Group Research & Development Centre, British-American Tobacco Co. Ltd., SOUTHAMPTON. LM/DM/46E PRINCIPLES OF MEASUREMENT OF VISUAL STANDOUT IN PACK DESIGN ,REPORT NO. RD.2039 RESTRICTED 23.5.1986 10'' AUfHOR: L. Miller GROUP LEADER: R.P. Ferris ISSUED BY: T. Hirji DISTRIBUTION: Dr. R. Sinns Copy No. 1 Mr. A.L. Heard 2 Mr. E.E. Kohnhorst 3, 4 Dr. P.J. Dunn 5 Dr. S.R. Massey 6 Mr. R.G. Nicholls 7, 8 4 0-k Herr E. Rittershaus 9 Dr. E. Kausch IS is 10 Dr. C.J.P. de Siqueira w U 11 Mr. H.V. Thomsen w 12 Mr. R.F. Gilderdale IN a 13 Library d a 14, 15 COPY NO. 1'5 1986 8. A T (U.K. and Exportl LAmettd. This repon must not he copied or shown to unauthonsed pe NIC BATCo document for Legal Services : Health Canada 20 October 1999 Group Research & Development Centre, British-American Tobacco Co. Ltd., SOUTHAMPTON. LM/DN/46E 23rd May, 1986 PRINCIPLES OF MEASUREMENT OF VISUAL STANDOUT IN PACK DESIGN REPORT NO. RD.2039 RESTRICTED SUMMARY 0, The increasing imposition of advertising rest.-, A zition on the tobacco industry inevitably results in the vis~al impact of the cigarette pack itself assuming more import-ance at a means of attracting consumer attention to the pr-:~duct. Whilst acknowledging that the aesthetic aspect of pa:k design must still be regarded as the principal foundation fz-. the develop- ment of brand imagery, this report makes the case 'For accompany- ing such a concern with a routine screening for visual impact during either new pack design exercises, the screening of series of pack variants on a theme, or modi f i catf ans/extens ions of established brand pack design. It is emphas4-z-=d that visual impact measurement should be taken to compleme,7.7- rather than supplant the aesthetic aspect of pack design. This report highlights the methodological issues involved in 401 using a tachistoscope for measurement of visual prominence. It is stressed that the main application of this technique is to measure the relative visual prominence oF the elements within an individual pack. It is not intended the= the technique be used to provide reliable information with respect to the relative overall visual impact of one cigarette pack over and C:) N; 1986 8 A. T I UK. sold Expurt) Lmiled This report must not be co;tcjur sht)%n ,: unauthurised persons C71-1 NIC 3W.-, .. . - - .--- I-Ic BATCo document for Legal Services : Health Canada 20 October 1999 above another. (This issue is currently the subject of a research project which will receive separate reporting). It is anticipated that this technique is more useful in the early stages of pack development to screen between a number of al ::ern- atives, taking into account factors such as the readabil4l:y of brand name or the standout of the company Crest. Five experiments are described. In the first two the ex;:.erim- ental procedure fnvolved in'tachistoscopic testing is spe&fied. 1he need to control, in this type of study, factors such as leve I of illumination is highlighted. In order to ccinduct these first two experiments, seven packs were selected from he U.K. market. The findings from the experimenn are discussed with reference to establishing guidelines for designing packs with a view to achieving maximum visual fmpact of their constituent elements. However, it is emphazised that no suggestion is made that the packs used in the test should be changed in any way as they were launched wi:-, the benefit of advertising support. As stressed previously, it is anticipated that the main -z:)pli- cation of the technique will be in' the development o~ pack designs. In a typical development situation, several :-Bsign alternatives may be ava,ilable all of which have the same brand name but the style of presentation is varied bezween t experiment outlined in the r-eport them. Thus, the third describes the methodology involved in testing sets of packs all with the same brand name. This involves the use of 0-1 matched groups of subjects selected on criteria speciFic to this type of task. This approach avoids the possibil izy of any biasing influence of learning effects on performance. Swiss packs are used to demonstrate the method. Although it is not intended that this technique is used t.:) make comparisons between packs concerning their overall level of visual impact, it may be desired to know, for example, how the N-) -ii- %10 (D 1996 3 AT (UK. and Export I Limited. This report must not be copied or shu*n iounaulhunsed --c-.ons \~D -r-M. NNIMMM BATCo document for Legal Services : Health Canada 20 October 1999 company crest stands out on development packs relative to comcetitionlexisting'packs on the market. In this example it is necessary to know whether or not the perceptual results will be biased by familiarity effects. Experiment Four des- cri~-es a cross cultural study comparing results between Finland and the U.K. It was a complete cross-over design where both Finnish and U.K. subjects were tested on both Finnish and U.K. packs. It was concluded that, apart from reacability of the brand name, familiarity did not affect the relative ordering of emergent pack detailing. The results also suggested that, within the westernised context, i f reacability of brand name is of prime concern then it is poss-ible to test the packs in another market without pre- jud'4cing the results. Experiment five considers the relationship of visual promin- encs of pack design elements to consumer recall of those elernents concluding that a strong positive relationship exists between these two factors. This finding indicates that as well as attracting attention to themselves, visually impacting design elements are more likely to be consolidated into the memory process. The style of tachistoscopic testing reported here represents a ~-"ahly time/cost effective addition to the process of pack deyelopment. It allows the assurance that the various symbol- isim carried by the pack is achieving the desired effect in terms of impact. This is no way detracts from the principaT importance of the aesthetic aspect of pack design to support brand imagery. -iii- C71% 1986 B-AL T W K. stid ElLporti Limited. This repon must not be copied or shown to unauthunsed persons BATCo document for Legal Services : Health Canada 20 October 1999 ~ KEY WORDS Pack Design Brand Imagery Cigarette Packs Pack Imagery Pack Prominance Memory Visual Impact Recognition Print Colour Smokers Brand Name Nonsmokers Consumer Product Testing Tachistoscope Subjective Assessment Consumer Research Techniques Visual Perception '0 1986 B.A.T (U X. and Expon) l mard. This reffli musc noŒt bc copied or shown io un…uthonsed persons r\j cy*% L~_J , M~. , ;o‹, BATCo document-for Legal Services: Health Canada 20 October 1999 1. INTRODUCTION The tobacco industry is increasingly faced with the problem of inability to market products using traditional 'above-line' means. Restrictions on advertising, brought about as an indirect legistlative consequence of the smoking and health debate, mean that many of the accepted routes of communicating with the consumer have been severed. The extent of the problem varies from one market to another. At the,present time the restrictions in most markets are only partial. In these markets, restrictions in comm- unicating through one medium can be partially compensated for by extra expenditure in another. However, in those markets where there is a total ban on advertising of tobacco products (e.g. Finland and Singapore) legislatively viable methods of communication must be developed in order to sustain dialogue with the consumer. These developments achieve even greater significance in the context of the present day market scenario where the increase in novel product propositions make adequate communication with the consumer probably more important than it has ever been. Advertising restrictions have, in many markets, coincided with attempts on behalf of the tobacco industry to meet the demands of the increasing low delivery sector. Another example is where products such as low sidestream cigarettes are required to be launched without the advantage of adequate media to describe the hitherto unfamiliar technical developments and their associated product benefits. Given the above points the consequences of an advertising ban require fullest consideration. Finland represents a good example of an advertising ban market where the ban was introduced in successive stages. . Initially, voluntary industry agreements on advertising restrictions before the tobacco law included cinema (1969), youth's magazines (1969), direct mail (1969), television (1971), and a ban on advertising of new brands (1977). These were the precursors leading up to a total ban I:= on advertising and consumer oriented promotions which took rl.~ CYN place from 1978. Many other contemporary markets are now 1986 8 A.T (U K. and Export) Umited This report mw,. -^K be %:Dplcd or shown to unauthonsai persons BATCo document for Legal Services : Health Canada 20 October 1999 receiving similar warning signals. It is during this period that the use of other media as a means of communicating the product should be maximised to its full potential. In particular, the cigarette pack itself assumes great importance since it is likely to be one of the last remaining means of communicating with the consumer in an advertising ban market. Historically, cigarette pack design has assumed a great deal of importance in the marketing process. This is because brand imagery is salient in the mind of the consumer. Certain consumer segments demand products which portray certain types of imagery. Much of the imagery has traditionally been developed through advertising. However, it has been understood that this imagery must be carried right through to the brand. The cigarette itself offers limited opportunity for conveying brand imagery although longer length, king size (latterly luxury length) and gold bands are examples of opportunities that have been taken. The main focus of attention, therefore, has been on the pack which carries the product. This has been achieved to such an extent that any pack designer understands intuitively that there are bounds which he can not break if he wants to convey certain messages. For example, in the U.K. market it is established in the minds of consumers that green is associated with menthol. This is a very powerful association since it is reinforced by its use in other product fields. Therefore, the designer who wishes to tell the consumer that the pack contains a full flavour U.S. blended type product would be unwise to place it in a green pack. Red is the colour which is pre- dominantly used to convey this particular image. In fact, colour has been used to considerable advantage in the creation of imagery. In the U.K. market, for example, it is now rarely the case to find a high class product which does not carry some gold on the pack. Lower delivery products tend to be featured in blue packs. Indeed, as one moves down the delivery sector then the closer to white a pack tends to become. This is because white is generally held to convey a clean, healthy C D association. Qri 1986 8 1, T It: K. &n4 Export) Limited Th;s report must not be copied' or shuwn to anawhonsird persons BATCo document for Legal Services : Health Canada 20 October 1999 -3- Brand name is also increasingly being used to convey imagery. An example from the U.K. market is Raffles. This name provides a direct associative link with the place from which it originates and, in this example, provides an inference of a classy product. Passport is an example from the Canadian market. This product incorporates reduced sidestream and the name Passport is designed to be associated with the notion of "Passport to Freedom". This may be a particularly attractive notion to the Canadian smoker who has possibly been harassed more over the passive smoking issue than smokers in any other market. To quote one last example, Kool on the U.S. market was designed to 6onvey not only something about the product but also to appeal to the particular market segment at which it is aimed. It is recognised that the use of pack design as a means of conveying brand imagery is now firmly embedded within the understanding of the consumer. However, it is felt that given the consequences of a total ban on advertising, a pack should be designed to give the product visual impact as well as brand imagery. This can be achieved at two levels: (a) The elements of the pack can be designed so that'. those that are most important in communication terms also achieve the highest level of visual impact. An example of this is Marlboro where the chevron, which has very high visual prominence on the pack, is used as a means of sYmbolising the product in advertising ban markets. (b) The pack itself can be designed so that it achieves more visual impact in the point of sale environment than its competitors. Approaches to the measurement of visual impact in the point of sale environment will be the issue of seperate reporting. With the luxury of advertising, pack design for visual impact has been a largely neglected area of research within the tobacco industry. Any research into pack design, therefore, needs to NJ 0-Is establish the ground rules as well as establishing a viable 11-C C 1986 8 A.T (U.K. and Export) Limited This report must not be :Opted or sho-n to unautr'Unsed persons BATCo document for Legal Services : Health Canada 20 October 1999 -4- technique. The first stage within the process of designing packs for visual impact is to look at them on an individual basis. To do this the technique outlined in this report uses an item of experimental equipment known as a conventional tachistoscope. 