RESEARCH LABORATORY REPORT NO. 183 COPY No. :~- IMPERIAL TOBACCO LIMITED RESEARCH & DEvELOpMENT DIVISjOrj MONTREAL RESTRICTED 0 A SUMMARY OF BRAND IMAGERY STUDIES ON CANADIAN PRODUCTS Author: C. McBride Issued by: Dr. S.R. Massey Date Issued: September 1987 DISTRIBUTION: Dr. P.J. Dunn COPY No. 1 Dr. S.R. Massey 2 Mr. A.L. Heard Dr. R. Binns 3 Mr. L. Reynolds 4, 5 Mr. R.G. Nicholls 6, 7 Herr. E. Rittershaus 8, 9 Dr. E. Kausch Dr. C-3-P. de Siqueira Library 12 13, 14, 15 -Cb. CD Ni (INI BATCo document for Legal Services : Health Canada 22 October 1999 TABLE OF CONTENTS Lue Summary Introduction 2 1. Methodoloov 3 1 .1 Delta Panel 3 1.2 Cigarette Selection 4 1.3 Experimental Design 4 1.4 Data Analysis 4 2. Discussion 7 2.1 Experimental Design 7 2.2 Individual Studies: a Brand-Imagery Study I a Brand Imagery Study 2 13 Brand Imagery Study 3 19 2.3 General Conclusions 24 3. Recommendations Concerning Brand Imagery Studies 24 0 Ni BATCo document for Legal Services: Health Canada 22 October 1999 SUMMARY The purpose of this report Is. to summarize some of the results and the conclusions drawn from the three brand imagery studies conducted at ITL since 1985. General methodology is outlined, experimental design is discussed in detail. Generalizable conclusions for the Canadian market arising from the three studies are presented. Recommendations concerning the technique are also included as well as brief comments concerning future work in this area at ITL. KEYWORDS: Subjective Evaluation Brand Imagery Delta Imagery Pack Design -1~:. (Z r\j (J-J BATCo document for Legal Services : Health Canada 22 October 1999 INTRODUCTION Consumers purchase tobacco products with specific names and package designs, that is. they buy brands. Brand name and pack design are usually extensively researched to ensure that the connotations associated with each and with the combination elicit the appropriate expectations and responses from the target consumers. Product alone is also extensively reseirched and often undergoes long and detailed product development. Ultimately however. the viability of a brand is not based solely on the acceptability of the Individual components: product, name and packaging, but on the acceptability of their combination. In brand Imagery studies, the contribution of the Individual components of a brand to the subjective evaluation of the brand as a whole are investigated systematically In order to provide information about this synergism. Since 1985, three separate studies concerning the effects of product name and pack design on the subjective evaluation of cigarettes have been conducted at ITL. It is the purpose of this report to present an overview of these studies, to illustrate some of their results, to summarize their individual conclusions and to point out generalizable conclusions. The report is divided Into sections discussing methodology common to the three studies and results and conclusions for each. Overall conclusions and a general discussion of brand imagery exercises are presented separately. CD 11V r\J BATCo document for Legal Services : Health Canada 22 October 1999 - 3 - I. METHODOLOGY In general, within the tobacco industry the objective of any Brand Imagery study is to investigate the effects of some of the elements of imagery, i.e. brand name and packaging on the sensory evaluation of particular tobacco products. However, an individual study will also have a specific objective which will be reflected In Its experimental design. The specific objectives of each of the Brand Imagery studies described In this report are given in Table 1. The methodology and experimental design employed to fulfill these objectives are discussed in the following section. Throughout this report, the three Brand Imagery studies are referred to as B.I. #1 (M. ay 1985), B.I. #2 (February 1986) and B.I. #3 (January 1987) following the chronological order of the internal reports (1-3). The methodology outlined in the following sections is based primarily on the initial work by Ferris (4). The Delta panel subjective evaluation technique, which is an integral part of Brand Imagery studies, has been described by Ferris in a number of BAT reports (4-6) and by ITL researchers In internal reports (1-3). ft will not, therefore, be presented in detail here. 1.1 Delta Panel In all three studies, 9 members of the ITL in-house Delta panel assessed 8 products of interest. Within a single study. the same panelists evaluated the products under all presentations. However, the same panelists were not necessarily involved in all three Brand Imagery studies. The product attrib u tes used by the panelists to describe the products are given in Table 2. Lri BATCo document for Legal Services : Health Canada 22 October 1999 - 4 - 1.2 Cigarette Selection In accordance with the Delta technique, all cigarettes used in the Brand Imagery studies were selected to a mean value (n = 100) for weight (+ 15 mg), pressure drop ( 3 mm) and when applicable, ventilation ( 3%). The cigarettes evaluated in each study are given in Table 3. All cigarettes when presented masked, were masked with coloured tape such that the logo was not visible to the panelists. 1.3 Experimental Desiqn In Brand Imagery studies. experimental design refers to the way in which the products to be evaluated are presented to the panelists. The presentations associated with each of the three Brand imagery studies are given in Table 3, as well as the time frame in which each study was completed. Each test was separated by at least one week. In the interim week(s), panelists assessed a cohort of products not related to Brand Imagery studies. This was to control for memory effects. A more detailed discussion of experimental desIgn Is presented in the Discussion. 1.4 Data Analysis The data was analysed I.n a similar fashion for all the Brand Imagery studies and Is Illustrated in the flowchart given in Table 4. The names of the computer programs are given in brackets in the following discussion. 1) Within Presentation Tests 2j Between Presentation Tests rQ (-P4 ,--,I ON BATCo document for Legal Services : Health Canada 22 October 1999 - 5 - Within Presentation Tests The consistency and reliability of the data within each presentation was determined by comparing the product evaluations of the individual panelists to the group product evaluation both by product and by attribute (DELTA). I This analysis provides an indication of the cohesiveness and hence reliability of the panel. Comparison by product Indicates the inter- panelist variation in perception of the product, whilst comparison by attributes indicates the inter-panelist variation in comprehension, and therefore use, of the product attributes. The results of these analyses take the form of correlation coefficients. Acceptable levels of correlation, indicating a suffictently cohesive panel must be at least 0.50-0.60 for overall correlation between an individual and the group and at least 0.40 between an Individual's evaluation of a specific product and/or use of a specific attribute and the group. Once an acceptable level of panel cohesiveness has been established, the relationships among products, attributes and between products and attributes are determined using a form of Principal Component Analysis (INGRID). This Is a two part analysis, the first part of which establishes the boundaries of the perceptual space and assesses the validity with which the relationships among products and attributes within that space may be expressed as correlations. The second part of the analysis details the structures of these