Tobacco Documents Online

Authors: -1991 " Datet-"June 18- , Department: Biochemical/Biobehavioral/Brands/Technical Support/ Notebook New Products Technology Pages: Previous Reports: RDM, No. 2, 1981; RDM, No 48, 1980 Project No./Name: Nicotine/Product Understanding R&DM 1991, No. of Pages: Human Smoking BehaviorJFour-City Studv OBJECTIVES The main objective of this study was to determine how full flavor nonmenthol (FFNM) smokers' smoking behavior changed when they smoked cigarettes which varied in assembled cigarette draft and blend nicotine. The secondary objective was to confirm results from a 1986 Desire for Cigarette (DFC) study which indicated that full-flavor smokers' can be divided into clusters based on their description of an ideal cigarette and that these individual clusters of smokers had distinct smoking behaviors. If indeed, clusters similar to those in the previous study were found to exist, we wanted to identify demographic, psychological and sociological descriptors ~ for•-the-cluster-s. -- ~ - SUMMARY One-thousand, three hundred and forty-six full flavor, non-menthol smokers from four different U.S. cities participated in the "Four- City Study." The study involved testing these smokers' perspectives of an ideal cigarette, perceptions of various test products and how they smoked various cigarettes. Additionally, smokers' psychological traits were characterized and their usual brands were identified. A hexagonal RSM design with blend nicotine and assembled cigarette draft as the independent variables was used. The center of the design reflected Winston 85 specifications. Marlboro and Winston Lights were controls. Results indicated that full flavor, non-menthol smokers could be divided into four clusters based on their descriptions of an "ideal cigarette." Cluster 1 wanted smoothness along with little tobacco taste and low impact; cluster 2 wanted a lot of tobacco taste and low imgact;, cluste.r. 3 wanted alm.o_st. no.,t9baic,e,o,., t.aste and, n.o, impact; and, cluster 4 wanted lots of tobacco taste and moderate impact. Different clusters exhibited different smoking behaviors and reacted differently to product changes. The particular smoking behaviors of cluster members related to their descriptions of an ideal cigarette.

In addition to the clusters having distinct product wants and smoking behaviors, they had distinct psychological/sociological profiles. Data gained from this study met our goal to gain understanding of consumer wants, their smoking behaviors and their psychological/sociological profiles. Division of full flavor smokers into clusters should enable us to meet the needs of individual customers instead of trying to be all things to all people. 2

STATUS This project is complete. Results have been reported to R&D, Market~ng and the Business Information and Analysis Department's (BIAD) management. Human Smoking Behavior methodology, as confirmed by this study, is being utilized by a large-scale study referred to as "Nicotine RSM." The Nicotine RSM study involves applying HSB technology to understand the effects of FTC "tar," FTC nicotine and assembled cigarette draft on consumer acceptance, satisfaction, perception and smoking behavior. Product-want clusters are also being defined as part of this study. In addition, a study is being conducted to determine the feasibility of linking marketing segments- as defined by the "Strategic Marketing Analyses Resource Tool," (S.M.A.R.T.) and product-want clusters as defined by the Nicotine RSM study to aid marketing products based on product-want clusters. KEYWORDS clusters, demographic, draft, full flavor, human smoking behavior, HSB, lifestyles, Marlboro, nicotine (54-11-5), perceptions, personality, product-want clusters, psychographic, RSM, tar, Winston, Winston Lights. The Business Information and Analysis Department was previously referred to as the Marketing Research Division. 3