2. EXPERIMENT 1: THE USE OF THE TACHISTOSCOPE FOR MEASURING THE YISUAL PROMINENCE OF PACK DESIGN ELEMENTS. The initial experiment using the tachistoscope was designed to evaluate the equipment as a means of measuring the visual prominence of pack design elements. A variety of packs from the U.K. market were selected for test and they were as follows: Benson and Hedges (Middle Tar) Marlboro (Low to Middle Tar) Rothmans (Middle Tar) Silk Cut (Low Tar) State Express 555 Medium Mild (Low to Middle Tar) Dunhill (Middle Tar) State Express 555 Filter Kings (Middle Tar) The king size version of each of these brands was used. %) 1986 B A T (U X and Eipan) Lrniied. This rc;x)rt must not be copied ur sho.n to unauthcwtsed persuni A conventional tachistoscope is a piece of portable equipment which allows the presentation of visual stimulus material for successively increasing short intervals of time. Initially, the exposure durations are so short that the stimulus material is presented below the threshold (known as sub-threshold). Threshold is the level at which exposure durations are'just' long enough for the individual to actually become conciously aware of the stimulus. As the exposure durations increase different elements of the pack reach the perceptual threshold level. This level is determined by the visual prominence of those elements. In this research the time intervals used were in the range of 0-20 milliseconds (msec). Previous pilot work had shown that within this time span the majority of individuals can see most elements on all packs tested. A series of studies using the tachistoscope are described inthis report highlighting. methodological and practical issues in tachistoscopic testing. BATCo document for Legal Services: Health Canada 20 October 1999 0 Sixty subjects were tested on each of these packs. The subjects were all consumers from the Charter Research Panel. Any subjects who normally wore spectacles to correct their vision were required to use them during the experiment. The order in which the packs were presented was randomised between subjects. The subject pool was broken down into sub samples of 20 smokers, 20 non-smokers and 20 0 ex-smokers. Since all three smoking status groups were included in the sample seme notion could be obtained on the effect of familiarity in determinina the visual prominence of the design elements. It is expected that smokers would be more familiar with the packs than ex-smokers who themselves would be more familiar than non-smokers. 2. M 'ApDaratus: Tachistoscope The tachistoscope consists of a. viewing device which has two fields (Figure 1) (A more technical description is given in Appendix 1). The first field (A) is an adaptation field used so that the subjects can accommodate to %the level of illumination an the second field (field B) which is the stimulus field. The stimulus cicarette pack is placed on field B. Illumination on both fields is provided by a circular lamp positioned around them. This ensures constant illumination over the field. A two way mirror is positioned diagonally between the fields. Whether the subject sees the adaptation field A or the stimulus field 8 when looking through the view finder is dependent upon which field is illuminated. Illumination of the fields is programmed using the control box shown in Figure 2. From'the wiring in the diagram it can be seen that on pressing the start but ton, timer 1 is triggered. This has the eff-ect of switching off the adaptation field and at the same time switching on the stimulus field for the length of time set on the timer. When the stimulus field has fired for the desired time interval, the programming switches the adaptation field back on immediately. There are other ways of programming the 19S6 B,A.T (U K. and Ez;un) Limited This report musc not be copeed or shv%n to unauthonsed persons CN BATCo document for Legal Services : Health Canada 20 October 1999 , -6- tachistoscope: For example, the light to the adaptation field can remain on whilst the stimulus field is exposed. If comparisons are to be made between experiments, however, then the method of programming must remain constant. For the studies in this report the tachistoscope was always programmed in the previously described way. The intensity of illumination over the fields can be selected using the intensity button shown in Figure 2. This level of illumination should be the same for both fields and remain constant since it can affect the amount of time taken to see the design elements (see Experiment 2). This also means that if comparisons are to be made between experiments then it should be ensured that the level of illumination is constant between the experiments. Unless otherwise stated, the level of illumination of stimulus and adaptation fields used in the experiments described in this report is 100 lux (100%) giving 5 lux approximately at the subjects eyes. 2. (ii) Method The subjects were familiarised with the tachistoscope and its purpose. Some background was also provided on the reasons for performing experiments on the design of cigarette packs. A pack, which was not being tested, was placed in the tachistoscope so that the subjects were able to have a practice session before being exposed to an actual test pack. The 'ascending method of limits' was used to expose the pack to the subjects. This involved presenting the pack for successively increasing intervals of time. One msec time intervals were used and the range was from I - 20 msecs. 1986 B.A.T (U K. snd Exptin) Limited This report must not be copied or sho-n to unauchortsed persons BATCo document for Legal Services : Health Canada 20 October 1999-- After each presentation of the Stimulus material the subject was required to describe everything that had been seen. The experimenter took precise notes on what the subject said and had the relevant cigarette pack available for reference. This was important because often it was difficult to be absolutely sure from the subjects descriptions which design element was being referred to. The method of scoring involved noting the precise time when the subject mentioned a particular design element. For copy print on the cigarette pack two timings were noted: the first was when the subjects saw the copy and the second was when they were able to read it. There is also a score for recognition of the brand/pack. Recognition is when the subject correctly identified the brandipack. If the subject was unfamiiiar with the brand/pack then the recognition time corresponds to the time taken to read the brand name. If, however, the subject was familiar with the brand/pack they would recognise it from its design elements. In such situati.ons the subject was requested to attempt to be objective and indicate when they could actually read the brand name. 2. (iii) Results For each cigarette pack there are four sets of results: the overall scores for all 60 subjects, and the indivjdual scores for smokers, ex-smokers and non-smokers.' Each set of results shows the mean recognition time for the brand"/pack, the mean time taken to see each design element, the standard deviations and the number of subjects who saw the design element. There is a close correspondence between mean time taken to see the element and the number of subjects who saw it. The longer it took for an element CD to emerge then the fewer the subjects who saw it. ON I-C 1986 B.A.T WX and Exponi Urnitcd. This repon must not be copied or shown to unauthonsed persons BATCoAocurnent for Legal Services : Health Canada 20 October 1999 Bids In these results colour is, unless otherwise stated, not taken into consideration. This is because at time intervals of less than 30 msec often the individual is only able to respond to the light energy from the stimulus. Thus, only those cells (rods) in the retina which register light 'on' or 'off' irrespective of colour (scotopic vision) are stimulated. This means that colour is very difficult to determine. Contrast between colours is much more likely to affect whether or not an individual sees any particular element. If there is a good figure ground relationship (high contrast) between the element and its background then it will be seen more quickly than if there is not. (a) Benson & Hedges From Figure 3 it can be seen that the mean time taken to recognise the pack for the total sample was 9.32 msec. When the sample is broken down into smokers, ex-smokers and non-smokers (Figures 4, 5 and 6) it can be seen that for each group the time taken to recognise the pack was in the region of 9 msec. Of the three groups, smokers took slightly longer to recognise the pack than others. However, this difference was negligible. In all cases the first element to emerge was the brand name copy "Benson and Hedges". However, on average it took 9 exposures of increasing duration before the subjects saw anything at all on the pack. The general response at these early stages was that they could see a pack shaped block. The elements on the Benson and Hedges pack, there- fore, do not achieve high visual prominence. The scores shown in Figures 3, 4, 5, and 6 indicate that, on average, the subjects recognised the Benson and Hedges 0N pack immediately after seeing the copy but before being N-0 able to read it. Many reported that they had the impression \-0 Qn 1986 RA.T W.K. and Expurt) Limited This repon must not be copied ur shown t4i unauthunsci persons BATCo document for Legal Services : Health Canada 20 October 1999 , -9- that it was Benson and Hedges because of the plainness of the pack. Seeing the logo on the bottom of the pack confirmed their impression. The time taken to read the brand name was also very similar between the smoking status groups. Comparisons between the figures shows that the logo "Special Filter" was always the last element to emerge. It is quite interesting to note that at these short durations individuals tend to think that all colours are red. This is because the colour receptors in the eye (cones) are more sensitive to the wavelengths associated with*the red part of the colour spectrum. Red, therefore, is a very visually prominent colour (Chamberlin and Chamberlin (1)). Since gold has a high red saturation the subjects thought the pack was red. This may have had the effect of masking the red copy on the pack. The order in which the elements emerge is exactly the same between the groups of subjects. The mean time taken to see them is also remarkably similar between the groups. This suggests that what is being tested using this piece of equipment is an innate perceptual process which is not subject to the effects of familiarity. (b) Marlboro Figure 7 shows the overall results for the Marlboro pack. It can be seen that it took 4.00 msec for Marlboro to be recognised. This is considerably faster than the recognition time for the Benson and Hedges pack. In fact it took less than half the amount of time for Marlboro to be recognized. Comparing Figures 8, 9, and 10 it can be seen that Marlboro was recognised equally quickly by all the smoking status groups. Whilst it took approximately CD 9 msec for any detail to be recognised on the Benson and N-) ON Hedges pack, subjects responded very quickly to the Marlboro %10 N-0 t--Ii 1996 B.A T IL K and Export) Limited. This repon must not be copiedor shown to unauthonsed persons ON BATCo document for Legal Services : Health Canada 20 October 1999 7 -I(- pack i.e. at approximately 3 msec. For each set of results it can be seen that between 3 and 4.5 msec the subjects had seen the chevron, the copy for the brand name and had recognised the pack. The chevron is a particularly prominent design element. This is due to two factors: (a) It is placed against a white background and since it has a high colour saturation presents a good figure ground relationship; and, W As already discussed, red is an extremely visually prominent colour. The copy,for Marlboro is also prominent because the potential of the parallel lines is maximised since the eye is programmed to detect edges. The use of black on white is another example of good figure ground contrast. It is interesting to note that, unlike other packs which present their parts from which the subject determines the whole, Marlboro presents itself as a 'Gestalt'. In most cases the subject sees the majority of detail on the pack. (i.e. the chevron and the copy for the brand name) almost simultaneously and consequently find the pack easily recognisable. For smokers and non-smokers (and consequently the population as a whole) the chevron emerges before the brand name copy. However, for ex-smokers the reverse is true. On close inspection of the figures it can be seen that the difference in time between the emergence of the two elements is very small (i.e. less than half a msec). Consequently, it is suggested that any difference between the ordering of these elements may be artefactual. Readability of the brand name is also much faster for this C_ pack than the previous one. Correspondent with this, more subjects were able to read the brand name. There is 1986 9 A T I U K. end Export) Limited This report must not be ;:opied or shown to unauthonsed persons I BATCo document for Legal Services : Health Canada 20 October 1999 -il- some indication that smokers were able to read it quicker than ex-smokers who themselves were faster than non-smokers. This is quite likely to be due to factors other than familiarity. In all but one case the subjects had recognised the pack and some may have been more conservative in their estimate of readability than others. After the chevron and the brand name, the crest is the next element to emerge. The time taken to see the crest is again similar for all three sub-groups. After the crest, the remaining three elements on the pack (red line, 20 class A cigarettes, Filter Cigarettes) emerge differently for each of the three groups. When looking at the time differences it can be seen that the maximum time difference in emergence between any of the three of these elements is approximately 1.5 msec. These elements, therefore, have very similar prominence values and which one emerges first is quite likely to be related to chance factors. (c) Rothmans In terms of ease of recognition, Rothmans lies between Marlboro and Benson and Hedges. Upon initial presentation of the pack the subjects can see a dark area in the centre of the pack. Unlike the Marlboro chevron, however, this does not provide a clue to the brand. Even after seeing the white brand name copy across the area the subjects still need more time to confirm that it is Rothmans. In all cases, Rothmans is recognised after the brand name is seen but before It is read. No other elements emerge between seeing the brand name and reading it. Therefore, the sight and style of the brand name does not act as an immediate trigger as in the case of Marlboro. A particularly interesting feature of this pack is the CP11 difference in the visual prominence of "Rothmans" and ~10 I go b B.A.T tUX and Export) Limstcd. This report must not be copied or shu-n to unauthonsed persons C71\ BATCo document for Legal Services : Health Canada 20 October 1999 "King Size". They, are both in the same area of the pack and it may be thought that the font (print) style for "King Size" is clearer than that for "Rothmans". However, in all cases "Rothmans" is the second element to emerge and "King