Article Text

Download PDFPDF

The silent salesman: an observational study of personal tobacco pack display at outdoor café strips in Australia
  1. Melanie A Wakefield1,
  2. Meghan Zacher1,
  3. Megan Bayly1,
  4. Emily Brennan2,
  5. Joanne Dono3,
  6. Caroline Miller4,
  7. Sarah J Durkin1,
  8. Michelle M Scollo1
  1. 1Centre for Behavioural Research in Cancer, Cancer Council Victoria, Carlton, Victoria, Australia
  2. 2Annenberg School for Communication, University of Pennsylvania, Pennsylvania, USA
  3. 3Cancer Council South Australia, Eastwood, South Australia, Australia
  4. 4South Australian Health and Medical Research Institute and Discipline of Public Health, University of Adelaide, Adelaide, South Australia, Australia
  1. Correspondence to Professor Melanie A Wakefield, Centre for Behavioural Research in Cancer, Cancer Council Victoria, 1 Rathdowne Street, Carlton, Victoria 3053, Australia; melanie.wakefield{at}


Objective We sought to determine the relative frequency and nature of personal display of cigarette packs by smokers in two Australian cities where 30% front-of-pack and 90% back-of-pack health warnings have been used since 2006 and comprehensive tobacco marketing restrictions apply.

Methods An observational study counted patrons, active smokers and tobacco packs at cafés, restaurants and bars with outdoor seating. Pack orientation and use of cigarette cases were also noted.

Results Overall, 18 954 patrons, 1576 active smokers and 2153 packs were observed, meaning that one out of every 12.0 patrons was actively smoking, and one of every 8.8 patrons displayed a pack. Packs were more frequently observed in lower socio-economic neighbourhoods, reflecting the higher prevalence of smoking in those regions. Packs were displayed less often in venues where children were present, suggesting a greater tendency not to smoke around children. Most packs (81.4%) were oriented face-up, permitting prominent brand display. Only 1.5% of observed packs were cigarette cases, and 4.2% of packs were concealed by another item, such as a phone or wallet.

Conclusions Tobacco packs are frequently seen on table-tops in café strips, providing many opportunities for other patrons and passers-by to be incidentally exposed to cigarette brand names and imagery. Use of cigarette cases is rare, suggesting that smokers eventually habituate to pictorial warnings on branded packs and/or find repeated decanting of each newly purchased branded pack into a case to be inconvenient.

  • Advertising and Promotion
  • Packaging and Labelling
  • Surveillance and monitoring

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 nations progress towards implementing comprehensive restrictions on tobacco marketing, including those required as part of Article 13 of WHO Framework Convention on Tobacco Control,1 tobacco companies have increasingly turned to cigarette packaging as a vehicle for brand communication.2–4 Unlike most other consumer goods, cigarette packs remain with consumers once purchased, and are taken out and used or displayed many times each day, often in social situations. This high degree of social visibility leads cigarettes to be known as ‘badge products’, associating the user with the brand's image2 ,5 and allowing young people to use cigarettes to help establish their social identity.6 In fact, cigarette packaging has come to be described as the ‘silent salesman’ of integrated tobacco marketing.7

In the broader literature, personal display of brands is recognised as an important aspect of marketing. In the course of daily encounters with other people, an individual is frequently exposed in an incidental fashion to various brands. Even though these encounters may be brief and the brand may not be the focal point of the encounter, the ubiquitous and pervasive nature of exposure means that these so-called ‘incidental consumer brand encounters’ (ICBEs) are an extremely common form of exposure to chosen brands.8 Furthermore, ICBEs generally occur below the level of conscious awareness, subtly influencing perceptions about the frequency of use of particular brands, as well as attitudes towards brands through their association with different types of consumers. For example, Bargh9 and Dijksterhuis et al10 argue that much consumer behaviour is in fact influenced by subtle environmental cues that activate cognitive and affective processes below the level of awareness. In addition, a series of four experimental studies by Ferraro et al8 found that exposure to simulated ICBEs increased the choice of the displayed brand among people not overtly aware of the brand exposure. A process of perceptual fluency, moderated by consumers’ automatic responses to the type of user observed with the brand, was found to underlie these effects, meaning that repeated exposure to the brand facilitated processing of subsequent exposures, thereby increasing the relative ease with which consumers could identify, and therefore select, the brand.8

