choices twice caused smokers to reduce their preferences for treatment products. Subjects did not actually receive their selected prizes, so their need for treatment products was not saturated after the pre-test. We speculate that those who selected treatment products on the first round may have experienced the second offer as pressure to quit, perhaps eliciting reactance and thus apparent movement away from treatment products. In any case, this result precluded complete evaluation of the effect of the test ads on movement towards assisted quitting. We did find, however, that the Feel ad yielded less movement away from assisted quitting products than was seen in the control group, suggesting that this message not only encourages movement away from cigarettes but may also prompt movement towards assisted quitting.

While changing people's beliefs and attitudes about Light and Ultra Light cigarettes is an important goal for a counter-marketing campaign (and prerequisite to behaviour change), behaviour change is ultimately the desired outcome and the ultimate measure of success. In previous studies examining the impact of messages designed to counter-market Light cigarettes, smokers were exposed to the message of interest within the context of a survey that assessed smoking history and examined beliefs and attitudes regarding Light and Ultra Light cigarettes. The context called respondents' attention to their smoking and was clearly one of a persuasion attempt, thus increasing the likelihood that participants' responses and reactions to the message may have been influenced by demand characteristics. Subjects in the present study, on the other hand, were unaware of the true nature or purpose of the study. The advertisement of interest was camouflaged among other advertisements and a television pilot programme. Subjects for this study were not specifically recruited on the basis of their smoking status, and smoking status data were collected only after subjects had completed the test. Thus, demand characteristics should have been minimised.

The limitations of the study include the lack of strict random assignment. Because of the way ARS Persuasion® tests are conducted, including their administration in large groups at geographically dispersed locations, random assignment was not feasible. However, care was taken that assignment to treatment was not systematic, but depended on subjects' convenience. The lack of group differences on demographic and smoking variables confirms the balanced assignment of subjects to groups. Further, in scheduling the test sessions, care was taken to mix weekends and weekdays, geographical sites, and the other commercials viewed during the test, to assure unbiased comparisons across groups. Another limitation was that the six test commercials were not strictly identical on other parameters such as visual imagery, language, number of words, etc. It is possible that these distinctions, and not the message theme itself, may in fact be responsible for the differential effects observed. However, the ads were very simple and quite similar. Also, the consistency between the findings of this study and those of our prior comparison, which used different message content and medium, but yielded parallel results, lends credence to the current findings.

In conclusion, this research builds upon prior studies to demonstrate that a particular message strategy is most successful for counter-advertising on the subject of Light and Ultra Light cigarettes. Addressing smokers' sensory perceptions of Light cigarettes and presenting this information in an impartial way is likely to be an effective communication strategy for counter-marketing Light cigarettes. This message and media approach can provide a clear strategy for public media campaigns against smoking.

## Acknowledgments

This research was funded by GlaxoSmithKline. The authors not employed by GlaxoSmithKline serve as consultants in the areas of smoking cessation and tobacco control exclusively for GlaxoSmithKline, and received compensation for their participation in this research. Mr Burton is Head of Smoking Control, Strategic Development, and Medical Promotions for GlaxoSmithKline. Mr Wardle is Senior Brand Equity Manager, New Smoking Control Products for GlaxoSmithKline. Mr. Koehler is Senior Category Research Manager for GlaxoSmithKline. The authors are grateful for the collaboration of rsc THE QUALITY MEASURMENT COMPANY in using the ARS Persuasion® test for research purposes.

View Abstract

## 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.