I. Introduction The Four-City Study combines technologies developed by R.J. Reynolds Tobacco Company's Research and Development (R&D) and the Business Information and Analysis Departments (BIAD). R&D developed the technology to measure how a person smokes a cigarettel. BIAD developed the technology to characterize smokers into product-want clusters;' i: e: ; grotzps-based• on- how, 'the - smokers,° characterize their "ideal cigarette." In 1986, a study, commonly referred to as the "1986 DFC* Study," which conducted. This study, which involved sixty consumers (evenly divided among male and female smokers, and among Winston and Marlboro smokers) was the first study to merge these technologies.** Results suggested, but due to the small sample size did not convincingly demonstrate, that smokers could be divided into clusters, based on their usual brand and their perceptions of an "ideal cigarette." Highlights from that study are presented in Table 1. Cluster 1's considered an ideal cigarette as one with a lot of tobacco taste. Cluster 2 wanted a lot of tobacco taste and a lot of strength. Cluster 3 wanted medium amounts of tobacco taste and strength. Smokers' smoking behaviors differed among the clusters and related to the clusters' ideals. Because of the importance of these findings, it was decided to verify the results by conducting a large scale study. This large- scale study, described here, is commonly referred to as the "Four- City Study." There~were three major objectives to the "Four-City Study." The first objective was to confirm the 1986 Small-Scale Study's results, i.e., that smokers could be divided into groups based on their description of an ideal cigarette and that the individual clusters' smoking behavior related to their ideals. Second, we wanted to determine if there are any psychological/sociological or demographic descriptors of product-want clusters to aid marketing to these clusters. Third, we wanted to determine if product-want clusters reacted similarly to product changes, in particular to changes in blend nicotine and to changes in assembled cigarette draft. To accomplish these objectives, characterizations of the full-flavor smoker's "ideal cigarette" were used to divide the * ** 1'DFC" is an acronym for Desire for Cigarette. The 1986-DFC study had the objective to learn about differences in Winston and Marlboro smokers. A addition to this study (mailout of ballots and Winston cigarettes to subjects after the human smoking behavior/desire for a cigarette phase was completed) allow this study's participants to„be. d~yided..ip~_Q,_clusters._, „ 4

subjects into clusters; changes in puffing behavior and perceptions with changes in "tar" and draft were measured. Cluster ideals were related to their responses to the psychological/sociological profile questionnaire items and to their smoking behaviors. The unique aspects of the Four-City Study are the large number of subjects considered and the depth of information gathered regarding subjects' attitudes, lifestyles, opinions and psychological make- up. This powerful data base was used to determine as completely as possible, the desires, interests and smoking behaviors of full flavor, non-menthol cigarette smokers. II. Experimental A. G_aneral _ Testing for the Four-City study took place in Buffalo, NY, Savannah, GA, St. Clairsville, OH, and Tucson, AZ. These four cities, whose locations are depicted in Figure 1, were selected because they were climatically diverse and because full flavor smokers could be recruited easily in these locations. Bellomy Research, Inc. ("Bellomy", Winston-Salem, NC) was responsible for proposing test cities. The goal was to recruit 300 persons from each city (it was anticipated that 90% of the recruits would complete the study.) Local marketing research firms in each city conducted the tests, under Bellomy's and BIAD's (Brand Information Department of RJRT) supervision. Names and addresses of the test centers are provided in Table 2. The four-city test consisted of two parts: first, an on-site test, which involved monitoring full-flavor smokers' smoking behavior and asking these same smokers to respond to questionnaires regarding their lifestyles, their personalities and their opinions. Second, a take-home part in which subjects indicted on NFO ballots (example given in Attachment 1) their acceptance and perceptions of test cigarettes`and their descriptioris of an ic~eaYcigarette. . B. Test Design A hexagonal RSM design, which consisted of six star points and a duplicated center point, was selected for this study. This design pattern was selected because it was the smallest practical pattern that would allow us to evaluate the effects of linear, squared and crossed terms in the various responses. The design was oriented so there were five different levels of draft, and three different levels of blend nicotine. Target draft levels ranged from 115 to 160 mm H2O; target nicotine levels ranged from 1.6 to 2.0 wt% (DWB). The design was oriented in this manner to minimize the number of blend nicotine levels required. Minimum number of blend 5

nicotine levels was desired because it was difficult to prepare blends that were similar with only nicotine level varXing. Center point~'specifications were J'based on Winston KS (with G7-2) specification. Winston Lights and Marlboro KS cigarettes were controls. The actual design differed from the target (Figure 2). Assembled cigarette draft levels were only slightly off from target; the mid- and high-nicotine levels were near target, but the low nicotine levels were definitely high, i.e., actual levels were 1.74 wt% -- target levels were 1.6 wt%. Correlations amongst the actual design levels were tested to determine if the linear, quadratic and interaction model terms were independent for the actual design. As shown in the correlation matrix (Table 3), except for a significant correlation between the linear and quadratic nicotine terms, all terms of interest remained independent.* The significant nicotine/nicotine-squared correlation was due to the two low nicotine level products being above target (a plot of nicotine versus nicotine-squared is provided in Figure 3). Although we did not meet our goal for complete independence, we continued the experiment using these products. This decision was made based on the fact that blends with nicotine at the 1.6 wt% target were not easily obtainable. But, --more importantly, 'we f elt - -that- -the la-ck- of- independence of - these two model terms would not affect ourr ability to address the study's objectives, i.e., the objectives were not to optimize the independent variables but to determine the relationship between the independent variables and human smoking behavior and perceptions. C. Products Ten products were utilized in this test. Eight were based on the design; two products (Marlboro KS and Winston Lights) were controls. Seven of the test ~roducts were from pilot plant run CT01568 (parts A, C through H) ; the other test product was from pilot test run CT01605A. Specifications for product CT01568, parts *Independence was calculated by running a correlation matrix on all points (in statistical design units) which comprised the design. Design units were determined by first calculating where the center of the design was located. The design center was determined by averaging over the levels of the variables for all ~ the design.points. --A design--unit• was--computed•-by••determining the-..- average distance from the center for the maximum and minimum levels for the independent variables. Lr **Pilot plant run CT01568, part B was remade in pilot test run a G CT01605A. 1-1 a