relationships. The criterion for validity Is fulfilled when at least 70% of the variation In the perceptual space is accounted for by the first two principal components. Results take the form of correlation matrices and two dimensional graphs or maps. C:D K) BATCo document for Legal Services : Health Canada 22 October 1999 - 6 - Between Presentations Tests As a first step, the attributes from each presentation were compared to assess the degree of consistency in their use (BANNISTER). This analysis determines whether the product attributes are used consistently by the panelists across all presentations. The results of this analysis take the form of correlation coefficients. An acceptable level (r .9 0.80) indicates that the use of the product attributes has not changed over time or due to the Introduction of new variables. Therefore, any change in product perception can be attributed to a true shift In the evaluation of that product and not to a shift in the scales (product attributes) used to describe that product. Given an acceptable level for the BANNISTER SCORE, different presentations were compared using DELTA. If the presentations were replicates, the DELTA comparison was used to assess the stability of the evaluations, as in the case of Masked replications. This is an Important consideration since the Masked evaluations act as the baseline against which all other presentations are compared. Masked and Different Presentation evaluations were also compared using DELTA. rn this case, If the level of correlation between Masked and any other presentation differed by at least 0.15 from the level of correlation between the Masked evaluation replicates, the effect of a brand Imagery variable was assumed to exist. Finally, the data from the different presentations were analysed using INGRID or ADELA* to define the relationships between product and attributes. In this way, the effect of a brand Imagery variable on the perception of a product could be assessed. ADELA is a computer program similar to INGRID. However ADELA Is specifically designed to analyse large asymmetric matrices. co BATCo document for Legal Services : Health Canada 22 October 1999 - 7 - 2. DISCUSSION 2.1 ExDerimental Design As previously noted, for the purposes of Brand Imagery studies ' , experimental design refers to the way in which products are presented to the panelists for evaluation. To fulfill the specific objective of any one of the studies described in this report, the experimental design of each must be closely followed with regard to replicate presentations and order of presentation. The specific objectives for each study are given in Table 1. the presentations and order of presentation in Table 3. Additional Information is presented in Table 5, outlining the specific purpose of each presentation. The most important presentation, common to all the Brand Imagery studies, is the Masked evaluation. This particular presentation must always be replicated as it serves as the baseline measurement for the product alone. A comparison of these replicates establishes the level of stability for the evaluation of the product without the Influence of any variables. If the Masked evaluation is stable, i.e. consistent over time, then any changes in the evaluation of the same product under a different presentation must be due to the introduction of a new variable. If the 'hen changes In product perception under Masked evaluation is not stable It different presentations cannot be reliably attributed to the introduction of a new variable. For these reasons, the Masked evaluation is always the first to be presented to the panelists. , The Empty pack evaluation was presented next in order to have as clear an indication of smoker expectations as possible without the influence of any product-pack interaction. Replication of this presentation is optional. However, the replicates should be separated by a considerable space of tipe since this presentation would be most susceptible to memory biases. N) %-0 BATCo document for Legal Services: Health Canada 22 October 1999 - a - Arguments can be made to order the remaining presentations in a variety of ways. However. the order followed in these Brand Imagery studies was selected to introduce new variables in a systematic fashion. 2.2 Individual Studies Throughout this discussion, it can be assumed that all criteria for reliability, as outlined In the section concerning Data Analysis, have been met In each of the Brand Imagery studies discussed in this report. All of these studies have been reported in greater detail in a series of Internal ITL reports (1-3). The specific objectives, results and conclusions will be presented for each study. However, the individual correlation matrices will not be given. Results selected to illustrate the conclusions of each study are shown in the form of graphs depicting the degree of association between a product and the attributes as Influenced by the different presentations (Figures 1-11). The degree of association between a product and any attribute may range from -1.00 to +1.00, Indicat4 tng a strongly negative or extremely weak association and a very close, strongly positive association respectively. As a general rule, the position of a line on the graph relative to the masked evaluation indicates the shift or lack of shift in the perception of a product caused by the introduction of an imagery variable. A product perceived as stronger than its Masked evaluation will be associated with a line above that of the Masked evaluation whilst a product perceived as milder will be associated with a line below that of the Masked evaluation. Brand Imagery Siudy I (B.I. P. May 1985)- The specific objective of this study was to explore the possible influence of brand identification and pack imagery on the evaluation of Canadian products assessed by the ITL Delta panel. There had been some thought that due to the homogeneity of the Canadian market (approximately 98% of Canadian products are 1001A Canadian flue-cured tobacco) differences in subjective evaluation due to imagery variables might not be detectable or would not exist. L-14 X::- r1_J BATCo document for Legal Services : Health Canada 22 October 1999 - 9 - To fulfill this objective, panelists assessed masked products on the first two occasions, unmasked on the third and unmasked products in their own packs on the last occasion. The Masked evaluation was repeated twice to assess overall baseline, stability (Table 6). From the information In Table 6, It is evident that all but two products, Rothman's Extra Light and Peter Jackson Extra Light, were evaluated consistently in masked format (correlations for Masked 2 1 0.60). These two products were retained in the assessment cohorts but not analysed in terms of brand imagery effects. It can also be seen from Table 6, that the correlations between the different presentations and the baseline were extremely high for Matinie Slim (0.97-0.99) and Player's Special Blend Light (0.92-0.94). It was noted in the dicussion of the Data Analysis, that a difference of at least 0.15 must exist between the correlation of the two Masked evaluations and the correlation of the first Masked evaluation and any other evaluation before a brand Imagery effect was assumed to exist. For Matin6e Slim and Player's Special Blend Light, there is no deviation from baseline stability and therefore, within this specific cohort, probably no brand imagery effect. Matinie Slim Is a cigarette with a circumference of 23 mm. Player's Special Blend Light was designed at that time to be a modified blend. Its recipe