Article Text

Download PDFPDF

Measuring the effect of cigarette plain packaging on transaction times and selection errors in a simulation experiment
  1. Owen B J Carter,
  2. Brennen W Mills,
  3. Tina Phan,
  4. Jonathon R Bremner
  1. Centre for Behavioural Research in Cancer Control, School of Public Health, Faculty of Health Sciences, Curtin University, Perth, Western Australia, Australia
  1. Correspondence to Dr Owen B J Carter, Centre for Behavioural Research in Cancer Control, Curtin University of Technology, Perth, WA 6845, Australia; o.carter{at}curtin.edu.au

Abstract

Introduction Australia has introduced legislation to force all cigarette packaging to be generic from 2012 onwards. The tobacco retail industry estimates this will result in transaction times increasing by 15–45 s per pack and is spending at least $A10 million of tobacco industry funds on an advertising campaigns claiming that the increased time and errors associated with plain packaging will ultimately cost small businesses $A461 million per annum and endanger 15 000 jobs. We undertook an objective experiment to test these claims.

Methodology Participants (n=52) were randomly assigned to stand in front of a display of either 50 plain or coloured cigarette packets and then were read a randomly ordered list of cigarette brands. The time participants took to locate each packet was recorded and all selection errors were noted. After 50 ‘transactions’, participants repeated the entire experiment with the alternative plain/coloured packs. Afterwards, participants were asked in an open-ended manner whether plain or coloured packaging was easier to locate and why.

Results The average transaction was significantly quicker for plain compared with coloured packs (2.92 vs 3.17 s; p=0.040). One or more mistakes were made by 40.4% of participants when selecting coloured packaging compared with only 17.3% for plain packaging (p=0.011). Qualitative results suggested that the colours and inconsistent location of brand names often served to distract when participants scanned for brands.

Conclusion Rather than plain packaging requiring an additional 45 s per transaction, our results suggest that it will, if anything, modestly decrease transaction times and selection errors.

  • Packaging and labelling
  • nicotine products
  • young adults
  • advertising and promotion
  • denormalisation
  • media
  • packaging and labelling
  • social marketing
  • smoking-caused disease
  • advertising and promotion

Statistics from Altmetric.com

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.

On 5 July 2011, a bill was introduced to the Australian parliament to become the first country in the world to have mandatory plain packaging for all tobacco products. This would see all distinguishing features removed from tobacco packs, including colours, logos, images, descriptors and fonts, to be replaced by uniform monochrome boxes of a standard cardboard texture, inscribed with a simple brand name and quantity written in a standard font, while retaining current colour graphic health warnings on the majority of the front and back of each pack.1

The tobacco industry is fighting the introduction of this legislation with every means at its disposal. On 27 June 2011, Philip Morris Asia launched a legal challenge to the legislation as being in violation of the bilateral investment treaty between Australia and Hong Kong, even before the bill was introduced.2 British American Tobacco has stated its intention to also mount a legal challenge and in the interim has been running a multimedia campaign entitled ‘Don’t let the taxpayer foot the bill for a bad Bill', arguing that the Australian government will have to spend millions of taxpayers' dollars in legal fees defending the legislation and potentially billions in compensation to the tobacco industry for loss of intellectual property.3 Meanwhile, Imperial Tobacco has been running its ‘Stop This Nanny State’ multimedia campaign claiming that the legislation ‘goes too far’, violating personal freedoms of choice and urges voters to complain to their members of parliament.4

However, due to the tobacco industry having very little public credibility, it has a long pedigree of also funding ‘small business’ organisations around the world to serve as its mouthpiece in order to politically agitate against tobacco control legislation. For instance, the attempted introduction of point-of-sale tobacco display restrictions has been opposed by the Association of Convenience Stores in the UK, Retailers Against Smuggling in Ireland, the Canadian Convenience Stores Association and the New Zealand Association of Convenience Stores, all receiving generous funding from the tobacco industry to run media campaigns opposing tobacco control legislation. Characteristically, these campaigns raise concerns about revenue and job losses should the proposed tobacco control legislation ever be enacted.

Thus, in the face of plain tobacco packaging legislation in Australia, the tobacco industry has jointly funded the creation of an umbrella organisation called the Alliance of Australian Retailers (AAR), representing the Australian Newsagents Federation, the Service Station Association and the National Independent Retailers Association. During the lead up to the Australian federal election in October 2010, British American Tobacco, Philip Morris and Imperial Tobacco funded the AAR to the tune of $A5 million (GB£3.25 million/US$5.27 million) to run a vigorous mass-media campaign opposing the introduction of plain tobacco packaging.5 The AAR ran a second wave of advertising in March 2011 of a similar scope without publically revealing how much was spent the second time around. However, all indications are that at least the same amount has been spent again, suggesting around $10 million has been spent to date.