Size" is the last. This is another example of the effect of contrast. Since Rothmans is scripted in white an a blue background it is highly visible. King Size, on the other hand, tends to merge in with the background since it is scripted in gold. Another interesting finding in relation to this pack is the positioning of the red area containing the copy "filter tipped" within the emergent detail. Given its relative size this is a particularly prominent element of the pack design. This is again due to the use of colour and contrast. The emergence of the crest, gold seal, and copy on the bottom of the pack is inconsistent between the groups of subjects. It appears that whilst the major elements on a pack are fairly consistent in terms of emergent detailing, smaller elements are more variable. However, the above mentioned elements all fall clearly into the category of low visual prominence. Silk Cut On average it took 10.78 msec for the subjects to recognise Silk Cut (Figure 15). However, there was some variation between the smoking status groups. Smokers tended to take longer to recognise this pack than both non-smokers and ex-smokers (Figures 16, 17, and 18). Corresponding to this, fewer smokers were able to recognise the pack compared with ex-smokers and non-smokers. These findings are contrary to a hypothesis which suggests that familiarity plays a part in recognition of pack design elements at these short exposures. C71 I q&6 a. AT I UK. and E%oiii; Limited. This report must not be copied or shown to unauthorised persons BATCo document for Legal Services : Health Canada 20 October 1999 -13- For all groups of subjects the first element on the pack, i.e. the dark central square is seen very quickly. However, the brand identifier positioned on the square takes much longer to emerge. This, again, is due to the use of the colour gold. This particular colour placed against the purple background does not provide the contrast necessary to make the name stand out. The subjects saw and were able to read the logo "Benson and Hedges" much faster than uSilk Cut". Even though "Benson and Hedges" is not centrally positioned and is written in smaller writing, its contrast with the background makes it more visible than "Silk Cut". The logo "Silk Cut", "King Size" and the crest are all elements which have very similar visual prominence values but the order in which they emerge differs between the smoking status groups. Since the time differences are so small, however, this change in ordering is not of any great significanc e. In all cases "Ventilated Filter" is consistently the last element to emerge. This element is disadvantaged because of its relative size, positioning and colouring. (e) Dunhill Dunhi I I presents itself visually in a very similar fashion to the Benson and Hedges pack (Figure 91. It takes some time for the subjects to see any detail at all on the pack. This is because the detail is presented in gold and is usually fairly fine. The amount of time taken to recognise this pack is similar to that for Benson and Hedges. Similar to Benson and Hedges, and unlike many of the N~ other packs discussed, the detailing on this pack does 1,C not fall into three categories. Usually, there is a 1986 B.A.T I U K. and Expon) Limited. This report must not be copied or shu-n to unauthorised persons BATCo document for Legal Services: Health Canada 20 October 1999 -14- highly visually prominent element of the pack design (e.g. the square on Silk Cut or the chevron on Marlboro) which is seen within the first few presentations. This is followed by less visually prominent elements which emerge after approximately 6 msecs. Latterly, there are those elements which emerge after approximately 12 msec which probably would not have caught the subjects attention if they had not been forced to look for them. In the case of Dunhill and Benson and Hedges there are no elements which achieve high visual prominence. Another feature that Dunhill shares with Benson and Hedges is the close correspondence between the time it takes to recognise the pack and the first sighting of the brand name. Although the subjects had an impression of a red colour pack with a gold coloured outline they needed to see the style and positioning of the brand name before being able. to recognise the pack. With other packs the subjects often recognise the brand without seeing the name by making inferences from the visible elements. Comparing across the smoking status groups (Figures 20, 21, and 22) it can be seen that the gold coloured frame is the most visible pack design feature. This is generally followed by the impression that the pack is a red colour. However, for smokers the brand name is reported before the colour. This did not represent a tendency for smokers to see the brand name quicker since all groups tended to see it at approximately 10 msec. The reason that the brand name stood out before the colour for the smokers was that they were slower than other groups to mention the colour of the pack. This could have been due to response bias on behalf of the subjects as through- out this experiment they were generally looking for design elements rather than colour. NJ %10 The crest is consistently the fourth element to emerge Q^4 1986 13 A. T I UK. ond Export I Limited. This report must -tvt :!e zc pied or sho%n to unaut~onscd persons BATCo document for Legal Services : Health Canada 20 October 1999 on the pack. However, amongst the other lower order elements (i.e. those that emerge between 13 and-20 msec) the ordering of emergence is variable. This tendency towards variable ordering with the less visually prominent elements of a pack has been noted previously. (f) State Express Medium Mild From Figure 23 it can be seen that this State Express pack is made up of three colours: gold, white and blue. The white acts as a background against which blue stands out very well but provides no contrast with gold. Consequently, the blue parts of the pack design are the first to be seen. Looking at the results of all three smoking status groups it can be seen that the blue circle, blue chevron and "Medium Mild" are invariably the first three elements to emerge. The blue circle and the blue part of the chevron are seen almost simultaneously. This is true for smokers (Figure 24), ex-smokers (Figure 25), and non-smokers (Figure 26) with the greatest difference in time between the emergence of these two elements being only 0.35 msec. This finding probably accounts for the fact that smokers, on average, report seeing the chevron before the circle whereas the converse is true for the other smoking status groups. Given the fact that the subjects saw the major features of the pack design very early in the task it is somewhat suprising that the mean recognition time for this pack is fairly long at around 8.27 msec. The reason for this is open to conjecture. It could have been that the subjects who participated in this study were not particularly familiar with this brand. Another plausible explanation is that subjects were aware that this brand was part of a family and were waiting to see elements NJ which would allow them to positively identify this 11b particular pack. It is felt that it is more likely, 1986 9 A T I U.K. and Exportl Limiwd. This report must not be copied at sho-n to unauthimsed persons BATCo document for Legal Services : Health Canada 20 October 1999 After the blue elements have emerged, the gold, details on the pack are seen by the subjects. The fluting around the blue circle is the next element to be seen and is consistently followed by "555". It is slightly disturbing to note that the brand identifiers take over 10 msec to be seen and are among the last elements to emerge. However, as has been discussed previously with the Silk Cut brand name and with the Rothman's "King Size", this type of result is not unexpected when gold is used on a darker colour. (g) State Express Filter Kings The total recognition time for this pack was 8.42 msec which was very similar to the medium mild pack (Figure 27). A learning effect was noted in this study with subjects recognising the second State Express pack (regardless of which one it was) more quickly than the first - this was counteracted by completely randomising the order of presentation of the packs. Comparisons between the subject groups (Figures 28, 29 and 30) show that there were differences in terms of the amount of time tdken to recognise this pack. However, with reference to the familiarity hypothesis, these results were not in the expected direction. Non-smokers were the fastest at recognising this pack and ex-smokers were the slowest. Unlike the other packs that have been discussed, there appears to be very little consistency between the subject groups in terms of ordering of emergent elements. The 0- blue elements of the pack design tend to emerge first. 1986 B A T (L K ond Eitpvm) Limited This report must not be copied or sho-n to unaudionsc~f ;>crson$ -16- r however, that the very visually prominent elements in this pack are not readily identifiable with the brand as, for example, the Marlboro chevron is. BATCo. document for Legal Services : Health Canada 20 October 1999 -17- The one exception to this is for non-smokers where "Filter Kings" emerges after the fluting around the blue circle. However, in all cases the blue circle and the blue part of the chevron are the first two elements to emerge. Generally, the circle is seen first at approximately 5 msec. Ex-smokers provide an exception to this general finding. For these subjects there is a marginal tendency to see the blue part of the chevron before the circle. This could be accounted for by the fact that these two elements are seen almost simultaneously by this group of subjects. The brand identifiers "555" and "State Express" are very slow to emerge on this pack. In all cases they take more than 10 msec. This finding conlirms previous findings that gold placed against a dark background does not provide the contrast necessary to achieve high visual prominence. Apart from the divergence of the elements based on colour (i.e. gold elements being less visually prominent than blue elements) the ordering of emergent elements is highly variable between the s--moking status groups. As discussed, this finding is contrary to the results from other pack designs. Also with previous findings it was apparent that, regardless of the order of emergence, for any individual element the time taken to see it was similar between all the smoking status groups. This tendency, however, was not seen in the data for this. pack. The reason for the discrepant findings is open to conjecture. One possible reason is that after the circle and the blue chevron, all the elements have approximately equal, low visual prominence values. Given this, the order in which.the subjects mention the elements is random. It is interesting to make a direct comparison between the results for this pack and those for State Express 555 Medium Mild. For both packs the design is exactly the same but the colour combinations are different. To confirm the contribution of contrast to yisual prominence separably 1996 B.A.T (U K. and Export) Limited. This report must not be copied or snw-n io unituthonsed persons BATCo. document for Legal Services : Health Canada 20 October 1999 -18- from the contribution shape the finding relating to the chevron on these packs is of importance. In both cases it is the blue part of the chevron which emerges before the gold part even though the area covered by blue reverses from one pack to the other. An additional finding of interest with respect to the contribution of colour is that on average (Figures 23 and 27), the time taken to see the blue aspects on the pack was nearly 2 msec faster on the Medium Mild pack than on the Filter Kings pack. This is because blue against white provides better contrast than blue against gold. On the State Express Medium Mild pack the emergence of the gold elements does show some consistency between groups. As well as showing consistency in order of emergence, the times taken to see each individual element is very similar between the smoking status groups. It is suggested that the difference in findings between the two packs is related to the fact that in one case the gold is presented mainly on a white backaround and in the other it is presented on a yellow background. The former, because of better contrast. effects, poduces more reliable results than the latter. 