Included 15% Maryland and 3Y. Oriental tobaccos. The other products in the cohorts were all 25 mm circumference and 100% flue-cured Canadian tobacco. These factors would cause both Matin4e Slim and Player's Special Blend Light to be highly recognizable relative to the other products in the cohort regardless of the presentation. From Figure 1. which Illustrates the various evaluations of, Matinde Slim, it can be seen that there does not appear to be an effect of brand imagery on the subjective evaluation of this product. U4 _~-i Xl~b INI) N) BATCo document for Legal Services : Health Canada 22 October 1999 - 10 - However. the other products in this study do show deviations 0.15 from baseline stability and therefore the presence of brand imagery effects. Three of these products will be presented In this report: du Maurier Special Mild Accord Ultra Mild Vantage The effects of brand imagery on the subjective evaluations of these products are illustrated in Figures 2-4. Each Is an example of a different type of effect and these are Summarized as follows. du Maurier Special Mild Revealing the brand name (Unmasked) caused a shift in the perception of this product (deviation from stability = 0.26, Table 6). The results are illustrated in Figure 2. Relative to the Masked evaluation, revealing the brand name caused a diminishment in the perception of Mouth Effects, Impact and Throat Effects but no change in Taste Level and an increase in Acceptability. The line associated with the Unmasked presentation is in a lower position in Figure 2 relative to the Masked evaluation. From this, it was concluded that there was a shift in perception towards a milder product with an equal Taste Level and therefore greater Acceptability. Placing the product unmasked in its own pack (Own Pack) caused a more dramatic shift In the perception of this product (deviation from stability = 0.35, Table 6). All correlations with the product attributes decreased with the exception of Draw Effort (Figure 2). Once again, a shift towards a milder product is evident. Acceptability also increased relative to the Masked evaluation. It would appear, therefore, that for this particular product revealing the cigarette logo which includes the words "Special Mild" alterA the _c::h CD BATCo document for Legal Services : Health Canada 22 October 1999 panelists' perception towards a milder product. The pack appears to reinforce the perception of a product that is relatively milder than in the Masked evaluation. In this case, brand name and pack alter perception in the same direction. Accord Ultra Mild The subjective evaluations of Accord Ultra Mild are illustrated in Figure 3. Deviations from stability are given in Table 6. When compared to the Masked evaiuation, revealing the brand name to the panelists caused them to chance their perception of the product towards higher strength. Although perceived mechanics and Taste Level remained similar, the levels of Irritation, Impact, After Effects were perceived as being higher. Relative Acceptability decreased. Presenting the product unmasked in its own pack (Own Pack) resulted in a shift towards a milder product relative to both Unmasked and Masked evaluations (Figure 3). The levels of all the product attributes with the exception of Draw Effort were perceived as being lower. Relative acceptability of Own Pack was higher compared to Unmasked but lower compared to Masked. The pack design includes the words nUltra Mild' which may be influencing the panelists. The logo on the cigarette does not. In this example brand name and pack appear to affect subjective evaluation in opposite ways. Vantace The results of the subjective evaluation of vantage for the three presentalt, tons are given -in Figure 4. Relative to the Masked evaluation, reveallng the brand name (Unmasked) to the panelists resulted in a very different perception of the same product (deviation from stability a 1.00, Tabie 6). Evaluation of mechanics (Smoke Quantity and Draw Effort) NJ r11 J U_J BATCo document for Legal Services : Health Canada 22 October 1999 I - 12 - was revers-e-d and levels of perceived Taste, Irritation. Impact and After Effects.were markedly decreased (Figure 4). It Is evident that there is a large per---eptual shift towards a milder product. In this case, relative a::eptabillty Is lower. When Vantaze was presented unmasked and In its own pack (Own Pack), it was percei-ied to be only slightly dissimilar to the Masked evaluation (Figure 4). Referring to Table 6, it can be seen that the Own Pack evaluation Is sufficiently similar to the Masked evaluation (deviation from stabi-.ity = 0.15) that the existence of an imagery effect due to the Own Pack variable could be questionable. However, 1-' it is accepted that the pack does not alter the subjective evaluation of this product, and from Table 6 and Figure 4, that the brand name alone (Unmasked) does affect subjective evaluation, It would follow that the Nn Pack evaluation would be more similar to the Unmasked evaluation and less similar to the Masked evaluation. This is not the case. An ilternative explanation Is to conclude that there is an effect due to the pack imagery, that it Is in conflict with the brand name alone (Unmasked) but reinforces the product alone '(Masked) and as such is sufficient*y powerful to overcome the influence of the Unmasked effect. Conclusions from B.I. #1 The followtng conclusions were drawn from the first study of brand imagery ef--ects: 1) Brand imagery effects do exist for Canadian products and can be as--essed by the in-house ITL Delta panel. 2) Nc: all products are affected by the introduction of imagery variables. There are some products which are highly recognizable e.:. Matinde Slim and Player's Special Blend Light. However, this is most likely cohort specific. Placed in a cohort where BATCo document for Legal Services : Health Canada 22 October 1999 the recognition trigger, such as modified blend or circumference, is not obvious the influence of brand Imagery variables would most likely be evident. 3) Individual brand imagery variables, such as brand name or pack, may Influence subjective evaluation in the same direction to simultaneously enhance or diminish sensory perception (du Maurier S.M.). The variables may also work in opposite directions (Accord U.M.). A third scenario indicates that Individual variables may not be of equal influence. the effect of one may be masked by another (Vantage). Recommendations from B.I. #1 The products evaluated in B.I. #1 were current market samples, readily available to all smokers. It was speculated that familiarity with these products was 'he factor responsible for changes in perception and not the influence of brand imagery variables. Secondly, in this study, products were presented unmasked In their own packages. Therefore any effect of pack design was not truly isolated from brand name. It was recommended therefore that a second Brand Imagery study be designed to address these points. Brand Imaaerv Study 2 (B.I. #2. February 1986) As noted, the second Brand Imagery study was initiated to answer certain questions arising from B.I. #1. The first, a question of the validity of brand imageri effects relative to product familiarity was addressed by including foreign (U.K.) products and little known Canadian trademarks In the assessment cohorts (Table 3). The second objective was to Isolate pack imagery effects from brand effects. This was achieved by Including an assessment of empty packages (Empty) and by presenting panelists with BATCo document for Legal Services : Health Canada 22 October 1999 - 14 - products masked in their own packs (Own) and masked in different packs (Alternative). In the latter presentation masked products were placed in packages other than their own but of similar standard tar delivery (Table 7). The Masked and Empty pack evaluations were replicated. The reasons for replicating the former have been discussed, previously. The purpose for replicating the latter was to determine whether empty packs could be assessed consistently over time and therefore indicate smoker expectations reliably. The evaluations of the Empty pack presentations were separated by three months. The overall correlation between these two assessments was 0.85 indicating a very high degree of stability In empty pack evaluation. The levels of stability between the two Masked presentations and between the Masked and other presentations are given in Table S. The main points may be summarized as follows: 1) An acceptable level of stability was achieved for all masked products except Dumont Select Mild and Century King Size (Table 7, Ist column). The latter are not Included in this discussion. 