A, C through H* and product CT01605A were based on speci~f* cations for the Winston KS product which utilized G7-2. The cigarettes' draft and blend nicotine were varied within the various parts which made up the run. Assembled cigarette draft was varied by changing the filter. Blend nicotine was varied by selecting Burley and flue-cured grades which had different nicotine contents. Marlboro KS cigarettes were banded and packed into white packs (Pilot plant run CT01568H). Winston Lights test products was prepared (CT015681) based on specifications for the "new Winston Lights," the product implemented in April 1988. Blend nicotine contents, drafts and FTC tar and nicotine deliveries from the various cigarettes are presented in Table 4. Test codes used at the test sites to refer to the products are also given in that table. Detailed product specification are provided in Attachment 2. D. Test Schedule There were six test sessions per day, f ive days a week resulting in thirty sessions per week. Ten smokers attended each session. All smokers were required to attend smoking sessions once a week for the five week-test period. Subjects were supposed to attend sessions at the same time each week. If a subject missed a session, he/she was allowed to make-up the session as long as it was not within 5 days of the next test. For example, a Monday subject had to make-up the missed session by Wednesday. Thursday and Friday subjects could make-up sessions, if necessary, on Saturday. Subjects smoked two products per session and were exposed to all products._ _ To minimize , the ,.possi,b.i,7,ity,..,of_.e.rrors ,..and_._to, ease, subject handling, all subjects attending sessions on a given day were given the same product. Product presentation orders varied did vary from day-to-day and from site-to-site. These orders are presented in Table 5. The test took place in the summer of 1988. Testing at the Tucson site took place form June 6 until July 15.*** Savannah, St. Clairsville and Buffalo tests were staggered one, two and three weeks, respectively, after Tuscan testing. * ** *** Part CT01568B was remade in run CT01605A. This particular specification was used due to the difficulty in ammoniating blends for the newer Winston KS specifications. Testing was not scheduled for the week of July 4th. 7

E. Subjects Thirteen hundred and eighty-six full flavor smokers were recruited for the study, from the population at large in and around the test- site cities. Subjects were recruited by the local recruiting firms using the screening questionnaire, developed by Bellomy Research Inc. and BID (Attachment 3.) The primary criteria were that the subjects were 18 to 59 years old, that they smoked a brand of cigarette which was 85-mm, non-menthol and full flavor, and that they smoked ten or more cigarettes per day. Age, sex and usual brand distribution targets were established by BIAD and were calculated to reflect national incidence. Recruiting targets are presented in Attachment 4. The demographic distribution of subjects, for whom complete data records were available, closely matched targets. Seventy percent (versus a target of 75%) of the subjects were male. Eighteen percent (versus a target of 20%) were between the ages of 18 and ,(versus, a target, of,,30$), between, the_~ges_ of 25,,an..d,.34, 32$,_, (versus a target of 30%) were between the ages of 35 and 49 and 18% were above 50 (versus a target of 20%.) Upon completing the study, subjects were compensated ($125). F. Procedure 1. Pre-experimental Handling of Subjects Subjects were instructed not to smoke within a half-hour of their scheduled test time. When subjects arrived at the various test centers, they checked-in and then waited in a waiting area. Subjects were allowed to read magazines but were not allowed to smoke during the waiting period. 2. Human Smoking Behavior Test The human smoking behavior test procedure, summarized in Figure 4, involved testing subjects' puffing patterns while they smoked two different cigarettes. Subjects completed various questionnaires between the times they smoked the first and second cigarette. The procedure -begari 'with sdbjects chec7cirig-i:ri"arid"thbri' waiting to"' ' be escorted into the testing room. When it was time for the session to start, subjects were seated in the testing room. Next, the test administrator read the instructions (Attachment 5) to the subjects. Detailed instructions were given during the first test week. These instructions allowed the subjects to familiarize themselves with the equipment. Briefer instructions were read to the subjects during sessions held during weeks two through five. Both sets of instructions informed the subjects of the test protocol and about the test equipment. After instructions, subjects ate crackers and drank water to cleanse their palates. 8