2. (iv) Discussion A technique has been described which allows measurement of the visual prominence of the various elements within a cigarette pack using a tachistoscope. With this piece of equipment it is possible to present cigarette packs to a subject for gradually increasing short durations of time. When this is done the elements of a pack emerge in order of visual prominence. The packs of seven established brands on the U.K. market were tested using this technique. Under more 0 usual circumstances it would be expected that the technique would be used to test novel pack designs that were under consideration for a market launch. "a 0 1996 8 A T (U.K. and Exporti Limited This repon must not be copied or sho%n to unauttkinsed persons BATCo,document for Legal Services: Health Canada 20 October 1999, , _auw_~ _19- The purpose here, however, was to establish the feasibility of the technique rather than the feasibility of the packs for the market place. None of the findings outlined here are intended to suggest that any changes in the pack designs described would be desirable. Such packs were launched with the benefit of advertising support and as a consequence the prevailing market situation at the time of launch did not demand that visual prominence of the elements be considered as a salient issue in the pack design. Having emphasised this, however, it is still possible to use the findings derived from these packs'to establish some of the ground rules involved in pack design for visual impact of its component elements. It is also important to note that although some comparisons have been made between the packs discussed it is not intended to make any conclusions about the relative visual prominence of one pack to another. The conventional tachistoscopic test is a test of the relative visual prominence of elements within any one pack and should be used solely for this purpose. Other methods are available which allow the comparison of the relative visual impact of packs and they will be described described briefly in the concluding section of this report. Sixty subjects were tested and the sample was broken down into equal numbers of smokers, non-smokers and ex-smokers. Comparisons between the subsamples allow some comments concerning the likely effects of familiarity in this type of task. It was hypothesised that smokers would be more familiar with the packs tested than ex-smokers who themselves would be more familiar than non-smokers. The results indicate that, in terms of recognition of the packs, there was no evidence to suggest that familiarity with the packs had any effect on performance. In fact, it was often c7N %10 the case that the non-smokers were able to recognise 1996 3 A.T (U X. and ExporO Limited. This report must not be copied or sho-n to unauthensed persons , BATCo document for Legal Services : Health Canada 20 October 1999 Rem, -20- the packs more quickly than the other smoking status groups. However, this finding concerning familiarity can only be regarded as an initial insight into the phenomena. All the subjects were from the U.K. and all the packs were from the U.K. market. Consequently, the non-smokers tested would have had some degree of contact with the stimulus material and therefore cannot be regarded as naive subjects. Comparisons were also made betwen the subgroups in terms of ordering of emergent detail. Again, it was hypothesised that familiarity with the packs may play a part in determining the order in which the pack detail was described by the subjects. From the results it appears RM that elements of pack designs tend to be ordered hierarchically in three tiers. There are those elements that are highly visually prominent and are seen almost immediately (at approximatley three msecs) upon presentation of the stimulus material. Slightly less prominent detail is seen at around six msecs and the fine detail on the pack tends to emerge after approximatley twelve msecs. (In the case of the latter, the detail would probably never be noticed with a quick glance at the packs on the point of sale gantry and it is therefore important that such detail does not carry any information which is highly relevant to the consumer e.g. brand identifiers). When comparing between smoking status groups the detail falls consistently into one of these three categories. Hence, an element which falls into the highly visually prominent category for one group of subjects does the same for another group of subjects. Also, within any category,.the time differences between the emergence of elements is often very small indeed. In such cases variability in order of emergence between the groups was noted. The more visually prominent the category, however, the greater the degree of consistency of CN ordering between the groups. ID 1986 B A.T (U K. and Export) Limited T4js rep-)rt must not be copied or shuwn to unauttiinseJ persons BATCo document for Legal Services : Health Canada 20 October 1999 In the case of 555 Filter Kings the above findings tended to break down. Generally, the blue elements of this pack design emerged first and the gold elements second. Comparisons between the groups, however, indicated that other than the divergence of the elements in terms of colour there was very little consistency between the groups as far as emergent detailing was concerned. It was suggested that the reason for this may have been due to the fact that most of the elements have approximatley equal, low visual prominence values and given this the order in which the subjects mentioned them may be influenced by random factors. The findings relating to the hypothesised effect of familiarity suggest that the technique is testing an innate perceptual process which at these time intervals in not subject to corruption from the effects of familiarity. However, as stressed previously, it is likely that all the subjects tested had some degree of exposure to the packs used in this experiment. The only true test of the effects of familiarity on the perception of pack detail would necessitate a cross cultural study. The design would involve comparing the results of a tachistoscopic test on subjects familiar with the packs used with the results from a population in another market who had not been exposed to these particular packs. Concerning the hierarchically ordered categories of emergence, it was noted that none of the detail on the Benson and Hedges pack or the Dunhill pack falls into the highly visually prominent category. It was suggested that this may be the reason why, of the packs tested, these two brands took the longest time to be recognised. Upon initial presentation 10 110 1986 B AT W K iind Export I Limited, This re-,x%rt must noi !!c z~-picd or shou a to unauthofised persons BATCo document for Legal Services : Health Canada 20 October 1999 (a) Size : On the 555 Filter Kings pack the gold part of the chevron is larger than the blue part but because CD rl%.; the blue provides a better contrast with the gold ON than the gold does with yellow it stands out better. I-C 1986 B AT I U K. and Exponl Limited This repon must noi be copied or 5hu~n to unauthonsed persons -22- of these packs all the subjects saw was a plain pack shaped block. Even if they had an impression of what the pack may have been they had nothing to substantiate this on. With the Marlboro pack, on the other hand, the chevron is highly visually prominent and as a result of point of sale advertising activity (which is often still allowed in markets with advertising bans) in the U.K. it provides a direct associative link with the brand. In many cases it is only necessary for the subject to see the chevron and they know the brand is Marlboro. The use of such design detail to symbolise the product is highly relevant in ban situations where it is not allowed to promote the brand itself. For the ploy to be effective, however, the symbol used must fall into the highly visually prominent category. The discussion so far has centred around the order of emergent detailing of pack design elements. Clearly, a most important question is what determines whether or not an element achieves high visual prominence? Is it size, positioning on the pack, colour ' contrast or shape? The results reported here allow some comment on the relative importance of these factors in determining the visual prominence of the design elements. The overall conclusion must be reached that contrast is the most important factor in determining visual standout. The rule is very simple, if the element has high colour contrast with its background it will stand out, if it does not then it will not. The importance of contrast over and above the other variables is highlighted in the following examples: BATCo document for Legal Services : Health Canada 20 October 1999 -23- (b) Positioning on the Pack : On the Silk Cut pack the brand name is in a central position and is relatively large. However, because the copy "Benson & Hedges" is black on white (rather than gold on, purple as seen for the brand name) it has more visual prominence despite its less advantageous positioning on the pack. (c) Colour: Comparing between the Benson and Hedges pack and the Rothmans pack it can be seen that although the red element "Special Filter" on the former is bigger than the red element on the latter, it takes twice as long to be seen. The background on which these elements are placed clearly has an important role to play in determining the effect. (d) Shape: A comparison between the two State Express 555 packs demonstrates the contribution of contrast to visual prominence over and above the contribution of shape. In both cases it is the blue part of the chevron which emerges first even though the area covered by blue reverses from one pack to the other. 3. EXPERIMENT 2: THE EFFECT OF CHANGING LEVELS OF ILLUMINATION ON EMERGENT PACK DETAIL When developing any technique the parameters that should be controlled need to be determined. Since the technique described here has been previously unexplored it was decided to replicate Experiment 1 on all the variables other than level of illumination. As far as the equipment itself is concerned this is the only variable which can be manipulated. It was felt that it was important to determine the effect of change in level of illumination to provide guidance on the extent to which the technique should be standardised. CD rl--) 01% 0 1996 8 A.T (U K. and Ezporl~ Limited. This report must not be copied or sho-n to unauthonsed persons BATCo document for Legal Services: Health Canada 20 October 1999, ~ 3. (i) Results and Discussion 46 Unlike Experiment I the group results for smokers, ex-smokers and non-smokers are not presented separately. This is because the familiarity hypothesis, as it relates to smoking status, was explored thoroughly in Experiment 1. Included in the overall results for all 60 subjects is the mean recognition time for the pack, the mean time taken to see each element, the standard deviations and the number of subjects who saw the element. CD N-) ON Q-11- 1996 B.A.T 4t; K grid Export) Ljm&tcd Thts report must not be copicd or shu-n to unauffionsed persons -24- Exactly the same procedure was adopted as described previously. However, on this occasion the level of illumination on both the stimulus field and the adaptation field was reduced to 500 lux giving 2.5 lgx at the subject's eye. A different group of sixty subjects were tested in the experiment. The sample was again drawn from the Charter Research Panel and was made up of 20 smokers, 20 ex-smokers and 20 non-smokers. The same seven packs as in Experiment I were used in this experiment. Note: W For the Benson and Hedges and Dunhill packs a pilot study showed that under conditions of lcwer illumination the 20 msec upper limit on exposure duration set in Experiment 1 resulted in subjects being unable to see many of the pack design elements. It was felt that this would be disheartening for the subjects as they like to feel that they are positively contributing to the experiment. Consequently, an upper limit of 30 msec was set for these two packs. However, any emergent detailing after _20 msec was not included in the analysis of the results. The customary limit of 20 msec was used for all other packs included in the study. BATCo document for Legal Services : Health Canada 20 October 1999. - -25- The results of this experiment are shown in figures 31 - 37. It can be seen that, with the exception of the least visually prominent elements, the effect of reducing the level of illumination considerably increased the amount of time taken to see each element. For the majority of elements it took over 10 msecs for them to be seen. Tables I to 7 show statistical comparisons between the results from Experiments 1 and 2. The elements are listed in the order in which they emerged in Experiment 1. Student's 't' tests were performed on the means and standard deviations and standard deviations and chi squared tests were used to make comparisons between the number of individuals who saw each element. 0 01% 11C 1986 0 A T (U K. and Export) Urnsted. This report must not be copied or shown to unwhonse-1 pmons C, BATCo document for Legal Services : Health Canada 20 October 1999 - -26- TABLE 1 BENSON & HEDGES Elements t-Value Level of CHI Level of Significance Square Significance Benson & Hedges Copy 9.07 P 0.0005 9.40 P 4 0.01 Benson & Hedges Ident 2.96 P 0.025 46.45 P 4 0.001 Embossed B & H 4.47 P 0.0005 43.18 P 4 0.001 Crest 0.97 NS 32.17 P 4 0.001 Special Filter Copy 1.28 NS 0.13 NS 4) 1986 8 A T (U K. and Expurtl Limited This report must not be copied or sho-n to unauthonsed persons BATCo document for Legal Services: Health Canada 20 October,1 999 -27- TA13LE 2 MARLBORO Elements t-Val ue Level of CHI -Level of Significance Square Significance Red Chevron 7.12 P 0.0005 4.38 P < 0.05 Marlboro Copy 9.61 P 0.0005 4.38 P 0.05 Ident 7.44 P 0.0005 12.19 P 0.001 Crest 6.24 P 0.0005 14.93 P < 0.001 Red Line 4.30 P 0.0005 26.41 P -C 0.001 Copy at Bottom 3.90 P 0.0005 34.48 P < 0.001 Filter Cigarettes Copy 5.01 P 0.0005 10.97 P r. 0.001 te I 1936 B.AT JUX and Export) Lamed. This report must not be copied or sho-n to Unauthonsed persons BATCo document for Legal Services : Health Canada 20 October 1999 -28- TABLE 3 ROTHMANS I Elements t-Val ue Level Of CHI Level of Significance Square Significance Blue Shape 7.51 P 0.0005 0.50 NS Rothmans Copy 9.46 P 0.0005 4.38 P < 0.05 Ident 7.56 P 0.0005 12.84 P 4 0.001 Red Shape 6.22 P 0.0005 25.65 P 4 0.001 Crest 4.20 P 0.0005 34.48 P 4 0.001 Copy at Bottom 2.68 P 0.0005 38.80 P " 0.001 Gold Shape 1.04 NS 1.24 NS King Size Copy 0.63 NS 18.65 P 4 0.001 Ident 2.36 P 0.025 27.78 P < 0.001 C 1,396 8 A T X K. and Export) Limited. This report must not be ccpied or shown to unautNnsed pensons t BATCo document for Legal Services : Health Canada 20 October 1999 -29- TABLE 4 SILK CUT Elements t-Val ue Level of CHI Level of Significance Square Significance Purple Square 7.36 P 0.0005 5.17 P.; 0.05 Benson & Hedges Copy 6.05 P 0.0005 40.95 P < 0.001 Ident 1.20 NS 48.15 P < 0.001 Silk Cut Copy 3.96 P 0.0005 34.51 P < 0.001 Ident 2.13 P 0.025 55.19 P < 0.001 Crest 1.52 NS 62.25 P '. 