2) There are marked changes In the evaluation of the products least some of the different presentations. The exceptions State Express Own pack and Alternative pack Sterling Own pack and Unmasked Belmont Mild all presentations except Empty pack For these products, the levels of stability between the two Masked evaluations and between the Masked evaluation and the particular pre5entation noted, are sufficiently similar that the existence of a brand imagery effect due to the imagery variable introduced Is doubtful (deviation from stability f- 0.15, Table 8). This has been briefly BATCo document for Legal Services : Health Canada 22 October 1999 - 15 - discussed In the previous sections concerning Matin6e Slim/Player's Special Bland Light and Vantage. However, results of this type must be interpreted in relation to the effects o~ the other imagery variables. It is possible that the, connotations associated with a particular imagery variable are in such complete agreement with the product sensory character, that they add no further dimensions to total product perception and therefore no effect of brand Imagery Is detected. However, in examining the data in Table 8 for State Express, Sterling and Belmont Mild, It Is more likely that the influence being tested is overcome by the product and, as a result, is ineffective in influencing subjective evaluation. As an example, Belmont Mild is a product which was perceived under all presentations except Empty Pack, to be virtually identical to its Masked evaluation. The deviation from stability for the Empty Pack evaluation (1.67, Table 8) indicates a marked difference in product perception. Thus, it would appear that although the pack for Belmont Mild is associated with a certain set of expectations/ perceptions, when masked Belmont Mild cigarettes were placed In the pack, the sensory character of the product alone dominated the subjective evaluation (devlaton from stability = 0.14, Table 8). This is supported by the evaluation of masked Belmont Mild cigarettes in an Alternative pack. Once again, the sensory character of the product dictates the overall subjective evaluation (deviation from stability a 0.06, Table 8). Similar arguments can be made for State Express and Sterling. There does appear to be some effect of brand name alone (Unmasked, deviation from stability - 0.55, Table 8) and pack (Empty Pack, deviation from stability = 0.20, Table 8) for State Express. However, the masked product still seems to overcome any other imagery effect (deviations from stability 0.02 and 0.03 for Own and Alternative packs, respectively). Sterling illustrates a slightly different example. The masked product Is still the major contributor to the subjective evaluation of the product In its own pack (deviations from stability = 1.44 and 0.05 for Empty and Own packs, respectively). However, In the alternative pack presentation (J-j r1o r\11 1-4 BATCo document for Legal Services : Health Canada 22 October 1999 - 16 - there does appear to be a greater influence due to the pack (deviation from stability . 0.29, Table 8). Since the correlation between the alternative pack and the original Masked assessment is greater than any other correlation, it might even be argued that the Alternative pack (Dumont Select) is more appropriate for this product than its own.pack. In this report, the following products will be discus these points further: Avanti Light Craven A Silk Cut Avanti Liqht The results of the subjective evaluation for Avanti Light under the different presentations are illustrated in Figure 5. Revealing the brand name (Unmasked) caused a shift In perception (deviation from stability 0.22. Table 8). As illustrated in Figure 5, the shift is towards a milder product compared to the Masked evaluation especially in terms of Irritation factors Impact and Taste Level (lower levels). The data in Table 8 Indicates that there is also a shift in perception of the pack relative to the Masked evaluation (Empty pack, deviation from stability = 0.24). The Empty pack evaluation illustrated in Figure 5 shows that smokers expected a considerably milder product than the Masked evaluation would imply. In fact, the Empty pack and Unmasked evaluations appear to be more similar to one another than either is to the Masked evaluation. The levels of Acceptability are similarly quite high for all three presentations. Presented to the panelists masked and in its own pack, Avanti Light was once again perceived to be a milder product than its Masked evaluation (Figure 5). The packed evaluation closely,follows that of the Empty pack. The influences of product alone (Ma~ked) are evident in the N) co BATCo document for Legal Services : Health Canada 22 October 1999 - 17 - perceived le'vel of Throat Effects (one of the irritation factors). However, it would appear that perception of the total product is most powerfully Influenced by pack perceptions. In the Alternative pack presentation there is a more dramatic shift In product perception (Alternative pack, deviation from stability - 0.91, Table 8). From Figure 5, It Is evident that there is a reversal In the overall perception of this product. Avanti Light presented In a Century pack was perceived as a very much stronger product relative to its Masked evaluation. The Empty pack evaluation for Century is also shown In Figure 5. The relative position of this evaluation can account for the apparent influence of this pack on Avantl Light. In this case, the total product evaluation Is being gr.eatly influenced by the imagery of the pack in which it is presented. Craven A The results for Craven A evaluated under different presentations are shown In Figure 6. From the figures for deviation from stability (Table 8), It is evident that there is a perceptual shift associated with the brand name (Unmasked). It appears, from Figure 6, that this shift is towards a somewhat milder product, although there was a slightly higher perception of Taste Level. Is The Empty pack evaluation is associated with the greatest deviation from stability (0.62, Table 8). As shown in Figure 6, panelists expected a considerably milder product compared to the Masked and Unmasked evaluations. Relative acceptability was highest for the Empty pack. It would follow from these points, that Craven A masked and in its own pack would be perceived as being milder than Its masked evaluation. However, from Table 8, the deviation from stability Is only 0.17. Furthermore, the line in Figure 6 associated with the