Although incidental brand exposure is known to be important in influencing brand choices and perceptions, to our knowledge there have been no studies that have assessed the frequency of incidental personal display of cigarette packs. Some tobacco control policies might be expected to reduce the likelihood of personal pack display. For example, 9 months after the introduction of pictorial warnings on Canadian tobacco packs, 11% of smokers reported at least sometimes putting their pack away because they did not want others to see the warning and 24% reported at least once either having placed a cardboard sleeve over their pack to conceal the warning or having transferred their cigarettes to another container.11 Little is known about the extent to which such discomfort persists over the years as smokers become accustomed to the warnings. Since 2006, 14 different pictorial health warnings (two annually-rotating sets of seven) have been used on packs in Australia. These pictorial warnings cover 30% of the front of the pack, while 90% of the back of the pack contains the pictorial warning as well as text providing further explanation about the health condition and the Quitline number.12 ,13

In an attempt to quantify the frequency of incidental exposure to branded cigarette packs, the objective of this study was to measure the relative frequency and nature of personal pack display at outdoor venues some 5 years after pictorial health warnings had been implemented.


We selected a convenience sample of 18 café strips in the suburbs of Melbourne in the state of Victoria and seven in suburbs of Adelaide in South Australia, based on our observation that they contained a high number of eligible venues and high patronage in a concentrated location, while ensuring a diverse range of socio-economic status (SES) areas were sampled. These two cities were selected because they were located in states without outdoor smoking restrictions.

Eligible venues within café strips included cafés, restaurants, bars and pubs with outdoor tables visible from the footpath. Take-away food, fast food, juice bars and ice-cream shops were not eligible for sample inclusion because such venues do not generally encourage patrons to remain seated at the venue for an extended period of time (thus, patrons are less likely to smoke while at the venue). Fieldworkers were instructed to select a starting location, and to walk along the busiest portion of the strip, observing every sample-eligible venue they passed. If the fieldworker saw other venues located on side-streets, they sampled these as well. Once fieldworkers had selected venues, they revisited the same venues every 2 weeks. For venues that ceased operating over the study period, the subsequent waves contain missing data for those venues. Venues that were closed at the time of data collection were considered missing for that wave (ie, the fieldworker did not return to the venue at a later date within the wave).

Beginning in the second half of October 2011 and ending mid-April 2012 (late Spring to mid-Autumn), fieldworkers completed nine observations in their assigned café strips (in South Australia, fieldworkers began data collection at the beginning of November, and submitted only eight sets of observations). Fieldworkers conducted observations every 2 weeks, avoiding days on which there were public holidays, large public events, the end-of-year period and very poor weather, when patronage would have been atypical. Fieldworkers could either conduct observations on any weekday or either weekend day. Most fieldworkers observed their venues on the same day of the week and at similar times of day for each wave, and all observations were completed during daylight hours. Fieldworkers selected their observation times based on local knowledge of when their areas were likely to be busy with patrons.

At each café strip, fieldworkers walked along the footpath. Using notepads or Smartphone data collection applications, at each venue, they discreetly noted: (1) the number of seated patrons (including children); (2) whether children of approximately primary school age or younger were present; (3) the number of people smoking or holding/rolling/lighting a cigarette (defined as ‘active smokers’); (4) the total number of cigarette packages (for manufactured cigarettes or roll-your-own tobacco) visible on the table; and (5) the number of cigarette packages oriented (5a) face-down; (5b) standing on their side, top or bottom; (5c) in a case or tin (not the branded original packaging); (5d) completely concealed by a phone, wallet or some other object, so that the fieldworker was unable to ascertain its orientation; and (5e) with an unknown orientation (ie, too far away). Fieldworkers also recorded temperature (in degrees Celsius) and wind speed (in kilometres per hour) for their city at the time of their observations (from