The AAR campaign has included television, radio, newspaper and poster advertising with the catch phrases ‘Good policy requires more than good intentions’ and ‘It won’t work, so why do it?' Figure 1 is one such example of an AAR newspaper advertisement. An example of a media release accompanying the campaign by the Australian Newsagents Federation states: “The removal of branding and making all cigarette packets look the same will make the life of retailers and shoppers hell as queues lengthen, creating confusion and other delays adding to the frustration of shoppers.”6

Figure 1

Example of an Alliance of Australian Retailers advertisement.

Similarly, a letter from the Service Station Association to the Senate Standing Committee on Community Affairs states: “For tobacco products, console operators need to be able to quickly identify a customer's request for a specific product and they do that by sighting the unique branding on the package. Without the unique packaging, console operators would only have stereotype words to help them locate the product of choice, which will take considerably longer and lead to a high error rate.”7

The AAR commissioned Deloitte to quantify the impact, and based upon this (non-peer reviewed), research suggests that plain packaging will increase each transaction by up to 45 s (range 15–45 s) incurring an annual cost to small business tobacco retailers of A$461 million (£299 million/US$486 million) and 15 000 jobs.8 ,9

Given the vigour with which tobacco retailers have made these claims and the likelihood that other ‘small business’ organisations around the world will make similar future claims when plain packaging legislation is considered in their own countries, we sought to empirically test the veracity of the claims. We therefore designed an experiment to determine exactly how much longer each transaction is likely to be and how many more errors are likely to be made with the introduction of plain packaging for tobacco products.

Method

Approval for the study was gained from the Curtin University Ethics Committee.

Participants

Participants were randomly approached on a university campus over a 1-week data collection period and asked to participate in a brief experiment. A sampling quota was set to ensure equal sex representation. In order to reduce confounding from prior familiarity with brand logos and colours, smokers and those with previous experience in selling tobacco were excluded from participation. Those not fluent in English were also excluded. In total, 52 participants were recruited, comprising 26 men and 26 women, with an average age of 21.8 years (SD=5.2; range 17–44).

Materials

A list of 50 transactions was compiled, with frequencies assigned to various cigarette brands on the list, based upon market share.10 In this way, the most popular brand in Australia with a 32% market share was assigned 16 transactions, the second most popular with a 24% share was assigned 12 transactions and so on, with transactions being rounded to the nearest whole number (brands with market shares <1% were rounded to zero) (see table 1). This resulted in a transaction list of the 19 most popular cigarette brands, ranging in frequency from one to eight transactions each that would represent realistic daily retail demand. A tobacco vendor with 6 years' retail experience then independently checked the list to confirm its commercial face validity. Each brand on the transaction list was assigned a number from 1 to 50. These numbers were then entered into a random sequence generator (http://www.random.org) to create 10 differently ordered transaction lists.

Table 1

Number of transactions per cigarette brand

For each brand and its sub-brands, mock cigarette packs were constructed in cardboard from a standardised cigarette pack template; one in full colour and one in plain packaging (see figure 2). Colour packs were constructed of white cardboard covered with printed cigarette packet images (acquired from http://www.tobacco.net.au). The plain packs were constructed in a dull brown coloured cardboard with the brand name and quantity of cigarettes printed in a standard, black Arial Rounded font size 22 on the bottom half of the face. In order to reduce confounding from health warnings, none were featured on either style of packaging. Two shelves were constructed to each hold 50 cigarette packs with faces fully displayed, arranged 10 across by five down. The 50 coloured packs were placed on one shelf and the plain packs on the other. Cigarette packs were arranged in a standardised alphabetical order, based upon main brand and then sub-brand name.

Figure 2

Examples of the coloured and plain packs used in the experiment.

Procedure

Participants were randomly assigned by the toss of a coin to start with either the coloured or plain packs. Participants were placed in front of an empty table at approximately waist height. A second table of equal height was positioned behind participants upon which one of the shelves of 50 cigarette packets was placed. Experimenters randomly selected one of the 10 transaction lists and proceeded to call out cigarette brand names one at a time. Participants were instructed to turn around, select the appropriate cigarette pack from those on display and then turn back and place it on the empty table as quickly as possible. Using a stopwatch, a second experimenter timed each transaction in split seconds (to two decimal places). The second experimenter had his or her back to the participant so as to be blinded to the coloured or plain packaging condition. The second experimenter began timing as soon as the first experimenter had finished calling out a brand name and stopped as soon as he or she heard the participants put a cigarette pack on the empty table—the packs were hollow and so produced an audible sound when placed on the table. The time of each transaction was called out by the second experimenter, and this was recorded by the first experimenter, who also made note of incorrect brands selected by the participant. Once the participant had completed 50 transactions, the shelf of coloured or plain cigarette packs was then substituted for its alternative. The first experimenter then randomly selected another transaction list from the remaining nine and repeated the entire experiment. At the conclusion of the second 50 transactions, participants were then asked in an open-ended manner from which type of packaging they found it easier to select and why. Responses were recorded verbatim. Participants were then debriefed and allowed to leave. The entire procedure took approximately 15 min per participant. In total, our procedure captured 5200 cigarette pack ‘transactions’.