Own pack evaluation follows the Masked evaluation more closely than it does the Empty pack _Pb- evaluation. It appears therefore, that the evaluation of Craven A masked (~D and in Its own pack is more influenced by the product than the pack. rQ r%J r1-j %-0 BATCo document for Legal Services : Health Canada 22 October 1999 - 18 - This was supported by the Alternative presentation, in which Craven A was placed in a Belmont pack. The Empty pack evaluation of Belmont Indicates 6hat smokers expected a very mild product. The deviation from stability lor Craven A in a Belmont pack confirms a perceptual shift due to the Belmont pack (0.38, Table 8). However, from Figure 6, It appears that the evaluation of Craven A In a Belmont pack is more similar to the evaluations of Craven A in its own pack, masked and unmasked than it is .o the Belmont pack. Thus, although Craven A Is influenced to an extent ty Imagery, the perception of the packed product Is more influenced by the sensory character of the product than by the connotations associated with the imagery variables. Silk Cut The subjective evaluations of Silk Cut are Illustrated in Figure 7. From iable 8, It is apparent that there is a considerable difference between tbe perceptions of Silk Cut masked and unmasked (deviation from stability 0-63). Compared to the Masked evaluation, the Unmasked evaluation indicates a considerably different product especially In terms of irritation factors, After tffects and Acceptability (Figure 7). The Unmasked product was perceived as being stronger than the Masked product. The Empty pack evaluation indicates that smokers expected a completely different product (deviation from stability a 1.37, Table 8), an extremely mild one, relative to both Masked and Unmasked evaluations (Figure 7). Presented masked and in its own pack, Silk Cut was perceived as still another product relatIve to the Masked evaluation (deviation from stability = 0.72, Table 8). According to Figure 7, Silk Cut In its own pack bears some resemblance to its evaluation unmasked but was perceived as stronger than its Unmasked and considerably stronger than its Masked evaluation. As a total product perception, Silk Cut appears to be more product or trand name oriented than pack Image oriented. This Is supported somewhat by the Alternative pack presentation in which Silk Cut was placed in a 1 State Express pack. The evaluation of Silk Cut in this pack was also perceived to be somewhat stronger than the Empty pack evaluation of Silk N Cut. N-1 BATCo document for Legal Services: Health Canada 22 October 1999 - 19 - (Figure 7 and deviation from stability - 1.17, Table 8). It does, however, bear some resemblance to the Unmasked evaluation of Silk Cut (Figure 7). Therefore, it Is possible that the total perception of this brand Is influenced more by the product characteristics when presented in Alternative packs or the brand name when presented in Its own pack than by the actual Imagery (colour, stripes, etc.) associated with the pack. Conclusions from B.I. #2 The following conclusions were drawn from the second Brand Imagery study: 0. 1) The majority of the products both Canadian and foreign were consistently evaluated in the Masked presentations. Product familiarity did not appear to be responsible for changes in product perception under different conditions. Changes in perception could, however, be attributed to the imagery variables introduced. 2) Empty pack evaluations could be assessed consistently over time and provided an indication of smoker expectation. 3) In some cases, the perception of the total product was more greatly influenced by the pack (Avantl Light). In other cases the product itself or perhaps the brand name was a more important factor (Craven A). Brand Imagery Study 3 (B.I. *3. January 1987) The third and most recent Brand Imagery study was initiated to examine a more realistic marketing scenario. The objective of B.I. #3 was to assess the changes occurring in the subjective evaluation of a series of products when smoked Masked, Unmasked, in their own packs and in each of the other packs (Table 1). The products are listed in Table 3 and the alternative pack presentations in Table 9. L'1~ BATCo document for Legal Services : Health Canada 22 October 1999 - 20 - During the course of analysis, it became apparent that the panelists were evaiuating this particular series of cigarettes as two distinct cohorts., a iow (3-4 mg tar delivery) and a high (8-9 mg tar delivery) tar cohort. It was decided therefore to analyse the two groups of cigarettes se.parately once a sufficient number of presentations had been evaluated. This is an important point to bear in mind when examining the results presented in this report. The association between the low" tar products and the Product Attributes appear to be extremely high and misrepresentative of historical values for these types of products. However, the results for the low tar products are relative numbers associated with cohorts made up entirely of low tar products. On no aczount should they be construed as absolute values. The product correlations between the various presentations and the Masked evaiuations are given in Table 10. The following points are evident from thts table: 1) All products were consistently evaluated In the Masked presentations (levels of stability 0.77-0.97, Table 10). 2) There are marked differences between the products evaluated Masked and the products' Empty pack evaluations and Unmasked evaluations (deviations from stability 0.24-1.69, Table 10). 3) In the high tar portion of the cohort, there Is really no variation in subjective evaluation across any of the Alternative pack presentations except for Matin6e and du Maurier Extra Light (devi.atlons from stability, Table 10). This may be a function of the narrow tar band (8-9 mg) and therefore little differentiation between products. 4) There is considerably more variation across presentations In the low tar portion of the cohort (deylations from stability, Table 10). This is an Interesting point since the width of this tar band (3-4 mg) is no different from that of the high tar portion. N) rQ BATCo document for Legal Services : Health Canada 22 October 1999 - 21 - T,I,e following products were of particular interest and so will be presented in this report: Mati nie - high tar portion Peter Jackson Extra Light - low tar portion Ma`nde The results of the subjective evaluations of Matinde are shown In Figure 8. Revealing the brand name of this product (Unmasked) caused a small perceptual shift (deviation from stability - 0.15, Table 10). In Figure 8. 