Statistical analysis

We used the Socioeconomic Index for Areas Index of Relative Disadvantage state rankings to determine SES of each postcode.14 Café strips from postcodes from the lower two quintiles of the index make up the low-SES sample, those from the third and fourth quintiles make up the mid-SES sample, and those from the fifth quintile are the high-SES areas. For analysis, we combined observations from October and November (Spring), December and January (early Summer), February (late Summer) and March and April (Autumn). Observations conducted on early weekdays (Mondays, Tuesdays and Wednesdays) were compared with those conducted on late weekdays (Thursdays and Fridays) and weekends. Temperature and wind speed were analysed as continuous variables.

The observational method described above does not permit analysis of individuals, as we have no information regarding which individuals were smoking and which possessed and displayed a cigarette package. Instead, venues are used as the unit of analysis. Multilevel Poisson models were employed in Stata 11.215 to regress the number of packs seen at a venue during a particular wave on a set of potential covariates. The count of patrons in the venue was used as the offset term, which allowed the data to be analysed as the rate of packs to patrons. Only venues for which at least one patron was present during that particular wave were analysed. Venue and café strip were included as random effects in all models. This method effectively adjusts SEs for correlations among observations within the same venue and among venues within the same café strip. The modelling procedure first ran a set of univariate models, followed by a full multivariate model with all covariates, and then eliminated covariates with significance levels above p=0.150 to yield a final multivariate model. We also used multilevel Poisson models to regress the number of packs oriented face-up and the number of cigarette cases on the same potential covariates. In these models, the number of packs observed was the offset term (and only observations where at least one pack was seen are analysed), allowing analyses of the rate of face-up packs and cigarette cases to packs displayed.


In total, 453 unique venues were observed over the study period, 347 in Victoria and 106 in South Australia. Since Victorian venues were potentially observed nine times each and South Australian venues eight times each, there were 3123 total possible observations from Victoria (347×9) and 848 from South Australia (106×8), making a total of 3971 possible observations. Of the 453 venues observed, 46 never had patrons present and could not be included in the analysis, leaving 407 eligible venues in total. From these 407 eligible venues, 2391 venue×wave observations are included in the analysis, as venues were required to be open for business and to have at least one patron present during the wave. Thus, on average, each venue had 5.9 actual observations (‘cases’) in the analysis (ranging from 1 to 9 observations from each venue).

Pack display

Over the study period, a total of 18 954 patrons were observed. Of these, 1576 (8.3%) were ‘active smokers’ at the time of observation. A total of 2153 packs were observed; this is approximately one pack displayed for every 8.8 patrons observed, or 1.4 packs for each active smoker. Since there were 1576 patrons seen actively smoking and 2153 packs observed, at least 577 packs (26.8%) would have belonged to patrons who were not actively smoking, suggesting that some smokers leave their packs on the table between cigarettes. Of the 2153 packs observed, 472 (21.9%) were observed at venues with no active smokers.

Table 1 presents descriptive information on the number of observations obtained for each covariate category (thus, this represents the total number of cases to be analysed within each category). Table 1 details the percentage of patrons actively smoking for each covariate category, as well as the percentage of patrons with a tobacco pack displayed on the table. Higher percentages of active smokers corresponded to higher percentages of patrons with packs displayed, though the percentages of patrons with a pack displayed was always higher than the percentages who were smoking.

Table 1

Descriptive statistics for eligible venue×wave observations (cases)

The univariate models in table 2 show that without controlling for other variables, rates of packs to patrons were significantly higher in Victoria than in South Australia. Also, low-SES areas had higher rates than mid-SES areas. Packs were displayed at a lower rate at venues where children were present compared with venues where there were no children. Pack display did not differ significantly between months or for different times of the day. However, late weekdays and weekend days had significantly higher rates of pack display than early weekdays. Neither temperature nor wind speed affected pack display.