Step-by-step instructions, to be followed during the test along with the questionnaires to be completed, were provided to the subjects. The instruction books differed from week to week due to the different questionnaires being used. Instructions books for all five weeks are attached (Attachment__6). ~The questionnaires, presented to the subjects on' a week-by-w~eek basis are listed in Table 7. After receiving instructions, and filling out their names, the date and the time on the instruction-book cover, the subjects were given the "go-ahead" to smoke the first cigarette. Cartoons were played for the subjects' entertainment during this smoking period. The panelists smoked the cigarettes through flow sensors in the form of a small plastic mouth pieces which were attached to the cigarettes (Figure 5). Smokers could easily hold and manipulate their cigarettes, within a radius of two or three feet (Figure 6). Details of the test equipment are presented in Section 6. After all subjects, participating in a given session, completed smoking the first product, a twenty-five minute waiting period started. During the waiting period, subjects ate crackers and drank water to prepare for the next cigarette and responded to various questionnaires. After the waiting period was over, subjects smoked another cigarette and watched cartoons. When testing was completed, subjects received a package containing cigarettes and questionnaires for the take-home portion of the test:_ . _ ,. .. _-_ .:..____ . .... . , ,... .._ .. ii. Take-Home Test After on-site testing, subjects were given a "take-home packet" consisting of two packs of cigarettes and two NFO ballots. ProductT in the take-home pack were the same products smoked in the session . Subjects were free to smoke the products in either order, but were instructed to complete one pack and smoke as many cigarettes as they could from the first pack and then complete the first questionnaire before starting the next pack. A second questionnaire was filled out after subjects smoked as many products as they could from the second pack. Subjects filled out both questionnaires at-home and then returned them in the envelope provided. Subjects were told that they had to return the questionnaires at their next scheduled test session in order for testing to continue. * The same products were in the take-home packs as were - smoked during a-given •.day • ta--eliminate -the•••need-.•for the various test administrators to handle more than two products per day. 9

6. Test Sites and Training of Various Marketing Research Firm Personnel a. Facility The test facility accommodated up to ten smokers per session. In the test room there were two 13-foot tables, constructed by bolting together a five and an eight foot table. Five subjects could sit at each. Tables were positions so that subjects at one of the 13- f oot tables had their backs to the subj ects at the other table. Six partitions were placed on each table to prevent subjects from disturbing each other. A TV was placed in front of each of the two tables. Both TV's were connected to a single VCR to allow simultaneous display of • the videcr entertainment-.- - - An air cleaner was installed in each test room. b. Training of Test Site Personnel A week-long training program was held to train the Senior Project Administrator and two additional persons from each of the various test sites. A schedule for the training program is provided in Attachment 7. Dr. John Reynolds and Mr. Alan Norman were the instructors. The training session allowed test site operators to familiarize themselves with the equipment and procedures. After the various marketing research firm personnel were trained, they, participated in the production of an instructional training video tape. The purpose of the training tape was to have a medium to train additional people at the test sites in case of employee turnover at the various marketing test sites. H. Human Smoking Behavior Testing Equipment 1. General .The, pxincipal_ feature of, the...eguipment- .use.d...,in._ this ,.tes_t versus. . that used in previous human smoking behavior testsl, such as the 1986 DFC study, is the revamping that took place to allow people less familiar with the instrumentation to operate the equipment in a reliable manner. The human smoking behavior (hsb) monitoring equipment, used in this study was designed by RJR personnel, principally Dr. David Griffith, and manufactured by Ferguson Manufacturing Company (Winston-Salem, NC). The method uses a device that measures and records the flow of air (measured by differential pressure) and draft (as indicated by ?????) at the mouth end of the cigarette as a person smokes. As shown in Figure 7, in addition to the mouthend piece, "probe", the equipment consisted a portable puff profilers, located at each station. Each puff profiler contains transducers, an analog to digital converter, an IBM compatible computer and a disk drive. These puff-profiler 10