116 can be seen that this shift is towards a somewhat milder product. ilhe Empty pack evaluation is associated with the product's greatest perceptual sh1f t (1 .03, Table 10) . It appears from Figure 8, that the ;a:-elists anticipated a product that would be considerably milder than 6..e Masked product. It would follow, therefore, that if the pack imagery of Matin6e was effective, it would greatly influence the subjective evaluation of the packed product. However, when placed masked and In its ck-i pack, Matinie was perceived to be virtually identical to Its Masked eyaaluation (deviation from stability z 0.05, Table 10). The subjective e-e-aluatlon of the packed product is dominated bi the product sensory c~aracter. Frzm the Alternative pack presentations (Table 10) it can be seen that Patin6e Is a very distinctive product. The only perceptual shift, albeit a small one, occurred when Matinde was placed masked in a du Maurier L'ght pack (deviation from stability = 0.15, Table 10). This Information Is Illustrated In greater detail in Figure 9. The relative position of Na-tinde in a du Maurier Light pack can be attributed to the influence of 1-e du Maurier Light packaging. Since the latter pack is influencing the perception of Matin6e while Matinde's own pack Is not, It could be argued 6..at du Maurier Light is a more powerful pack than is Matin6e. jtb- IN-1 BATCo document for Legal Services : Health Canada 22 October 1999 - 22 - ra-er Jackson Extra Light e subjective evaluations of Peter Jackson Extra Light are I ilustrated Figure 10. There are really no differences in the subjective evaluation of this product Masked, Unmasked, In Its Own pack or as an EmIly pack (deviation from stability 0.02 to 0.09, Table 10). The Information in Figure 10 shows that the Unmasked product was perhaps percelved to be slightly stronger. and that the Empty pack evaluation is r associated with higher levels of Impact, Throat Effects and Taste Level. K--wever, these differences are marginal. The Own pack evaluation shows even less difference compared to the Masked evaluation (Figure 10) and is associated with the highest relative acceptability. It would follow then t-nat Peter Jackson Extra Light is either a very distinctive product or ~aL its pack and brand name imagery are In close agreement wi6h its ;-~,duct sensory character. W-,en placed in Alternative packaging, it is evident that this product can 4 1 ity Influenced by different pack Imagery (deviations from stab, 17ternative 41, Craven A S.M., 0.18; Alternative #2, Viscount E.M., 0.53 -~:d Alternative #3, Matinde E.M 0.47; Table 10). It Is unlikely =,erefore, that Peter Jackson Extra Light is such a distinctive or re ognizable product. such as Matin6e, Matin&e Slim or Player's Special Fend Light. From Figure 10, it appears that Peter Jackson E.L. in a C7aven A S.M. pack was perceived as somewhat stronger, but not to a very Sreat extent. In both Matin6e E.M. and Viscount E.M. packages, however, ;~etar Jackson Extra Light was perceived to be considerably milder than .:s Masked evaluation. 77e relative characteristics of the different packagings are illustrated ~n Figure 11. From this figure, and Figure 10, it is obvious that the re-tallive perception of Peter Jackson Extra Light in either a Matinde E-tra Mild or Viscount Extra Mild pack must be a compromise between the extremely mild expectations suggested by these packs and the relatively s:.-ong sensory chara~ter of the Masked product. -9~- BATCo document for Legal Services : Health Canada 22 October 1999 - 23 - Thie :raven A Special Mild package was perceived to be somewhat weaker ttz~ the Peter Jackson Extra Light package but stronger than Matin6e E.M. an-- Viscount E.M. From Figure 10 and Table 10 the evaluation of Peter 3a:xson Extra Light In a Craven A Special Mild pack is similar '60 the Unnasked Peter Jackson Extra Light evaluation. WE particular product has been shown to be influenced to a certain ex:-znt by the pack in which It is presented. The degree of this lrr"uence is a function of difference between the expectations elicited by :ne packaging and the sensory character of the product itself. Since Pe:ar Jackson Extra Light can be influenced by packaging, It is perhaps nc,: as distinctive a product In sensory terms as Matinde was in the high tar ~ortion of this cohort. However, Peter Jackson Extra Light Is sufficiently strong In its sensory characteristics, that the perception of tnls product in very different packagings (Matinie E.M. and Viscount E.F.) is modified but not in complete disagreement with its Masked e,,E:uation. C-s-clusions from B.I. #3 1) There appears to be more variation In subjective evaluation atttributable to imagery variables in low delivery cigarettes relative .o higher delivery cigarettes. 2) 3rand name does have connotations which may shift product perception. However the more important influences appear to be the product itself and the pack In which it Is presented. 3) Subjective evaluation can be manipulated by imagery variables. However, the extent to which perception of the total product can be altered depends on the relative dominance in sensory terms of the individual product and packaging elements. I (-P4 N-1 (A U1 BATCo document for Legal Services : Health Canada 22 October 1999 - 24 - 2.3 General C:icTusions Based on :ns three Brand Imagery studies summarized in this report, the followin-- --aneral conclusions can be drawn: 1) Branc :,nintagery effects are measurable In the Canadian market. 2) These effects have directional characteristics and may vary in m a gn t -::., d e . 3) There 's not necessarily a consistent change In perceived attributes due :: brand imagery variables. Some attributes may be affected more or Iess by the introduction of these variables. Irritation factors, After Effects and Acceptability appear to be the most readily inflzenced. 4) There 4s some evidence that the perception of certain brands is damilnz:ed by product characteristics while for other brands imagery is tne most influential factor. 5) The re'ative dominance of products and packagings can be measured and their -nd1vidual contributions to the perception of the total product can tZ assessed. 3. RECOMMEND17IONS CONCERNING BRAND IMAGERY STUDIES The Del'-= :echnique as followed in these studies is relatively straight- forward. "However there are certain methodological points which should be considerec'. N) Ll_~ ON BATCo document for Legal Services : Health Canada 22 October 1999 - 25 - Cohort Des-an Design of :he product cohort to be assessed is of primary importance. Ideally, a cohort should be a continuum of products representing the maximum variety available for the products of interest. If possible there shoud be no anomalies i.e. highly recognizable products, unless these are -.o be specifically tested. Inclusion of groups of extremes should als-!~ be avoided. The latter point was Illustrated by the third brand imagery study in which the -roducts of interest included two distinct groups of product delivery (2--9 mg and 3-4 mg). Although the results for the two groups were rigor.-us in their individual domains and brand Imagery effects could be assess&:, situations of this type are best avoided. Panelists The seconc point to be considered is the age and history of the panel. Brand Imagery studies tend to be long and intensive assessment exercises. maintaining a high calibre of panel performance can become difficult iihen no immediate feedback is forthcoming. Furthermore, panelists uay form biases towards products when exposed repeatedly to the same product in different presentations. If panelists come to doubt their own expertise in subjective evaluation, their reliability as a panel may .4e placed In jeopardy. This is especially relevant if the panel Is r-.utinely used in-house. There may ze some merit In using external panelists providing they are sufficient.:y trained in the technique to be reliable. Relative to in-house panelists. an external panel would be "naive* with respect to product/to:acco Industry familiarity. Their subjective evaluations of products nay then be more susceptible to brand Imagery variables and more relevant t: consumers at large. However, in addition to the cost and time frame of such studies, questions concerning the potential Impact on public relations, consumer attitudes and the extended vafue of the irformatict should be considered. N.) L%, ___J BATCo document for Legal Services : Health Canada 22 October 1999 - 26 - The last point may require further clarification. It must be remembered that the Delta technique provides