Results

Transaction time

Average times per transaction for coloured and plain packs are displayed in figure 3. This describes the average transaction time for coloured packs being greater than plain packs, but with this difference diminishing over time such that by the end of 100 transactions, the average transaction times for both pack types had converged at around 2.5 s each.

Figure 3

Comparison of mean transaction times for plain versus coloured cigarette packaging.

To simulate the likely effect of transitioning from coloured to plain packs that will occur, the results of only that half of participants who undertook the first trial with coloured packs and the second trial with plain packs were compared. A paired samples t test, comparing participants' first 50 transactions with coloured packs compared with their second 50 transactions with plain packs, suggested transaction times decreased for plain packaging by a statistically significant average of 0.47 s per transaction (t(25)=7.783, p<0.001) (see table 2).

Table 2

Average transaction times for coloured versus plain packs with evidence of a training effect and selection errors

It is evident from figure 3 that there was a training effect, with participants becoming progressively quicker per transaction over the course of the experiment, regardless of packaging type. A visual inspection of figure 3 suggests that most of this training occurred within the first 25 transactions, after which time improvement continued but at a more gradual pace. A paired samples t test confirms that, regardless of packaging type, the average time for the first 25 transactions took 0.82 s longer than for the second 25 transactions (t(51)=7.539, p<0.001). A comparison of the first 50 transactions to the second 50 transactions, also regardless of packaging type, suggests a smaller but still statistically significant difference of 0.46 s (t(51)=10.131, p<0.001) (see table 2).

In order to control for the training effects, two further analyses of the data were undertaken. First, participants’ first 50 transactions for either plain or coloured packaging were compared, revealing that the average transaction time for colour packs remained 0.44 s longer than plain packs (t(33.513)=2.167, p=0.037). The average transaction time for each package type was then combined for all participants, regardless whether it came from their first or second 50 transactions. This confirmed that the average transaction time per colour pack remained 0.25 s longer than per plain pack (t(51)=2.103, p=0.040).

Selection errors

Participant selection errors were rare for either packaging type, representing only 30 of 5200 observed transactions (0.6%). Even still, significantly more selection errors were made for coloured packs (1.5% of all transactions) compared with plain packs (0.4% of all transactions) (Wilcoxon signed ranks test z=2.558, p=0.011). Although rare, in terms of proportions, over twice as many participants made one or more selection errors with colour packs (40.4%) compared with plain packs (17.3%).

‘Easier’ packaging type

When asked “From which packaging type did you find it easier to select packs?”, there were 27 participants (51.9%) who nominated plain packs, 23 (44.2%) who nominated coloured packs and two who remained undecided (3.8%). A χ2 analysis suggested no significant difference in preferences (χ2 (1)=0.320, p=0.572), but there was a tendency for participants to prefer whichever packaging type they used in their second trial compared with their first trial (62.0% vs 38.0%, respectively), albeit this difference only approached statistical significance (χ2 (1)=2.880, p=0.090). There was also a tendency for participants who trialled plain packs second to prefer it compared with those who trialled coloured packs second (66.7% vs 57.7%, respectively). Similarly, this difference approached but did not achieve statistical significance (χ2 (1)=2.981, p=0.084).

Thematic analysis

Participants' open-ended discussions about which type of packaging was easier to use were clustered into recurring themes, from which four consistent topics emerged. The most common topic, discussed by 32 participants (61.5%), was that it was easier to visually scan the plain packs when searching for the name of a particular brand due to the consistent location and writing used for labelling and a clear figure/ground contrast (eg, “the text [for plain packs] is easier to read,” “the print against a plain background was much easier to read” and “plain packs were easier to read because you immediately knew where to look for the name”). One participant also pointed out that plain packs would greatly assist the colour-blind.

An advantage of colour packs, suggested by 25 participants (48.1%), was that it was easier to visually scan by colour than words (eg, “when I heard ‘gold’ I looked for gold,” “I first searched by colour and then by words” and “most were colour co-ordinated making them easier to identify”). However, this was not always considered an advantage; 13 participants also suggested that similarly coloured packets could be distracting (eg, “when you said ‘blue’ I was looking at all the blue packets and got distracted”) or brand names discordant with the colour packaging could be distracting (eg, “sometimes I was looking for a colour when the packet wasn't actually that colour”). Colour appeared mainly useful for brands where a colour was incorporated in the name and featured predominantly in the packaging (eg, “Winfields [predominantly name-colour coded] were easier in colour but Benson & Hedges [predominantly gold packs] were easier in plain”). Many participants also suggested colour was only useful for those cigarette brands with which they were familiar (eg, “I found it easier to use colour for the ones I knew but plain was easier for the ones I didn't know” and “common ones were easier in colour but the uncommon ones were easier in text—especially if looking for brands in the same colour”).