0.001 King Size Copy 1.31 HS 56.28 P < 0.001 Ventilated Filter Copy 0.87 NS 23.64 P 4 0.001 0 CD 1996 B A.T (U K. mind Export) Limited. This report must not be copied or shown to unauthonsed persons cc BATCo.docurnent for Legal Services : Health Canada 20 October 1999 -30- TABLE 5 DUNHILL Level of CHI Level of Elements t-Val ue Si gni f i cance Square Significance Gold Border 6.60 P 0.0005 3.34 NS Red Pack 7.46 P 0.0005 5.05 P ; 0.05 Dunhill Copy 5.33 P 0.0005 33.34 P < 0.001 Ident 5.23 P 0.0005 53.34 P < 0.001 Crest 1.84 P 0.05 49.81 P -C 0.001 Copy on Flip Top 0.80 NS 36.95 P < 0.001 Copy Under Crest 2.87 P 0.05 10.70 P < 0.01 King Size - - 24.45 P 4 0.001 Copy Under Dunhill 2.30 P 0.05 19.83 1P ( 0.001 0 . C, '0 1986 13 "~ 'r tU X and Exporti L.mited. Th.s report mus% not bc copicrj or sho~n tu un3uth,,nsc~j pers,,"% BATCo document for Legal Services Health Canada 20 October- 1999 -31- TABLE 6 555 MEDIUM MILD El ement t-Val ue Level of CHI Level of Significance Square Significance Blue Circle 15.16 P 0.0005 2.32 NS Blue Chevron 8.44 P 0.0005 4.38 P 0.06 Medium Mild Copy 9.06 P 0.0005 23.08 P 0.001 Ident 0.68 HS - 45.94 P 0.001 Ident 1.22 NS 40.43 P 0.001 Fluting 2.94 P 0.005 17.34 P 0.001 555 Copy 2.72 P 0.005 36.30 P 0.001 Ident 2.00 P 0.025 64.61 P < 0.001 State Express Copy 1.71 P 0.05 16.92 P 0.001 Gold Chevron 0.30 NS 0.10 NS Crest 0.70 NS 24.54 P 0.001 le, c:::> rQ C71% 1-10 %10 Q.1-4 1986 B A T (U.K. and Export) Limited. This report must not be copied or sho%n to unaullmnsed Persons CX:) rNj BATCo document for Legal Services : Health Canada 20 October, 1999 TABLE 7 555 FILTER KINGS Elements t-Val ue Level of CHI Level of Significance Square Significance Blue Circle 8.28 P 0.0005 17.5 P < 0.001 Blue Chevron 6.96 P 0.0005 23.08 P 4 0.001 Filter Kings Copy 5.90 P 0.0005 60.32 P < 0.001 Ident 0.33 NS 50.14 P 4 0.001 Fluting Z.82 P 0.05 38.63 P < 0.001 555 Copy 2.22 P 0.025 61.69 P -C 0.001 Ident 0.13 NS 51.62 P c 0.001 Gold Chevron 0.44 NS 1.82 NS Crest 0.14 NS 24.74 P < 0.001 State Express Copy 0.70 NS 37.33 P < 0.001 Of London Copy 0.11 NS 0 NS '~ 14S4 8 .4. T (L: K and Expurt) L.rn~ted Th:% report must r3o bc ccooed or shuun tu unauthonsed pcr5cms BATCo Aocurnent for Legal Services : Health Canada 20 October 1999; ' -33- From the t-values and significance levels given in the tables it can be seen that lowering the level of illumination has a considerable effect on the amount of time it takes for each element to emerge and also on the number of subjects who saw each element. Interestingly, where there are no statistically significant differences in time taken for the elements to emerg the elements had achieved very low visual prominence values under the higher illumination level. It can be hypothesised, therefore, that the consequence of lowering the level of illumination is to reduce the constrast effect necessary to see the more visually prominent elements. It can also be seen from the tables that as well as increasing the time taken-to see the elements a reduction in illumination also means that fewer people actually see each element. Contrary to the effect observed with time, it was noted that failure to find statistically significant effect on this measure occured on the more visually prominent items. This is not suprising when consideration is given to the fact that because these elements took much less time to emerge more subjects have time to see them within the given exposure period. The results of this study show quite clearly that level of illumination is a variable that needs to be controlled very tightly in this type of experiment. This is particularly important if the results of one experiment are to be compared with those of another. The level of illumination adopted in Experiment I was chosen to represent the approximate level of illumination to the consumer's eye when a pack is placed on a shelf under artificial lighting conditions. This level is therefore taken as the standard. C7*1 ~Ic 1996 B.A.T (U K. arid Export) Urnittd. This rcpurc must not be copied or shown to Unauthonsed pcrsoos C'. BATCo document for Legal Services : Health Canada 20 October1999 -34- 4. EXPERIMENT 3: A METHODOLOGY FOR TESTING A GROUP OF PACKS WITH THE SAME BRAND NAME The research described so far has been principally concerned with the development of the tachistoscopic technique. A series of different packs taken from the U.K. market have been tested on a one-off basis. In the case of 555 State Express both the middle and the low to middle tar versions of the pack were tested. It was noted that with these two packs there was a strong learning effect such that, regardless of which pack it was, the subjects were always able to recognise the second pack faster than the first pack. This problem was overcome to some extent by completely randomising the order of presentation of the packs between subjects. In the experiments described, this solution to the problem was acceptable since the principal aim of the study was to establish the technique rather than to draw firm conclusions from the results. In many practical applications of tachistoscopic pack testing the problem of dealing with sets of packs with the same brand name will often be encountered. This is because it is often the case that in the pre-launch situation the brand name will have been selected but the actual design of the pack is under test. In this case an accurate reflection of the visual prominence of the brand name in each design is required. Consequently, randomisation of the presentation order is not a solution to the problem since it doesn't prevent the learning effect occuring. In this experiment, therefore, a methodology is described which allows the learning effect problem to be overcome. A selection of packs from the Swiss market have been used to demonstrate the methodology. 4. (i) Method One of the ways of overcoming the problem of learning effects is to use matched groups of subjects. Depending on how many versions of the packs there are to be tested, an appropriate number of groups C. 1986 8 A T I L K ..,,! E.pn) L.,,tcd Th.s rer~,rt must nt,t be coptcd or shown to unauthortsed persons BATCo document for Legal Services : Health Canada 20 October,1999 -35- of subjects are selected. The individuals across the groups are matched on the basis of a number of variables thought to be important with respect to performance on this particular type of task. Each group of individuals are then tested on one version of the pack design and the results are compared across the groups in order to determine which of the packs achieves the design objectives in terms of visual prominence. The success of such a methodology is, of course, dependent on the selection of the appropriate matching variables. In this experiment the Ishihara (2) colour blindness test was used as a screening device. This is obviously an important variable when testing coloured packs. All subjects who passed the screening test subsequently had their eyesight tested using the standard Snellen Chart. The subjects were allowed to wear spectacles to correct their vision for the eyesight tests on condition that they wore them in the main experiment. On the Snellen chart a series of letters are arranged linearly so that the letters get smaller as the subject visually moves down the chart. Each line is allocated a score and this represents the angle at which the letters subtend to the eye when the subject stands twelve feet away from the chart. The left and right eyes are tested individually and the score allocated to each represents the last successful line completed. The third and final test involved measuring the subjects recognition thresholds. This is where a series of words are presented to the subjects tachistoscopically and the time taken to read them is recorded. Six words were selected from a word,count (3) all of which have approximately the same frequency of occurence in written English. The order in which 0N the words were presented to the subjects was randomised. 1-10 1986 &A.T (U.K. and Eipon) Limited. This report must not be copied or shoun iu unauthonsed persons cc 0-1 BATCo document for Legal Services: Health Canada 20 October,11 999 -36- The first word was always taken as a practice trial and its score was not included in the analysis. In this experiment the following sets of packs were tested: f.1 1. Marocaine Extra (A), Marocaine Super (B), Marocaine Mild (C) 2. Marylong Double Filter (A), Marylong Extra Fine (B), Marylong Natural (C) 3. Parisienne Super (A), Parisienne Mild (B), Parisienne Extra (C) 4. Brunette Extra (A), Brunette No 3 (B), Brunette Double Filter (C) 5. Muratti Ambassador Multifilter (A), Muratti 2000 (8), Muratti knbassador Extra (C) 6. Select No 2 (A), Select No 3 (B), Select (C) Since there were three packs in each set there were three groups of subjects included in the experiment. There were twelve individuals in each group and all 36 subjects were selected from GROC personnel. The subjects scores on each of the criteria used to select them are shown in Appendix 2. The full design of the experiment is outlined in Table 8. As a further precaution against possible biasing due to superior performance of any one group, the presentation of packs from a set was counterbalanced between the groups. The packs in a set were arbitrarily assigned A, 8, and C and these assignments are shown in the table. Reading across the table then, for Pack 1: Subject 1, Group 1 received the A version; Subject 2, Group 2 the B version; and Subject 3, Group 3 the C version. Although Subject 7 was also in Group I he received the B version of pack 1. A full description of the design of the experiment is given in Appendix 3. ; 0- SIC, : C C 1996 B.AT (U.K. and Expon) Limited. This report must not be copied or shu-n to unauthonsed persons BATCo document for Legal Services : Health Canada 20 October .1999 Do > 0 -37- 0 CL 0 TABLE 8 (D EXPERIMENTAL DESIGN >W r+ PACKS PACKS PACKS 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6 0) SUBJECT I A B C A B C SUBJECTS 2 8 C A B C A SUBJECTS 3 C A B C A B 4 A C B A C B C B A C B A 6 8 A C B A C 4) 7 B A C B A C 8 A C B A C B 9 C 8 A C B A .01 i :z -q 10 B C A B C A 11 C A B C A B 12 A B C A B C 13 C B A C B A 14 B A C B A C 15 A C B A C B (A '91 16 C A 8 C A B 17 A B C A B C 18 B C A B C A 19 A B C A B C 20 8 C A B C A 21 C A B C A B (D 00 22 A C B A C B 23 C B A C B A 24 B A C B A C 0 25 B A C B A C 26 A C B A C B 27 C B A C 8 A V. 28 B C A B C A 29 C A 8 C A B 30 A B C A B C 31 C B A C B A 32 B A C B A C 33 A C B A C B 0 r, CL 34 C A B C A B 35 A B C A B C 36 B C A B C A 0) rl - I GROUP I GROUP 2 GROUP 3 0 R, 0 CIL 0 cr CiD cc c~ 0 r- ~ ( ,q7p., 1 -38- The procedure adopted in this experiment was the same as described for Experiment 1. Thus, subjects were given one pack to practice on so that they could familiarise with the technique. Using the ascending method of limits, each pack was presented to the subject for a-maximum of 20 msec. As with the previous experiments described, this study was not designed to make recommendations concerning the designs of the test packs used. However, since in this study direct comparisons are posssible between a number of very similar pack designs some useful comments can be made concerning visual prominence. It is once again stressed that these comments are made purely as an aid to establishing principles for pack design. It is recognised that these products are from a market where the scenario at the time of launch did not demand,pack design for visual impact to be considered as a primary factor. Also, when considering the recognition scores for these packs the fact that some of the packs were familiar to many of the subjects tested whilst others were totally unfamiliar must be taken into consideration. 0 4. (ii) Results The results are presented in the standard format where for each pack the following data is given: (a) The overall recognition score for the pack; (b) The mean time taken to see each element and the standard deviation; and, 0 (c) The number of subjects who saw each element. ~J I G_ (D 1986 8 A T tU X oni Exportl Limited This report must not be copied of sho-n to unAuthortsed persons OQ _4* BATCo document for Legal Services : Health Canada 20 October 1999 -39- The findings are presented on a pack set basis. In most cases the points made serve to reiterate those made in the discussion of results from Experiment 1. Again, it must be emphasised that the findings do not indicate which of the packs in a set will be more visible on the point of sale gantry but rather which elements within any single pack will be more visible. (a) Marocaine Marocaine Mild and Marocaine Super provide an interesting comparison between two packs that are very similar in design (Figures 38 and 39). In the case of the former, the red stripe on the pack runs through the "MA" of the brand name. In the case of the latter,'it is the "ROCAINE" section of the brand name which is embedded in the stripe. In both examples this has the effect of splitting the brand name such that the part which is positioned on the white area is considerably easier to see than that part positioned on the red area. This is obviously a contrast effect which is exacerbated by the particular colour combinations used. As stated previously, at very low exposure durations the eye is particularly sensitive to the red wavelengths in the colour spectrum. Gold is a colour with a high saturation of red. Consequently, the brand name is stimulating the cells responsive to red and, therefore, in the area where it is covered by a red stripe (which is also stimulating the red responsive cells) it becomes indistinguishable from the background. The brand name on the Marocaine pack is seen very quickly and read at least five milli- seconds earlier than on the other two packs. This indicates that the above findings were not due to such factors as, for example, Marocaine presenting Itself as a difficult name to pronounce (Figure 40). ON 1986 B.A T I U K and Eitportl Limited This report must not be zopied or shoun to unauthortsed persons BATCo document for Legal Services : Health Canada 20 October 1999 -40- The masking effect of the red background against gold can also be seen with respect to the copy "Maryland Fin". This is featured on all three packs but is completely masked on the Marocaine Super pack. i Marylong It is difficult to attempt to make comparisons between the Marylong packs as they present very different designs (Figures 41, 42, and 43). However, it can be seen that the black on white for the brand name on the double filter pack is slightly more effective than the white on red in the naturel pack . The female head on the naturel pack is more visually prominent than on the double filter pack. This is probably due to a combination of effects related to contrast, size and positioning on the pack. (c) Parisienne 1 In comparing the three Parisienne packs (Figures 44, 45 and 46) there is some suggestion that a white figure on a dark background is less visually discernable than a dark figure on a light background. Looking at the packs, it can be seen that the highest contrast between brand name and background is found on the "extra" pack. It is on this pack that the faster emergence time for brand name is observed. Also, although the brand name on the "super" pack begins to emerge at approximately the same time as on the "mild" pack, it takes longer to be able to read it. However, the difference between the results for these two packs could be accounted for b~ such factors as the font style and positioning of the brand name on the pack. CD N-) %10 %10 1986 B.A.T I L: K and Export) Limited. This report must not be copied or shoun to unauthorised persons BATCo document for Legal Services : Health Canada 20 October 1999 -41- (d) Brunette The findings from the Brunette packs (Figures 47, 48 and 49) suggest that when