Table 2

Results from multilevel Poisson regression models examining rates of tobacco packages to patrons: incidence rate ratios (IRR) and significance tests (p values)

The preliminary multivariate model including all variables (model 1) produced similar results, with some noteworthy changes. When controlling for other covariates, Victoria's rate of packs to patrons was no longer significantly higher than that found in South Australia, and the main effect for SES decreased in significance. The presence of children continued to be associated with lower rates of pack display, while observations on late weekdays and weekends were again associated with significantly higher observed rates of packs to patrons. In the final multivariate model (model 2), all effects either remained the same or strengthened as those described for model 1. Venues in mid- and high-SES areas had substantially lower rates of pack display than those in low-SES areas. When children were present, the number of packs on display was significantly lower than when there were no children present, and late weekdays and weekends had significantly higher rates of pack display than early weekdays.

To further explore possible reasons for the lower rate of pack display when children were present, we conducted an analysis of the subset of 831 venue×wave observations (out of the total of 2391 cases) where at least one active smoker was present. In this analysis, the rate of pack display to active smokers did not differ in venues with children compared with those with no children present (multivariate incidence rate ratio (IRR)=1.028, p=0.703), suggesting that the lower rates of packs to patrons in venues with children were due to the fact that there were fewer patrons smoking. Indeed, in further analysis, we found the rate of active smoking to patrons in venues with children were 24.3% lower than in venues without children (multivariate IRR=0.757, p=0.001, n=2391).

Pack orientation

Of the 2153 packs observed over the course of the study, 81.4% were oriented face-up, 8.0% were oriented face-down, 1.7% were standing upright or lying on their side, and 4.2% were concealed by phones, wallets or other objects. Fieldworkers could not determine the orientation of another 3.2% of packs. The remaining 1.5% of packs were cigarette cases.

We tested the rate of face-up packs and cigarette case use to total packs observed in venues with at least one pack displayed (n=1015). In the full multivariate model regressing the rate of face-up packs to all packs displayed in a venue, no covariates had significance levels for main effects below p=0.150. Thus, a reduced multivariate model was not created.

When testing the rate of cigarette case use among packs, only SES (p=0.141) was eligible for inclusion in the reduced model. In the reduced final model, cigarette case use rates were lower in high-SES areas than those in low-SES areas (IRR=0.343, p=0.050).


This study indicates that branded cigarette packs are commonly on display where patrons are seated at venues with outdoor seating, with one pack observed for every 8.8 patrons. This frequency of incidental exposure to branded cigarette packs for other patrons and passers-by might be considered surprisingly high in light of Australia's comprehensive restrictions on tobacco marketing in traditional channels, including bans on the retail display of cigarette packs in all jurisdictions. Incidental brand impressions serve to undermine the purpose of traditional marketing restrictions and illustrate the need for plain packaging of tobacco. A large body of literature has demonstrated that the use of colour, varied pack shapes and different design elements serve to make packs more appealing to potential consumers (eg16 ,17) and that plain packaging reduces brand appeal.18 ,19 Our findings underline the need for plain packaging to reduce the frequency of incidental brand exposure.

Observed rates of active smoking were lower than adult smoking prevalence rates in these two states,20 indicating that smokers are not always smoking or preparing to smoke when sitting at outdoor cafés. On the other hand, rates of pack display were consistently higher than rates of active smoking, indicating that many smokers leave their cigarette packs on display even when not engaged in the act of smoking, and this was the case for at least 26.8% of packs. Consistent with higher smoking prevalence in lower SES areas, the pack to patron ratio was higher in low-SES areas than in mid- or high-SES areas. In addition, when children of primary school age or younger were present, the rate of packs to patrons decreased significantly. Our additional analysis suggested that this was mostly driven by the fact that when children were present, fewer patrons were actively smoking than when children were not present. Some of this difference is likely to be due to the fact that the children themselves (who were included in the count of patrons) were not smoking. However, fieldworkers agreed that children typically comprised much less than a quarter of the patrons observed at any venue. Thus, some of the reduced rate of smoking may be due to social norms restricting people from smoking when children are present, or due to norms within child-friendly venues which discourage patrons from smoking. Packs were observed at higher relative frequencies on late weekdays and weekends compared with early weekdays. It is possible that more social events are likely on these days when people may feel more able to smoke freely, stay longer at venues and to leave their packs on display.