The final topic related to a training effect, with 13 participants (25.0%) suggesting that after several repeated transactions of the same brand they quickly learnt its location such that it became irrelevant whether the pack was coloured or plain (eg, “after a while I just began to remember where they were,” “the second time around was so much easier because I already knew where they all were” and “I remembered where the ones you called lots were in the end anyway”).

Discussion

With the introduction of plain cigarette packaging, the tobacco retail industry predicts an additional 15–45 s per transaction, resulting in long queues, customer frustrations, diminished profits and job losses. However, rather than plain packs causing increased transaction times and more selection errors, our results clearly suggest there are likely to be over three times fewer selection errors and even a slight gain in time efficiency with the introduction of plain packaging, albeit one that rapidly diminishes over time. Our results suggest tobacco retailers hiring new staff might expect a 0.44 s time saving per transaction for the first 50 cigarette packs sold with the introduction of plain packaged cigarettes, after which time the employees will likely have memorised the location of most cigarettes regardless of whether they are coloured or plain. Thus, rather than losing money from inefficiency, we estimate that this time saving will result in tobacco retailers experiencing a one-off efficiency gain to the value of $A0.11 (US$0.12/GB£0.07) per new employeei.

Our qualitative results suggest that colours and logos can serve as a useful cue to locate some tobacco brands, with improved familiarity of logos and colours certainly appearing to facilitate participants' location of them. However, colours and logos also serve to distract when participants attempted to locate other brands, especially those unfamiliar to them. This highlights a potential limitation of our study; we did not include health warnings on our packs, which in Australia occupy 30% of the front of the pack and contain large colour images and text. The distraction of health warnings could potentially increase the difficulty of identification of plain packs by brand name alone, but the same distraction would equally apply to coloured packs. Indeed, participants suggested that the uniform layout of plain packs allowed them to immediately know the location of brand names, as opposed to colour packs, which required scanning over the entire pack face. For this reason, coloured packs would always offer more distraction than plain packs, regardless of whether health warnings were present or not.

The fact that we did not use health warnings on either pack raises another question. In Australia, only 70% of the front of packs is currently reserved for colour branding but our coloured packs covered 100% of the front, allowing larger text, logos and images. Therefore, our coloured packs may have been easier to locate than would be normally the case. However, since conducting our study, it has been announced that the proposed text size of plain cigarette packets will be font sized 16, rather than the 22 we used in our study, suggesting that the smaller font on the proposed plain packs may take longer to read than those in our study. Thus, both the plain and coloured packs in our study are likely to have been easier to read than their real-life counterparts.

Another limitation of our study is that our recruitment process deliberately screened out those most likely to be familiar with cigarette brands. Our results suggest that plain packaging is most useful for searching for unfamiliar cigarette brands. Thus, it is likely that our participants were quicker at locating plain packs than coloured packs because of their general unfamiliarity with cigarette brands, and it remains possible that if the experiment was repeated with experienced tobacco retailers, any advantage conferred by plain packaging would be lost.

That is not to say that we believe coloured packs would be significantly quicker than plain packs for experienced tobacco retailers. Our clearest result by far was the training effect revealed by the experiment, whereby participants rapidly started memorising the location of cigarette brands, regardless of whether they were coloured or plain. Our results suggest that employees with experience of 50 or more cigarette pack transactions should experience negligible differences in transaction times between coloured and plain packaging, as long as the locations of various cigarette packs remain unchanged.

Overall, our results cast serious doubt over the essential arguments underpinning tobacco retailer opposition to plain tobacco packs. It points to the desperation of the tobacco industry that it invested in a multimillion dollar fear campaign based upon a set of arguments so spurious they could be refuted by even a modest experiment undertaken with no budget and a group of volunteers over the space of 1 week.

What this paper adds

  • Australia is poised to become the first country in the world to introduce plain packaging of tobacco products but its government is undergoing immense pressure by the tobacco industry to abandon the policy.

  • The tobacco industry is using politically sensitive ‘small business’ retailers to carry much of its argument.

  • This paper refutes most of their arguments.

References

Footnotes

  • Competing interests None.

  • Ethics approval Curtin University Ethics Committee.

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

  • Data sharing statement All data will be gladly and openly provided by request to the main author.

  • i Based upon a 0.44 s time saving over 50 transactions (22 s), valued at the current Australian minimum wage for casual employees of $18.15 per hour.