individuals are presented with stimulus material they look for design detail and it is the most visually prominent design detail which stands out first. The brand name on all three packs is presented in exactly the same colour, font style, position, and on the same colour background. However, on Brunette Extra and Brunette No 3 it is the background which is seen first but on Brunette Double Filter it is the brand name itself. It is suggested that the reason for this is that the background on the former two packs form part of the design detail which in itself is more prominent than the brand name. On the Double Filter pack, on the other hand, the background is not a design detail but part of the overall colour of the pack. (e) Muratti Ambassador A comparison between Muratti Multifilter (Figure 50) and Muratti Extra (Figure 52) tends to confirm that a dark figure presented on a light background is more visually prominent than the reverse style of presentation. The brand name on these two packs is exactly the same, the font style is the same and so are the colour combinations used. However, on the "extra" pack the brand name is presented on a light background whilst on the "multifilter" pack it is presented on a dark background. This has the effect not only of making the brand name on the latter more difficult to see initially but also making it more.difficult to read. Drawing from the results on the Brunette pack, it is suggested that this effect is probably accentuated when the dark background on which the object is placed forms part of the pack detail itself. Turning now to the "2000" pack (Figure 51) it can be seen that the most visually prominent element on N-) this pack is the "2000" itself. It is interesting to CN 1996 13.A.T (U.K. and Expon) Limited. This report inust not bt zopied or sho-m to unauthortsed persons BATCo document for Legal Services : Health Canada 20 October 1999 -42- speculate as to the reason for this finding. One possibility is that it is presented in the colour red which, as we have discussed, is highly visually prominent. Another possibility is that the use of the double edged lettering significantly contributed to the effect. M Sel ect The now familiar contrast effect is once again seen to be operating on the Select packs (Figures 53, 54 and 55). Although it is not shown in the presentation of the results, the stripe which runs through the brand name had a masking effect which made it much more difficult to read. The difference in findings betdeen the No.2 and the No.3 packs reflects this masking effect. On the latter pack No.3 was seen and read quite quickly but the No. 2, which was placed on the stripe, was much more difficult both to see and to read. 4. (iii) DISCUSSION A methodology has been described where the problem of the learning effect when testing packs with the same brand name is counteracted by using matched groups of subjects. The subjects were screened for colour blindness and then matched on a standard eye - sight test and on recognition threshold scores. As a further precaution against the possibility that the subjects in any one group may perform better than those in another, the presentation of the packs was counter-balanced across the groups and between the individuals. Six sets of Swiss pack families were used in the study. The results of the study were reported only in so N.) far as they provide further information towards ON \0 guidelines concerning factors which contribute to *__0 \_0 0' 1986 B A.T (V K and EiLportl Limited This re;x)rt must not be copied or sho-n to unatithonsed persons ('~ BATCo document for Legal Services : Health Canada 20 October 1,999 -43- visual impact of pack design elements. The effect of contrast was reaffirmed and the masking effect of certain colours was highlighted. An additional finding of interest was the superiority, in visual impact terms, of placing a dark figure on a light background as opposed to a light figure on a dark background. This point is particularly relevant if the dark background on which the figure is placed forms part of the pack detail itself. 5. EXPERIMENT 4: THE EFFECTS OF FAMILIARITY ON PERCEPTION OF PACK DESIGN ELEMENTS - A CROSS CULTURAL STUDY The effect of familiarity on perception of pack design elements was briefly discussed in Experiment 1 where the results of smokers, non-smokers and ex-smokers were compared. It was hypothesised that smokers would be more familiar with the packs than ex-smokers who themselves would be more familiar than non-smokers. From the cross comparisons there was no suggestion that familiarity had any effect on either the order in which the elements emerged or on the overall recognition scores for the packs. In fact, it was often the case that non-smokers were, on average, quicker to recognise some of the packs than smokers. However, it was concluded that since all the packs were from the U.K. market and all the subjects were from the same market that this was not a true test of familiarity. This was because all subjects would have had some degree of exposure to the packs tested. The question of the effect of familiarity obviously assumes a great deal of importance in those situations where it is desired to compare the visual prominence values of those elements on a new pack design with those on packs already existing in the market. Although it would be more usual to use the tachistoscopic test in a situation where several new r\ pack design options are to be compared it is often the case U 0 1996 8 AT It; K and Export) Limited This report must not be copied or shown to unauthorised persons BATCo document for Legal Services: Health Canada 20 October 1999 -45- For the first 4 packs seventy two U.K. smokers and thirty Finnish smokers were tested. On the last three packs, however, only 30 U.K. subjects were tested. In the second stage the results of 20 U.K. smokers and 30 Finnish smokers are compared on each of the following packs taken from the U.K. market: 1. Benson & Hedges Middle Tar 2. Silk Cut Low Tar 3. Rothmans Middle Tar 4. Dunhill Middle Tar 5. State Express Filter Kings Middle Tar 6. State Express Medium Mild Low to Middle Tar In both stages the research with the Finnish consumers was commissioned by Suomen Tupakka Oy to IFH Research International Oy. The tachistoscope used ~y IN was to the same specification as that described in ,Appendix 1. The procedure adopted was as described in Experiment 1. One difference between the work conducted in Finland and in the U.K. was related to recognition of the Finnish packs. In both the U.K. and Finland the recognition score for the respective unfamiliar packs is actually the time taken to read the brand name. However with respect to the familiar packs, in Finland no distinction is made between being able to read the brand name and recognising it. In the U.K. study three scores are given for the brand name and they are: 1. The time taken for the name to emerge; 2. The time taken to recognise the pack; and 3. The time taken to read the brand name. The latter score represented the time at which the C:> subjects felt that they could read the brand name N.) even though they may have recognised the brand from the pack design previously. (D 1936 D A T (U.K. snd Export) Limitt:4- This report must not be copied or shown to unauthonsed persons BATCo document for Legal Services : Health Canada 20 October 1999 -47- Form Similar to the Belmont pack it can be seen from Figures 58 and 59 that the order of emergence of elements for both sets of subjects is exactly the same. Another feature that this pack shares with Belmont is the ease of readability of the brand name. Again a large majority of the U.K. subjects were able to read this unfamiliar brand name. Although the time taken to do this was longer than for Belmont. The brand name on the pack is well positioned as it is the most visually prominent element of the pack design. 0 North State Mild Fioures 60 and 61 show the results for North State Mild from Finnish and U.K. smokers respectively. The order in which the elements emerge is the same for both sets of subjects. However, an effect of familiarity can be seen in the results from this pack. In the case of the Finnish subjects the vast majority of individuals are able to recognise this pack as North State. However, the results from the U.K. subjects, where only half were able to read the brand name, suggest that in many cases the Finnish subjects were able to guess the brand from factors other than being able to read the brand name. The effect of familiarity on this pack, therefore, seems to be in relation to the recognisability of the brand rather than to the order of emergence of the elements. Kevytsavuke 5 and Ultra Kevyt 1 Kevytsavuke 5 (Figures 62 and 63) and Ultra Kevyt 1 (Figures 64 and 65) are very similar in design. The main feature-of the design is the brand identifer : 5 ON and 1 respectively. The use of numbers to convey the brand name in an advertising ban market like Finland ON 0 1986 D.A T (U.K and Export I Limited. This report must nm be copied or shL)%n to unauthonsed persons BATCo document for Legal Services : Health Canada 20 October 1999 -49- subjects seeing the gold border marginally (0.1 msec) faster than the number 1. Again the use of a number as the brand identifier seems to be successful in terms of identifying the brand since a high number of U.K. subjects were able to do this. However, the percentage of subjects correctly identifying the number 1 is lower on this pack for the U.K. subjects than it was on the Ultra Kevyt 1 pack. It is thought that this is due to the differences in shape of the two number 1's and to the fact that on the Kevytsavuke pack the figure ground relationship between the 1 and its background is obscured to some extent by the inclusion of the gold border in the background. It is also interesting to compare the visual prominence of the crest on this pack with the one on the Ultra Kevyt pack. Although the two crests are positioned in the same location on the pack, the crest on the Kevytsavuke pack takes much longer to be seen. The explanation for this finding lies in the relative lack of contrast between the crest and its background on the Kevytsavuke pack. Marlboro Lights The way in which the data for the Finnish subjects (Figure 68) and for the U.K. subjects (Figure 69) is presented would suggest that there is one difference between the two groups with respect to order of emergence of the elements. It appears that the copy "lights" emerges faster for the U.K. subjects than for the Finnish subjects. This, however, is due to the slight differences in procedure mentioned earlier. In the case of the Finnish subjects the scare given for "Lights" represents the time taken to read the word. For the U.K. CD subjects, however, two scores are given: one for the time r\.~ ON taken to see the copy "Lights" and the other for the time I-IC taken to read it. If the latter score only had been given, 11C 1986 B.A.T I L; K. and Expon) Limited. This report must not be copied or sho-n to unauthovised persons BATCo document for Legal Services : Health Canada 20 October 1999 -51- results it can be seen that only two subjects were able to read the brand name. This suggests that the font style and contrast provided by the brand name is not particularly readable. It also tends to suggest that many of the U.K. subjects would never have been able to read the brand name "Benson A Hedges" unless they had known that it was Benson and Hedges. A similar result was observed with respect to Silk Cut (Figures 16 and 71) where only one Finnish subject was able to correctly identify the brand. Again it is noted that the order of emergence of the elements between the two groups of subjects is very similar. However, in this case there is one slight difference between the results with the crest being seen before the Benson and Hedges copy by the Finnish subjects and vice versa for the U.K. subjects. Comparing the results for the two groups of subjects on the Rothmans pack (Figures 12 and 72) it can be seen that the order of emergence of elements is identical. It is also noted that,more Finnish subjects were able to read the copy "Rothmans" than were able to read "Benson and Hedges" or "Silk Cut". The results for Dunhill (Figures 20 and 73), State Express Filter Kings (Figures 28 and 74) and State Express Medium Mild (Figures 24 and 75) tend to be slightly more conflicting than the results on any other packs that have been researched. This tendency towards conflicting results was also noted when comparisons on these packs were made between the groups of U.K. subjects in Experiment 1. Comparisons between the results of the Finnish and the U.K. consumers suggest that for the more visually prominent elements the order of emergence is the same. However, as we move down the scale of visual prominence the order of emergence tends to vary slightly between the two groups. Reasons for this have been suggested C7N in Experiment 1. 1986 B A.T (U.K. and Expertl Limited. This report must not be copied or shown to unauthonsed persons BATCo document for Legal Services: Health Canada 20 October 1999 -53- Overall the results would suggest that, as far as European cultures are concerned, familiarity does not affect the relative visual prominence of pack design elements. Apart from a few exceptions (usually on low prominence elements) the order in which the design elements were seen by Finnish and U.K. subjects was exactly the same on both Finnish and U.K. packs. The one area where familiarity does play a part is on recognition/readability of the brand name. Although the order in which the copy of the brand name was seen was the same regardless of whether the subjects were familiar with the pack or not, the ability to read it was very dependent on familiarity. Benson & Hedges and Silk Cut are good examples of this type of effect. The ordering of emergence of these brand names was the same for both groups of subjects. However, many more U.K. subjects were able to read the brand names than Finnish subjects. If the U.K. subjects had not known what the brand name was then it is likely that they also would not have been able to read it. The cultural independence of perception of emergent detailing indicates that if readability of the brand name is of particular interest to the pack designer then within the westernised context at least it is quite feasible to test the packs in another market. This is because the findings reported here highlighf the fact that apart from the readability of the brand name, the results obtained on a tachistoscopic test in one market would be the same as those obtained in another market whether or not the subjects are familiar with the pack design. How far these results are applicable to non westernised markets where the reading style is completely different is a matter for further research. N; ON 1096 B A.T (U.K. and Export) Limited This repan must not be CoPied or sho-n it) unauthonsed pcrsons BATCo document for Legal Services: Health Canada 20 October 1999 -55- which the elements were recalled may correlate highly with visual prominence of pack design elements. Consequently, as well as noting whether or not an element was recalled, the order in which the elements were recalled was noted. 