information for a given point in time- for a specific series of products. This information is relative to, and only relevait for, the products being compared. It cannot be extrapolated beyond Its specific study. These constraints are not unique and also usually apply to other product development, marketing research and mapping tools. Data Analvs-*s 0 Throughout :he studies described in this report, results from the various presentations have been compared to a baseline (Masked evaluation). Some consideration should be given to further comparisons among the different presentatiois such as Unmasked vs Own pack, Own pack vs Alternative pack etc. In thtos way the extent to which similarities or differences exist between various presentations could be determined relative to one another which may be more relevant than comparisons to a baseline. However such a refinemen: in data analysis would require either replications of the various presentations or acceptance of the assumption that the level of stability far replicates of any presentation will be at least equal to that of the baseline (Masked evaluation). If a larger panel was used (n .1 30) the significance of the differences between the evaluation of different presentations could be assessed by conventional statistical analyses. It would also be possible to determine the significance of differences between various points i.e. perception zf individual attributes for the different presentations. Therefore 'lie most readily Influenced attributes could be determined. In summary. --the effects of brand imagery variables on subjective evaluation can be measured systematically and quantitatively. Results from these Studies offer information concerning the forces to which smokers are exposed when branded products are purchased and smoked. Thi s may be of value to product development and marketing. N~ N) L~lj co BATCo document for Legal Services : Health Canada 22 October 1999 - 27 - From an R & D point of view, brand imagery studies provide another vehicle with which to study smoker-product interactions. In this context, a limited brand Imagery study is planned for the Human Smoking Behaviour work area. Beyond this, no further work on the effects of brand Imagery is envisioned in the near future. 0 0 C^1 110 BATCo document for Legal Services : Health Canada 22 October 1999 - 28 - REFERENCES 1. Boisjoly, C.; An Investigation of the Possible Influence of Imagery Variables an the Sensory Evaluation of Selected Canadian Product May, 1985. 2. Boisjoly, C-; Brand Imagery - Phase 2. February, 1986. 3. Boisjoly, C.; McBride, C.; Brand Imagery Study of Four Mid-Tar UL Products, January, 1987. 4. Ferris, R.P.; BAT Report No. R.D. 1752-C, 18.7.80. S. Ferris, R.P.; BAT Report No. R-D. 1667, 18.4.79. fi. Ferris, R.P.; BAT Report No. R.D. 1724-C, 25.2.80. 0 Xl~!. r*Q -Ilt6 CD BATCo document for Legal Services : Health Canada 22 October 1999 - 29- TABLE I SPECIFIC OBJECTIVES FOR THE ITL BRAND IMAGERY STUDIES B.I. SPECIFIC OBJECTIVES I To determine whether the ITL in-house Delta panel would be affected by the Introduction of imagery variables. To assess the direction and magnitude of these effects. 2 To assess whether the changes in product perception as evident in B.I. v2l were truly attributable to the effects of brand Imagery or caused by product familiarity. To determine whether the effects of brand name and pack imagery could be isolated from one another. 3 To assess the effects of alternative Imagery on the same products. 0_1 '_Nj BATCo document for Legal Services : Health Canada 22 October 1999 - 30 - TABLE 2 PRODUCT ATTRIBUTES USED BY THE ITL DELTA PANEL I - SMOKE QUANTITY: The perceived proportion of smoke to air In the mouth dur'ng a normal puff. 2 DRAW EFFORT The amount of physical effort required to get the smoke 0 out of the cigarette. 3 - MOUTH EFFECTS : Any physical feeling or sensation occurring In the mouth other than taste. 4 - IMPACT : Immadiate hit in the throat upon inhalation. 5 - THROAT EFFECTS: Any physical feeling or sensation (other than impact) occurring in the throat while smoking. 5 - TASTE LEVEL : A measure of the total intensity of flavour. .90 7 - AFTER EFFECTS : After smoking, a residual taste In the mouth or texture effect (dryness or coating of the oral-cavity). ACCEPTABILITY : The degree of acceptability based on the subjectivity evaluation. BATCo document for Legal Services : Health Canada 22 October 1999 - 31 - TABLE 3 .SUMMARY OF THE METHODOLOGY FOR ITL* BRAND IMAGERY STUDIES i I I Lengt1i and Declared Presentations Time # of Delta I Products Tar Delivery of replicates (wks) B.14 I Panelists I or sessions) Frame Matin6e Slim 3 Masked (2) du Maurier S.M. 11 Rothman's E.L. 8 4 1 9 Player's E.L. King Size 11 Unmasked Vantage 10 Accord U.M. 3 Player's Sp.31.L. 9 Unmasked in Peter Jackson E.L. 7 Own Pack I State Express* 9 1 Masked (2) Sterling" 7 Belmont 11 Empty Packs (2) 2 9 Avanti Light King Size a 14 Silk Cut* 9 Masked in Own Packi Craven A 12 (Respective Pack) Dumont 9 Century" 7 Masked in Alternative Packs Unmasked Viscount E.M. King Size 3 Masked (3) du Maurier Light I Regular 9 Peter Jackson EL** I King Size 4 Empty packs 3 1 9 1 Matin6e, I Regular 8 1 1 13 0 1 1 Craven A S.M. I King Size 4 IMasked in Own Pack I I I du Maurier E.L. I Regular 8 1 1 Matin6e E.M." King Size 4 Masked in Player's E.L. Regular 9 Alternative Packs (3) Unmasked I - Product name abbreviations: E.L./E.M. - Extra Light/Extra Mi Id S.M. = Special Mild Sp.Bl.L. Special Blend Light - U.K. products. - / Denotes identical products 'in that cohort. L,% BATCo document for Legal Services : Health Canada 22 October 1999 - 32 - TABLE 4 DATA ANALYSIS. A/ WITHIN PRESENTATION B/ BETWEEN PRESENTATIONS 1) DELTA By Product 1) BANNISTER By Attribute STOP Acceptable STOP NOT Acceptable Acceptablle 2) INGRID 2) Delta for Replicate I Comparison of Masked acceptable = X (a 0.6( Correlation Matrices Graphs & Maps Different Presentati 4- Comparison of Masked Presentations leve 41 Ix-yj ;; 0.15 3) INGRID/ADELA Correlation Matrices Graphs & Maps a. BATCo document for Legal Services : Health Canada 22 October 1999 - 33 - TABLE 5 SUMMARY OF PRESENTATIONS I I I I PRESENTATION IMAGERY VARIABLE PURPOSE TESTED 1. Masked Products None To establish a baseline against which subsequent tests are compared. Replicate: To assure stability of the baseline evaluation. 2. Empty Packages Pack Image To evaluate the products' sensory characteristics as suggested by the pack image i.e. smoker expectations. Replicate: To ascertain the reproducibility of the results obtained in the first test. 3. Masked Products Pack Image To evaluate the effects of pack In their Respec- Interaction image on product evaluation. tive Packages witq Product 4. Masked Products Pack Image To investigate the extent of pack in Alternative Interaction image influence by evaluating the Packages with Product effect of an alternative pack. 5. Unmasked Products Brand Awareness To evaluate the effect of brand name and/or product awareness on the sensory perception of a product. BATCo document for Legal Services : Health Canada 22 October 1999 - 34 - TABLE 6 CORRELATIONS BETWEEN THE FIRST MASKED EVALUATION AND THE REMAINING PRESENTATIONS FOR BRAND IMAGERY STUDY I PRESENTATION PRODUCT DEVIATION I I DEVIATION I MASKED 2* JUNMASKED I FROM I OWN PACK I FROM I STABILITY" I I STABILITY" Matinke Slim 1 0.99 1 0.99 1 0 1 0.97 1 0.02 I I i I I du Maurier S.M. 1 0.67 1 0.41 1 0.26 1 0.28 1 0.35 Rothman's E.L. 0.13 1 0.63 1 0.50 1 0.74 0.61 Player's E.L. 0.93 0.61 0.32 0.34 0.59 Vantage 0.85 -0.15 1.00 0.70 0.15 Accord U.M. 0.83 0.61 0.22 0.40 0.43 Player's Sp. Bl. L. 1 0.92 0.94 0.02 0.97 1 0.05 Peter Jackson E.L. 1 -0.02 0.53 0.55 1 0.27 1 0.29 Stability is achieved If these correlations 1 0.60. Absolute value of the difference between the first column and other appropriate columns of correlations. This value must ;; 0.15. CN BATCo document for Legal Services : Health Canada 22 October 1999 35 - TABLE 7 ALTERNATIVE PACK PRESENTATIONS FOR BRAND IMAGERY STUDY 2 ORIGINAL STANDARD ALTERNATIVE STANDARD PRODUCT TAR DELIVERY PACKAGING TAR DELIVERY I (mg) I I (mg) State Express 1 9 Silk Cut 9 Sterling 7 Dumont Select 9 Belmont Mild 11 Craven A 12 Avanti Light 8 Century 7 Silk Cut 9 State Express 9 Craven A 12 Belmont Mild 11 Dumont Select 9 Sterling 7 Century 7 Avanti Light a 0 -9:~ BATCo document for Legal Services : Health Canada 22 October 1999 >-5~ LM 10 M M c cn Z le --j I~ . 