The majority of packs observed were oriented face-up, thereby prominently displaying branding and concealing the larger graphic health warnings on the back of the packs. It is notable that the tobacco industry argue most vehemently against placement of health warnings on the front of packs,21 ,22 suggesting a recognition that this is the pack face most noticed or valued by smokers and potential consumers. Our observational study confirms the importance of the front of pack face for incidental brand display.

In addition, despite concerns from some that graphic health warnings would prompt smokers to use decorative cigarette cases more often,23 ,24 it appears that this was extremely uncommon 5 years after the 2006 Australian legislation was implemented, as only 1.5% of the 2153 observed packs were external cigarette cases. Covering of packs by wallets, phones or other objects was also infrequent. Surveys between 2 and 9 months after implementation of pictorial warnings have found between 6% and 21% of smokers to report using a different cigarette container at least once as a result of the warnings (Victorian Smoking and Health population survey, Cancer Council Victoria, unpublished data, 2006).25 ,26 New warnings may have prompted trial use of cases, but it appears that this has not persisted, perhaps because continually having to place each new pack of cigarettes in a case is inconvenient and/or because after 5 years, smokers have become habituated to the warnings. Interestingly, several countries including Australia with its soon-to-be implemented plain packaging law have banned cigarette sleeves, tins or cases that feature tobacco branding.27 Other countries such as India, Brazil and Spain have banned decorative cigarette pack sleeves that conceal health warnings (Rob Cunningham, personal communication, 21 June 2012), and yet others such as Madagascar28 and Djibouti29 have banned cigarettes cases of any description.

Our study method has strengths and limitations. First, our observational method is likely to have missed many smokers if they were not smoking at the precise time that the fieldworker walked slowly by. However, our method of assessing the ratio of observed packs to patrons gave an overall metric of the relative frequency of personal pack display and, importantly, we found this measure to vary predictably by SES and the presence of children, thereby providing validity. These two factors, as well as day of week, would be important to measure and adjust for in any observational effort to monitor policy-related changes in personal pack display. Our observational method afforded an advantage in quantifying brand exposure compared with population surveys that could yield data on reported exposure of respondents to cigarette packs in public places. However, such self-report measures would only measure subjective exposure rates among those respondents who consciously took note of such packs, a rate likely to be much lower than actual exposure rates. By contrast, our observational method affords comprehensiveness by using a more objective method that takes into account the tendency for individuals to be influenced below the level of conscious awareness by stimuli such as branded packaging.8–10

In summary, branded tobacco packaging is commonly on display at cafés, restaurants, bars and pubs with outdoor seating. This is a concerning finding given the established link between tobacco marketing and favourable perceptions of tobacco.30 The introduction of plain packaging in Australia in December 201227 is expected to reduce the extent to which personal pack display in social environments can easily communicate branding information.

Key messages

  • In an observational study of outdoor café strips in two Australian cities, one out of every 8.8 patrons displayed a branded tobacco pack on their table—with most being face-up—providing many opportunities for others to be incidentally exposed to cigarette brand names and imagery.

  • Use of cigarette cases was rare (<2%), suggesting that smokers eventually habituate to pictorial warnings on branded packs and/or find repeated decanting of each newly purchased branded pack into a case to be inconvenient.


We thank the staff of Cancer Council Victoria and Cancer Council South Australia who undertook data collection.


View Abstract


  • Contributors MAW and MS conceived the study, MB, EB and SD improved its design, MZ, MB, JD and CM coordinated aspects of data collection, MZ and SD undertook data analysis, MAW and MZ drafted the manuscript; all authors revised the manuscript and have read and approved the final version.

  • Funding This study was funded by Cancer Council Victoria and Cancer Council South Australia.

  • Competing interests None.

  • Ethics approval This was an unobtrusive observational study counting café patrons and cigarette packs and was conducted in a public setting.

  • Provenance and peer review Not commissioned; externally peer reviewed.