6 (ii) RESULTS For each cigarette pack a table shows which elements were recalled and how often, the order in which they were recalled (using the average rank scores) and the rank scores according to visual prominence (ds measured in Experiment I). (a) Marlboro TABLE 1 Average Rank Rank Scores Element No. of Times Scores According Recalled According to Visual to Recall Prominence Red Chevron 20 1.00 1 Marlboro 20 2.05 2 Crest 4 2.75 3 In contrast to the Benson and Hedges pack where all the elements on the pack were recalled by at least some of the subjects, only three elements on the Marlboro pack were ever recalled (Table 1). These were the red chevren. the brand name copy Marlboro and the crest. The findings of Experiment I indicate that these three elements are also the most visually ~-o -4N C) 1986 Et A T (U.K. &nd Export) Ltinoted This reporl must not be copied or %hawrt to unauthorised persons 0 BATCo document for Legal Services Health Canada 20 October 1999 -57- scores for visual prominence are somewhat displaced relative to those for recall. The first reason for this is that colour was not measured using ihe tachistoscope. However, if it had been measured it certainly would not have been the first aspect of the pack to be noticed since gold is not a highly visually prominent colour. The reason for this apparent discrepancy between the visual prominence of gold and it's cognitive salience as far as Benson and Hedges is concerned is most likely due to advertising influence. Gold is the theme of all Benson and Hedges advertisements in the U.K. , 0 The second discrepancy in comparing visual and recall order concerns the embossed "B & H" emblem. Visually this has more prominence than it has cognitively. However, it should be noted that although the emblem was mentioned before the crest and "Special Filter" it was actually recalled by fewer subjects. Apart from this, however, of the common elements measured in both experiments the results tend to support the hypothesis that visual prominence is related to recall salience. Rothmans TABLE 3 0 Average Rank Rank Scores Element No. of Times Scores According Recalled According to Visual to Recall Prominence Blue Central Area 19 1.00 1 Rothmans is 2.20 2 Crest 8 2.43 4 ~10 ~10 1996 B A.T 4 U.K. and Export) Limited. This report must not be copied or sho~n to unauthonsed persons _c::. CD BATCo document for Legal Services : Health Canada 20 October 1999 -59- contrary to the tachistoscopically produced hierarchy of elements In terms of visual prominence. If it is hypothesised that the most visually prominent elements are recalled first and more often then, based on the results of Experiment 1, it would be expected that "Benson and Hedges" would fare better than "Silk Cut". It is suggested that one of the contributing factors to this result may have been due to the use of familiar brands. The more appropriate experimental design would have been to show the subjects hitherto unfamiliar brands for a specified period of time and then to ask them to recall them. In this experiment, however, the subjects were required to recall the brand "Silk Cut". This type of task obviously may enhance recall of the brand name over the above small detail such as "Benson and Hedges". (e) Dunhill TABLE 5 Average Rank Rank Scores Element No. of Times Scores According Recalled According to Visual to Recall Prominence Red Pack 20 1.10 2 Gold Outline 18 1.94 1 Ounhill' 18 2.89 3 Crest 2 3.00 4 4D 1986 B.A.T (U.K. and Exporil Limited This report must not be copied or shun to unauthonscd persons BATCo document for Legal Services: Health Canada 20 October 1999 -61- TABLE 7 Average Rank Rank Scores Element No. of Times Scores According Recalled According to Visual to Recall Prominence Dark Circle 7 1.43 1 555 12 1.50 4 Yellow Pack 3 2.00 - Blue Chevron 4 2.25 2 Fluting 4 2.25 3 Gold Chevron 2 2.50 5 State Express a 2.63 6 From Tables 6 and 7 it can be seen that both State Express packs simulate contradictory recall data. For the Medium Mild pack at least half the subjects never recalled any pack detail whatsoever. One of the reasons for this may have been due to lack of familiarity since this pack had been withdrawn from the U.K. market for at least six months by the time this experiment was conducted. Of those subjects who did remember some design aspects of the State Express 555 pack (Table 6) it can be seen that the blue chevron was recalled more often than any other element with a tendency for it to be recalled first. However, apart from this fairly consistent result the other elements were not only recalled just a few times but the order in which they were recalled was highly variable. 1996 B.A.T (U.X. and Export) Limited. This report must not be copied or shon to unatitticinsed persons BATCo document for Legal Services : Health Canada 20 October 1999 -62- The lack of consistency in the recall results is also apparent in the data for the 555 filter kings pack. Although the dark circle tends to be recalled first, "555" was recalled more often. However, as indicated for the Silk Cut pack, it is open to conjecture whether the relatively high recall of this element was stimulated by the type of task. This comment is also relevant to the relatively high recall rate seen for the logo "State Express". Apart from these three elements the recall rate was comparatively low for the design elements on the pack. Additionally, it can be seen from theaverage rank scores that the order in which the elements were recalled was highly variable between subjects. As well as a failure to find consistency in order of recall of the design elements from these packs it was noted in Experiment I that comparisons between the experimental sub-groups (Smokers, Ex- smokers, and Non-smokers) revealed that there was little consistency in terms of their ordering in visual prominence terms. It was suggested that this result may be related to the fact that many of the elements on these packs are all of equal, low visual prominence. It is possible that corresponding to this the elements all have equal, low cognitive salience and it is for this reason that order of recall is highly discrepant between subjects. 6. (iii) DISCUSSIGN This experiment represents a preliminary enquiry into the relationship between visual prominence and cogni tive salience. The underlying theme of this type of investigation is to address the question as C:D to whether or not the achievement of high visual NO CYN prominence of particular important elements within 0 1936 BA.T (U K. and E,portj Limited, This repon must not be copied or shu-n to unauthonsed persons CD BATCo document for Legal Services: Health Canada 20 October 1999 , -63- a pack design will be associated with the consolidation of that information (whether it be a symbol, brand name, corporate crest etc.) by an individual. The experiment reported here can only be regarded as a preliminary investigation because of the ad-hoc manner in which it was designed. The U.K. packs used in the tachistoscopic Experiment I were tested on a number of U.K. Smokers. There are several confounding variables which may cause pit-falls in this type of approach where subjects are tested on already existing brands from the market place. The first uncontrollable variable is due to the effect of familiarity. Some of the subjects tested may be more familiar with the brands than others. It was suggested, for example, that the results relating to "555 Medium Mild" may be related to the fact that this brand had been withdrawn from the market at least 6 months before this particular experiment was conducted. In the situation where the subjects are tested on novel pack designs (i.e. modelling a new launch situation) the confounding effects of familiari ty would be negated and a more accurate reflection of the relationship between visual prominence and recall salience would be gained thereby. One further finding which may have been due to the effect of familiarity was related to the recall of crests. Invariably a crest was recalled by at least one individual for each pack. It is suggested that this may have been due to a minority of individuals believing (quite rightly) that all cigarette packs in the U.K. feature a crest rather thin actual recall of any particular crest. 1 0, NIC 1986 8 A T W K. and Export) Limited. Thks report must not be copied or sho-n to unauthorised persons BATCo document for Legal Services: Health Canada 20 October 1999 -65- It is also interesting to note that the greatest inconsistencies between the subjects in terms of order of recall were found an those packs where there had been difficulty in obtaining consistency in terms of rank ordering of visual prominence of the elements between the sub-groups in Experiment 1. Additionally, on these packs (State Express 555 Filter Kings and State Express 555 Medium Mild) although a large number of elements were recalled in total each element was actually recalled by very few subjects. This finding is contrary to the results of other packs where a small number of highly visually prominent elements were recalled by a large number of subjects. It was suggested that the reason for the above finding may be that low visual prominence elements correspondingly achieve low cognitive salience. 7. SUMMARY AND CONCLUSIONS One of the growing problems facing the tobacco industry in developed markets is the increasing restrictions on advertising. This is exacerbated in the context of a gradual move on behalf of the industry to develop innovative products. On this basis it is perhaps more important than ever that communication with the consumer is maintained. 0 Given the above scenario the cigarette pack itself assumes a great deal of importance since this is likely to be the last means of communicating with the consumer to be affected. As far as pack design itself is concerned, there are two aspects which are important and they are brand imagery and visual impact. With the luxury of advertising most of our effort in pack design has been on the side of imagery. Thus, the imagery created using the advertisng media is carried right through to the pack design itself. Brand imagery has now been developed to such an extent that as 1986 B.A.T I U K. and Export) Lmsted. This report must not be Copied or shu-n to unauthoriscd persons BATCo document for Legal Services : Health Canada 20 October 1999 -67- The aim of the research described in the report has been to develop the methodology involved in tachistoscopic testing. However, the results have been used to establish some of the principles involved in the design of cigarette packs in order to achieve visual prominence of the elements. In the first experiment the precise methodology involved in tachistoscopic testing was outlined. Sixty U.K. subjects were tested on seven U.K. packs. The subject sample was broken down into three groups of twenty smokers, non-smokers and ex-smokers. The sub division of the samples allowed a cursory look at the effects of familiarity in this type of task. It was hypothesised that smokers would be more familiar with the packs than ex-smokers who themselves would be more familiar than non-smokers. It was predicted that if familiarity did affect performance it would take one of two forms. Firstly, it was felt that if consumers recognised a pack they would describe the elements of the pack design as they recall them rather than as they saw them. Secondly, it was predicted that familiarity would enhance the recognition scores on these packs. The results were discussed in terms of the familiarity effect and of the factors which contribute to the visual prominence of the design elements. As far as familiarity was concerned it was concluded that in this sample there appeared to be no effect at all. Comparisons between the subgroups presented no evidence to suggest that the more familiar subjects saw the pack design elements any differently than the less familiar subjects. The order of emergence of the elements between the groups was very similar. Basically, the design elements on the packs tended to fall into three categories. There were the highly visually prominent elements that emerged after approximately three msecs, the less visually prominent elements that emerged after 6 msecs and, lastly, the elements with low visual prominence which took approximately 12 msecs to emerge (in the case of the latter these elements would probably never be seen with just a quick glance at packs on the point of sale and therefore they should never carry any important detail). It was found (D 1986 B A T W K and Expon) Urnacd. Thts report rpust not be copicd or shoun to unauthorised persons BATCo document for Legal Services : Health Canada 20 October 1999 -69- Since tachistoscopic testing of pack designs is a previously unexplored area the second experiment described in this report was designed to test the effect of varying the level of illumination on the stimulus field. As far as the tachistoscope itself is concerned this is the only parameter which is variable. In Experiment 1 the level of illumination was 1000m which gives 1000 lux on the stimulus field and 5 lux at the subject's eye. This level was chosen because it approximately represents the level of illumination found on the shelf of a point of sale gantry. In Experiment 2 the level of illumination was turned down to 50% which gives 500 lux on the stimulus field and 2.5 lux at the subjects eye. The procedure adopted was exactly the same as in Experiment 1. Thus, 60 naive subjects were tested on the seven packs described in the first experiment and the results were compared. It was found that the major effect of reducing the level of illumination was to considerably increase the amount of time taken to see each element. This effect was less pronounced on the least visually prominent elements. It was hypothesised that this result was a.consequence of the fact that lowering the level of illumination reduced the contrast effect necessary to see the more visually prominent elements. This result has obvious implications as far as merchandising in the instore environment is concerned. The third experiment in the series was designed to deal with one of the practical applications of tachistoscopic testing. It is anticipated that the main function of tachistoscope testing will be in the pre-launch phase where the brand name has already been selected but there are several possible design options. A typical question will be concerned with which font style best conveys the brand name. It was noted in Experiment 1 that of the two State Express 555 pack versions, the one that was presented second was recognised more quickly. Thus a learning effect was operating in this experiment. In Experiment 3 a methodology is described where the learning effect is dealt with by using matched groups of subjects. 