9 I~ 19 li 1:~ - - - - - - - - - - - - - - - - - - 61J I.2 r M C> Ln -CI 10 1 VI f- Co -e CD &n ~n 12 c~ c~ Q - - - - - - - - - - - - - - - - - - c%j ~2 CD :; M c7 0 ;; ~: = c~j L- c'~ c:~ I~ li Co a CD 0 CD - 0 0 CD VI) - - - - - - - - - - - - - - - - - L&J o ~- >d Ul$ CD cr 0~ 04 &~ Ln M VI < -Cr 1-1 60 M Co CY M UI~ C%i Cu -t = 4-- = Q- c~ c~ c~ c~ c~ c~ c~ CD 6w d p- CD < ~: C%j w1b WO cy r- r- CY cII~ 1-- 1:~ I~ M L65 ~ = CD CD CD CD CD CD 0 &- V> 6~J I: - - - - - - - - - - - - - - - - - CD CD ~e W -C 10 ~M ;~ 10 %0 ~D CD Cu I= M.. 61J C=; c~ CD CD r_ 1 ------------- 0 = 0 CA 1 m lI ii 61J = =,. - - - - - - - - - - - - - - Ci C.j r_ cr Li m C m cv cn tD 0 Q = CD CD CD CD CD CD W 6~ - - - - - - - - - - - - Co - &r) " Co C-') r- tu Cu ~ wj II~ II~ c~ II~ r~ I~ q: li e w~ ~ E V~ CD CD CD CD CD CD CD CD - - - - - - - - - - - - - - - - - - - - - M > CL) :: m J 0 #A U L- CD x CM -J 4, e VI CD 6.J r- 4~ m oi -W - -X > a 41 m 0 IE I= - m E r- -i~:- " eu > 1- = C3 VI V) = - Z; t- - (J4 BATCo document for Legal Services : Health Canada 22 October 1999 - 37 - TABLE 9 ALTERNATIVE PACK PRESENTATIONS FOR BRAND IMAGERY-STUDY 3 ALTERNATIVE PACK PRESENTATIONS ORIGINAL I PRODUCT #2 #3 Viscount E.M. Matin6e E.M. Peter Jackson E.L. Craven A S.M. du Maurier Light I Matinde I du Maurier E.L. I Player's E.L. I I I Peter Jackson E.L. Craven A S.M. Viscount E.M. Matinie E.M. Matinie I du Maurier Light I Player's E.L. I du Maurier E.L. I I I Craven A S.M. Peter Jackson E.L. Matin6e E.M. Viscount E.M. du Maurier E.L. I Player's E.L. I du Maurier Light I Matin6e I I I Mat4n6e E.M. Viscount E.M. Craven A S.M. Peter Jackson E.L. Player's E.L. du Maurier E.L. Matinde du Maurier Light -Ct~- --D IN) L-1 -Nj BATCo document for Legal Services : Health Canada 22 October 1999 0 0 CL - 38 - I jABLE 10 CORRELATIONS BETWEEN T11E FIRST MASKED PRESENTATION AND T11E REMAINING PRESENTATIONS FOR BRAND IMAGERY STUDY 3 -PRODUCT PRESENTATION EMPTY PACK OWN PACK A!TFRNATIVr P CKS UNMASKED HIGH TAR IMASKED 2* 1 1 #2 1 #3--1 z I -ID.S.**l 1 1 O.S. I- I D.S. S. 0 B.S. 0 du Haurier L. 1 0.09 1 0.65 1 0.24 1 0.90 1 0.01 1 0.85 1 0.04 1 0.831 0.06 1 0.721 0.17 1 0.75 1 0.14 0 1 1 1 1 1 1 1 1 1 1 1 1 1 Matin6e 1 0.80 -0.23 1 1.03 1 0.65 1 0.05 1 0.65 1 0.15 1 0.001 0.0 1 0.751 0.05 1 0.65 1 0.15 x I I I I I I I I I I I I Q Player's E.L. 0.94 0.68 0.26 1 0.93 0.01 0.95 1 0.01 .1 0.891 0.05 1 0.941 0.0 1 O.7(j 1 0.16 du Maurier E.L. 1 0.87 1 -0-89 1 1.76 0.63 1 0.24 1 0.69 1 0.18 1 0.701 0.17 10.841 0.03 1 0.72 1 0.15 . .. . .... ..... 2) 1 1 1 1 1 1 1 1 1 1 1 1 1 LOW TAR I I I I I I I 11) Matin6e E.M. 1 0.07 1 -0.6? 1 1.69 1 0.45 1 0.42 1 0.27 1 P.60 1 0.701 0.17 10.631 0.24 1 0.48 1 0.39 Viscount E.M. 0.97 0.62 0.35 0.63 0.14 1 0.70 1 0.27 1 -0.301 1.27 1 -0.831 1.30 1 0.56 0.41 0 1 1 1 1 1 1 1 0 Cravon A S.M. 0.77 0.41 0.36 0.50 0.19 1 0.69 1 0.08 1 0.361 0.41 1 -0.431 1.20 1 0.75 1 0.02 V* Cr Peter Jackson E.L. 1 0.78 1 0.76 1 0.02 1 0.00 1 0.02 1 0.60 1 0.18 1 0.251 0.53 1 0.311 0.47 1 0.69 1 0.09 (D Stability is achieved if these correlations k 0.60. D.S. - Deviation from Stability - Absolute difference between the first column and other appropriate to of correlations. This value must 9 0.15. co OqF,~/-U,ov 0 0 CL Figure I SUBJECTIVE EVALUATION OF MATINEE SLIM r+ MASKED 0.9 - 0.13 - fu 0.7 - UNMASKED 0.6 - 0.4 - OWN PACK 0.3 - 0.2 0.1 1: -0 X -0.2 - 0 -0.3 - 0) -0.4 - -0.5 - CL -0.6 - 0 -0.7 - -0 's - -0.9 0 0 QU o1v cr VAN* woo ofell-11 vvjo~- ofec"0-01'" (D co co 0 0 CL 0 Figure 2 (D SUBJECTIVE EVALUATION OF DU MAURIER SPECIAL MILD r+ a, MASKED 0.9 (a 13) 0.7 - UN14ASKED CA 0.6 - 0 0.5 - Z 0.4.- OWN PACK 0 0.3 - CD (A 0.2 - \ x -0 (D i 0) -0.1 - -0.2 - -0.3 - 0) -0.4 - -0.5 - 0) CL -0.6 - -0.7 0 0 for, (,Y ,f IAp N, off cr m of"' ol lks, W-(IFA NCGV to CD v 1 0 v m 0 CL Figure 3 C) SUBJECTIVE EVALUATION OF ACCORD ULTRA MILD CD 0.7 - UNMASKED 0.8 0.5 CD 0.4 - 0 WN PACK 2 0.3 - i, 0.2 - CD CD su -0.3 -0.4 % -015 - CL -0.7 - f* ~jjLl'i CAT Ffrls -Lw l a vf FE cls 1 &X eFv e f E-G, 6 .41E%- - 1 wtt-il %kouit, ego,, 11,51e wilEn ep Cr Ntc CD 1 - CD Ci) CD 0 0 CL 0 Figure 4 SUBJECTIVE EVALUATION OF VANTAGE (D HAUKLP 0.9 Q cc 0.7 UNMASKED lb 0.6 - 0.5 - 0.4 - OWN PACK 0.3 (D 0.2 - 0 -0 - (D la) X -0.2 - =r -0.3 - 0 0 0) -0.4 - -0.5 - -0.6 - -o.7 -0.8 0 -0.9 0 YPNIC3ev-fecjs XlE cr fo",olvill, (D VS~PZH,o~ 0 0 CL Figure 5 0 SUBJECTIVE EVALUATION OF AVANTI LIGHT (D MAUKU) 0.9 (D 0.7 - ~ UNMASKED 0.6 ca 0.5 1 11 1 0.4 - 0.3 - OWN PACK 0.2 - CHIT1111Y VAIX -0 x -0.1 - ------ *-, -0.2 - EMPTY PACK =r -0.3 - 0 -0.4 - -0.5 - EMPTY CENTURY PACK -0.6 r.L 0) -0.7 K) -0.8 - N) -0.9 - 0 0 PIOL SOO Dom l4oul" 06 ofv7s 0, too-- 0 0 CL 0 Figure 6 SUBJECTIVE EVALUATION OF CRAVEN 'A' M i r+ MASKED 0.0 0.8 0.7 0.6 UNMASKED CA 0.5 0.4 - OWN PACK 0.3 C) (D 0.2 (0) 0.1 DFI-14ONTPACK X -0 (D -0.1 -0.2 - EMPTY PACK =r -0.3 - 0 -0.4 - -0.5 - EMPTY BELMONT PACK -0.6 - I CL --0.7 0 --0.9 F PIL ECIS 0 sumn%" omc" "'m t%A cr Dow ~00 TO VA (D co co > 0 0 CL Figupe 7 SUBJECTIVE EVALUATION OF SILK CUT 0.9 MASKED r 0 0.9 cc 0) 0.7 CA 0.5 UNMASKED (D 0.5 0.4 0.3 CD CA 0.2 - OWN PACK 0.1 - x -0 (D sm -0.1 X -0.2 STATE EXPFIFSS PACK 0 -0.3 --0.4 fu -0.5 EMPTY PACK Q -0.6 - -0.7 - -0.8 - EMPTY STATE EXPRESS 0 -0.9 - PACK cl I I I I I -A-.-- I I I 0-01 cr 6,~o 0)wtT!;pw14 f ff Co. vocts lwwt I footstmo 16 Of is ovalo (D ~00 :400 ~cw co co ZS?,vz~ov CL 0 Figure 8 0 c SUBJECTIVE EVALUATION OF MATINEE 0 1 MASKED 0.0 0.0 0.7 OWN PACK 0) 0.6 CD 0.6 0.4 - DUML PACK 0 0.3 - PLEL PAGK (D 0) ;z -0.1 - -0.2 - DUN EL PACK 0 -0.3 - ~ 0 m -0.4 - 0 ASKED CL -0.5 a) -0.8 EHPTY PACK 0 -0.8 9 -.0.9 0 Cr CD oweUM,efoffi effeslw ?,CI~ ., eftz", oe toor evo> Iwo 00 Co CD Co RS~pz Hop Ua 0 0 CL Figure 9 COMPARISON OF MATINEE AND DU MAURIER LIGHT IMAGERY Ep 0.0 - I HAMM) 0.7 to 0.6 OWN PACK (1) 0.5 (D 0.4 - 0.3 - 0.2 - DUML PACK (D (0) 0.1 / I . -0 x -0.1 (D EMPTYHAT1111--E PACK -0.2 - -0.3 - -0.4 - EMPTY OU MAURIER LIGHT PACK 0) -0.6 - -0.7 - -0.8 - N N -0.9 - 0 --- L A 0 WTIV*i ro~ effoRN voysQU, jvkvol -tv, wl ~ccvpl 0 ORO ~Vjw B'se ,lot fyvec3s (D (o (o co 6~-7,p/~oV 00 0 0 CL Figure 10 0 SUBJECTIVE EVALUATION OF PETER JACKSON E.L. (D -h 0 MArsKFn (D (a 0.7 0) 0.6 0.5 - OWN PACK 0.4 - 0.3 - (D 0.2 - CRAVEN A SH PACK 0) 0.1 - 3: -0 (D 10 -0.1 X -0.2 - VISCOUNT EM PACK -0.3 - 0 0) -0.4 - -0.5 - MATINEE EXTRA MILD 0) PACK CL -0.6 Di -0.7 N -0.8 - 0 -0.9 - UNMASKED 0 cr VIA oolul" Iwo IT 010% f1roc swo ~slv too- wstls, (D _plpa.0-11 EMPTY PACK co 0 9F, vz ~~ov 00 .0 0 0 0 CL Figure li 0 COMPARISON OF LOW TAR EMPTY PACK EVALUATIONSr (D 211 VISCOUNT EXTRA MILD r- co CRAVEN A SH cu 0.7 0.6 - 0.5 - 0.4 - PETER JACKSON EL 0 0.3 - 0.2 0.1 MATINEE EXTnA MILD 3: -0 CD =r -0.2 - 0 -0,3 - 11) -0.4 - 0 13) -0.5 - -0.7 -0.8 0 -0.9 0 cr off-r-All ivp?A" off-r-T6 ELOEL . (~ M ofsa