011 %~c 11C 1q$6 B.A T (U K. and Export) Limited. This report must not be copied or shuun to unauthortsed persons BATCo document for Legal Services : Health Canada 20 October 1999 -71- It has been stated that the tachistoscopic test should not be used to make comparisons between packs in order to achieve a relative overall measure of visual impact. However, it may be the case that in designing an individual pack the objective is to make the brand name stand out on that pack better than it does on an already existing pack on the market. A situation may also arise where it is desired to gain some understanding of how other packs on the market work in visual impact terms before any new designs are embarked upon. In these situations the possible biasing effects of familiarity on performance could provide distorted results. In order to test the familiarity effect a cross cultural comparison was conducted between Finland and the U.K. A complete cross over design was employed where both Finnish and U.K. subjects were tested on both Finnish and U.K. packs. The findings indicated that as far as emergent detaiiing was concerned, familiarity did not affect the results. However, familiarity did affect readability of the brand name. Although the order in which the copy of the brand name was seen was the same regardless of whether the subjects were familiar with the pack or not, the ability to read it was positively affected by familiarity. However, the fact that perception of pack design elements is, within the westernised context, culturally independent means that if brand name is of particular interest then the packs can be tested in another market. How far these results are applicable to non-westernised markets, where the reading style is completely different, is a matter for further research. The final experiment outlined in this report (Experiment 5) was designed to assess the correspondence between the visual prominence of pack design elements and their cognitive salience. The procedure involved taking the brands used in Experiment I and asking subjects to describe from memory GN their design detail. This experiment was ad-hoc in nature ~10 1986 B A T (U X. and Expun) Urnsted This report must not be copied or shown to unauthorised persons BATCo document for Legal Services: Health Canada 20 October 1999 -72- and is consequently confounded by a number of pitfalls such as the style of presentation of the task, the biasing effect of advertising on performance on this type of task, and the effects of familiarity. Despite all it's drawbacks, however, the results of the experiment are encouraging and indicate that, generally, the elements recalled by consumers are those that achieve high visual prominence scores. FUTURE DEVELOPMENTS The technique described so far has utilised existing Packs from the market place. In order to extend the concept of tachistoscopic testing so that it is more versatile in terms of the cost of pack variant 'mock-ups', a facility is under development for testing new designs using a high resolution computer graphics system. With this system it is not necessary to go to pack mock-up stage. Using a micro-computer, the system allows you to paint on a high resolution display in any one of 64 colours. With the aid of a software package, anything that can be drawn on paper using a pen can be reproduced on the display. Once a new pack has been designed it can be photographed off the display and tested in the conventional tachistoscope in the,usual way. Further research in the area of testing for visual impact will mainly involve testing the relative prominence of one pack against another within the merchandising context. A method using a projection tachistoscope is being developed. A projection tachistoscope is in principle the same as a conventional tachistoscope except that when using this piece of equipment the stimuli are projected onto a screen. This means that the size of the stimulus is only inhibited by the size of the projection surface. Consequently, it is possible to project a point of sale display onto the screen for successively increasing intervals of time. Controlling for 01- 110 pack layout and position on the display, the relative prominence. 0 1986 B.A T JU K. and Expon) Limited. This report must not be copied or shoun to unauthurised persons BATCo document for Legal Services: Health Canada 20 October 1999 -73- of target packs can be measured in terms of the time it takes to locate them. Another approach to enhancing the prominence of own brands at the point of sale is to place them in the most visually impacting areas. Eye gaze monitoring techniques are currently being developed to allow us to determine where these areas are. These researches will be subject of formal reporting in due course. CN I--0 %10 1956 B.A.T (U.K. and Exporti Umned. This report must not be copied or shown to uniuchonsed persons BATCo document for Legal Services : Health Canada 20 October 1999 0 APPENDIX 1 0 0\ I---0 11.0 1986 B.A.T (U.K. and Expun) Limited. This report must not be copied or shown to unautthmsed persons BATCo document for Legal Services: Health Canada 20 October 1999 -74- PORTABLE TWO FIELD TACHISTOSCOPE General Description: The Tachistoscope used in this study is a Two Field Uld) model for displaying material presented on cards, having an active viewing area 7J " wide X 4J " high (190mm X 114mm) - The cards themselves should be cut from 0.030" thick stiff white card as follows: Fld. 1. card: 8V approx. (210mm) X 8" exactly (203.2 mm) Fld. B. card: 10J" approx. (266.7mm) X 5" exactly (127 mm) 0 The cards are presented at a distance of 20" (508 mm) from the subjects eyes. The visual angle subtended at the eyes is approximately 21* horizontally and 13* vertically. Small solid objects up to 7" wide (180 mm) 4" high (100 mm) and 3" thick (75 mm) may be accommodated in the Fld. B. lamp compartment on a small wooden block. The Fld. B. card holder has four terminal nuts to allow access for this. Each field is illuminated by a fluorescent lamp which is circular for even illumination. The adaptation field (mirror image) Is seen by reflection in a surface coated glass half-mirror. The front of the Tachistoscope may be raised for easier viewing by means of a pair of lifting feet. The adaptation field card is slid into a card holder beneath the instrument. The stimulus card is fitted into a card holder at the back of the instrument. to The card illumination intensity is adjustable over the range 250-1000 lux giving 1.25 - 5.0 lux approximately at the subjects' eyes. The instrument employs silicon semiconductors and integrated circuits throughout. To ensure cool running of the unit the fluorescent lamps are switched on from a high voltage supply and maintained from a lower voltage one. A comprehensive CD range of programmes is obtainable, including repetitive and r1o hold operation. 1996 B.A T (U K and Expurti Limited, This repon must not be copied or shu~rk to unauthonsel persons BATCo document for Legal Services : Health Canada 20 October 1999 -76- For programme checking each Timer has a LED pilot lamp showing when an output pulse is being delivered by the Timer.. Field Lamp Drivers: Each Field Lamp Driver has two DRIVE sockets and two INHIBIT sockets. The DRIVE sockets will turn the lamp driver on for as long as + 5 V is present. The INHIBIT sockets will hold the lamp driver OFF for as long as + 5 V is present even if DRIVE is simultaneously present. The threshold is + 2.5 V. The + 5 V socket is provided for patching into a lamp DRIVE to hold that lamp driver on. Next to the Lamp Driver sockets are mounted the lamp INTENSITY controls. They control the intensity of the illumination in each of the Optical Unit lamp compartments over a range of 4/1. Each lamp driver has a neon lamp to indicate the state of its associated fluorescent lamp in the Optical Unit. Programming: To obtain a sequence of timed operations it is necessary to interconnect the Timers and also the Start Pulse Circuit. Timer interconnection is by connecting the INVERSE OP socket of one Timer to the TRIG socket of the next. The first Timer input may be fed from the Start Pulse circuit START OP socket for single shot push button operation or from the INVERSE OP socket of the second Timer for continuous cyclic operation. Connection from Timer to Field Lamp Driver to switch ON fluorescent lamps for the duration of the timed period is by connecting a Timer OP to a Field Lamp Driver DRIVE. Connection of a Timer OP to a Lamp Driver to turn the field lamps OFF for the duration of a timed period may be made by first connecting the + 5 V socket to a lamp DRIVE socket. 1986 BAX i U K. and Export) Limited. This report must not be copied or shu-n to unauthonsed persons BATCo -document for Legal Services: Health Canada 20 October -1999 -78- Field illumination: When fitted with U.S.A. circular daylight fluorescent lamps, the field illumination is adjustable over the range: 250 - 1000 lux (approximately) The corresponding luminous flux density at the subjects' eyes using white stimulus cards is: 1.25 - 5.0 lux (approximately) Light Output Rise & Fall Times; Rise time: 200 microseconds to 90% Fall time: Less than 200 microseconds.to 15% followed by fall from 15% to zero in 20 mS. Availability The stockists in the U.K. are W.C.R. and B.I. Withers Ltd., Platts Eyot, Lower Sunbury Road, Hampton, Middlesex. The current list price for the portable two field tachistoscope is f640 + VAT. Purchasers from abroad may have to pay an extra charge to cover packaging, insurance and freight. €-r- 0 1986 8 A T (U K. and Export i Limited. This report must not be copied or shown to unauthorised persons BATCo document for Legal Services: Health Canada 20 October 1999 0 APPENDIX 2 0 'D 1996 B.A.T (U.K. and Export) Limited- This report must not be copied or sho-n to unaughorised persons BATCo document for Legal Services : Health Canada 20 October 1999 W > 1 0 0 -79- CL 0 TABLE 9 CD SUBJECTS SCORES ON SNELLEN CHART AND RECOGNITION THRESHOLDS PqL S S S u u u Recognition 8 Recognition B Recognition Snellen Threshold j Snellen Threshold i Snellen Threshold CO) (D E E E I C Ri ght Left (msec) C Ri ght Lef t (msec) C Right Left (msec) C7 T Eye Eye T Eye Eye T Eye Eye ir CD S S S 3 4 4 3.2 2 4 4. 3.0 3 4 4 3.0 CD t: 4 4 4 3.4 5 4 4 3.4 6 4 4 3.4 X 7 4 4 3.8 8 4 4 3.8 9 4 4 3.6 Cr 10 4 4 4.2 11 4 4 4.2 12 4 4 4.0 0 1 13 4 4 6.0 14 4 4 6.0 Is 4 4 6.0 16 4 5 4.2 17 4 5 3.6 18 4 5 3.2 fu 21 CL $ 19 5 4 3.0 20 5 4 3.2 21 5 4 4.0 22 4 4 3.0 23 4 4 3.0 24 4 4 3.0 C) Fr 25 5 4 3.4 26 5 4 3.4 27 5 4 4.2 0 0 28 4 4 4.4 29 4 4 5.6 30 4 4 5.6 0 31 4 4 2.2 32 4 4 3.2 33 4 4 3.4 " I Cr 0 CD 34 9 4 3.8 35 9 4 4.8 36 9 4 6.0 to / 1 WM 1 0 APPENDIX 3 Or's %10 1986 8 A.T cU.K. and Expon) Limi(ed. This report must not be copied or shown to unautbonsed persons Cc. BATCo document for Legal Services : Health Canada 20 October 1999 0 C TABLE 10 GROUP 1 0 Pack s Subjects 1 2 3 4 5 6 ®2 A B C A C ta Q~ Marocaine Extra *M L Extra Parisienne Extra Brunette Extra Muratti 2000 Select i M 0 i Œ 4 A C 6 A C B 1 a Marocaine Extra M L Naturel Parisienne Mild Brunette Extra Muratti AMD EX* Select No.3 0 r 7 8 A C B A C CD 2 Marocaine Super M L Doppfilt* Parisienne Extra Brunette No.3 Muratti AMD* Select 10 C A C A --i Marocaine Super M L Naturel Parisienne Super Brunette No.3 Muratti AMB EX Select No.2 2* M Il 13 C B A C A CD lb Marocaine Mild M L Extra Parisienne Super Brunette D F* Muratti 2000 Select No.2 16 C A B C A B Marocaine Mild M L Doppfilt Parisienne Mild Brunette D F Muratti AMB Select No.3 e 19 A B C A C 0 0) Cr Marocaine Extra M L Extra Parisienne Extra Brunette Extra Muratti 2000 Select 1 22 A C A C 1¯ 1 -2. Marocaine Extra M L Naturel Parisienne Mild Brunette Extra Muratti AMB EX Select No.3 CL i gi. 25 B A C B A C Marocaine Super M L Doppfilt Parisienne Extra Brunette No 3 Murattl AMB. Select r,43 1 :r 28 B C A B C A Marocaine Super M L Naturel Parisienne Super Brunette No 3 Muratti AMB EX Select No.2 Il 0 i r 31 C 8 A C A f.> i C Marocaine Mild M 1. Extra Parisienne Super Brunette D F Muratti 2000 Select No.2 C 0 34 C A B C A Cr Marocaine Mild M L Doppfilt Parisienne Mild Brunette D F Muratti AMD Select No.3 CD *M L Marylong w *Doppfilt - Double Filter to *0 F - Double Filler *Ambex = Ambassador Extra *AMB = Ambassador 7, 0 0 TABLE 12 GROUP 3 CD Packs Subjects 1 2 3 4 5 6 r; C A B C A 8 (a Marocaine MIld M L Doppfilt Parisienne Mild Brunette D F Muratti AMB Select No.3 1¯ a 6 . 6 A C B A C r ta Marocaine Super M L Doppfilt Parisienne Extra Brunette No.3 Muratti AMB Select 9 . C s A C 8 A Marocaine Mild M L Extra Parisienne Super Brunette 0 F Muratti 2000 Select No.2 Z Er 12 A B C A C CD Marocaine Extra M L Extra Parisienne Extra Brunette Extra Muratti 2000 Select 0 A C B - A C B a Marocaine Extra M L Naturel Parisienne Mild Brunette Extra Muratti AMB EXT Select No.3 A C A CD CM Marocaine Super M L Naturel Parisienne Super Brunette No.3 Muratti AMB EXT Select No.2 21 C A C A B Marocaine Mild M L Doppfilt Parisienne Mild Brunette D F Muratti AMB Select No.3 24 A C A C C) - Marocaine Super M L Doppfilt Parisienne Extra Brunette No.3 Muratti AMB Select 27 C A C B A Marocaine Mild M L Extra Parisienne Super Brunette D F Muratti 2000 Select No.2 CD : 1 0 30 A 8 C A B C Marocaine Extra M L Extra Parisienne Extra Brunette Extra Muratti 2000 Select e 33 A C A C B 9 Marocaine Extra M L Naturel Parisienne Mild ' Brunette Extra Muratti AMB EXT Select No.3