Tobacco Documents Online

® a 4 ACCESSI0N N0, COPY N0~ I SSUED TO~~:~_ CHARGE NQ. & TITLE: TYPE REPORT: Q ANNUAL OSEMIANNUAL 0 COMPLETION O SPECIAL DATE: June 5, 1981 PERIaDl COVERED: June 1980 - June 1981 WRITTEN BY , REPORT TITLE:' qa F. J Ryan P H I L I P K 0 R R I S U!, S, A, <RESEARCH' CENTER ~. __ . Behavioral Research - 1600 Behavioral Research Laboratory Annual Report Part I J~ ~ . W. L. Dunn, 81 127 » ~ S. d. Dunn , s SUPERVISED BY APPROVED BY DISTRIBUTION: H.. Wakeham B. Seligman F. Resnik W. Farone W. Gannon L. Meyer T. Osdene R. Thomson P. Eichorn D. Lowitz M. Bourlas K. Burns H. Daniel F. Daylor P. Gauvin A. Kassman B. Kosakowski' W. Kuhn W. Mutter J. Osmalov E. F. J. J. W. R. R. R. W. D. Sanders Watson Whidby Wickham Dunn, Fagan Ikeda Carpenter Claflin Clark Jenkins H'. Spielberg Lilly V. DeNoble Merri t S. Dunn. Palmer F. Gullotta Laszlo J. Jones Goodale F. Ryan, Kallianos Manzelli' Wi'l1 Hausermann KEYWORDS: Social Psychological, Psychological Factors, Ultra Low Delivery Smokers, Firmness, Acceptabi~lity

F F I- - TABLE OF CONTENTS - A. SOCIAL PSYCHOLOGICAL PROGRAM PROJECT ON PSYCHOSOCIAL FACTORS INFLUENCING SMOKING BEHAVIOR F I I I- I I I- F F L L ~ L PAGE Abstract ........................................................... 1 Introduction ........................................................ 3 Method............................................................ . 4 Results and Di'scussion.......... ....................... 6 Future Directions of the Social Psychology Program ................. 19 Appendix A.......... ................................................ 20 References ........................................................ 31 B. GENERAL EXPERIMENTAL PROGRAM CIGARETTE FIRMNESS AND ACCEPTABILITY Abstract.......................................................... 1 Introduction........................................................ 2 Firmness Panel .................................................... 3 Subjective Firmness vs. Importance of Firmness. FWS Machine Scores ........................

F F F F I- F I I I F F I- F I ANNUAL REPORT - PROJECT 1600 Part T - A SOCIAL PSYCHOLOGICAL PROGRAM

F F F I F I I I I I F I I I ~ I- 1 L PROJECT ON PSYCHOSOCIAL FACTORS INFLUENCING SMOKING BEHAVIOR ABSTRACT . Dunn). The exploratory study of social psychological f actors that influence smoking has provided considerable data for the effort to understand the nonpharmacologi- cal f actors influencing cigarette smoking. The study employed 112 interviews that consisted of questions covering the six areas of social psychological correl!ates of cigarette smoking and several measures of smoking behavior. The responses within each of the six social psy- chollogical areas and~ the smoking behavior measures were factor analyzed'. The i nformati on obtai'neI in the factor analysi s shows that there are several dimen- sions of smoking behavior that should'! be studied. Four of the dimensions or measures of smoking behavior were selected for use as variables to be explained in multiple linear regression analyses. The factors derived from the six psycho- sociali areas were used as independent variables to explain the vari'ation in each of the smoking behavior measures. - The smoking behavior measures selected for analysis are profitable areas for further research. One smoking behavior measure appears to be related to a smoker's quitting history. Another is related to a smoker's concern about the delivery level of the brand being smoked. The third measure relates to the num- ber of cigarettes consumed. And the fourthi measure rel!ates to the number of different brands a smoker has tried. The amount of variation in each of the smoking behavior measures that can be explained by the psychosocialT factors is significant at the .0001 level. Nbw. 10003921'74

I- F I I I I i I I I I I I F i F L that we have found relationships between smoking behavior and psychosocial vari- ables we must determine what these relationships mean. Once the findings have been thoroughly analyzed', we plan to use the insight and direction gained from the results to pursue in a focused fashion the indicated rel'ationships between the psychosocial' variables and smoking!behavior. Information obtainedifromithese investigations will ultimately contribute to an understand~ing of such, smoking parameters as smoking rate, smoking style and brand choices. Not only will such an understanding be of scientific interest, but it may well have implications for marketing research and product development within our Company. L

F I I I I- I I I I t I- I- I I- 'I- L INTRODUCTION Last year we reportedi the intent to develop a research program concerned with the soci'al psychological influences on cigarette smoking (Behavioral Research, Laboratory Annu!al Report 1980, Part II). Moreover, as a result of a review of the literature, it was proposedithat the focus of the program be on the development of a general' model of psychosociial factors and'smoking,behaviors. It was proposed that the development of a new model be undertaken because the tradi- tional expl'anations of social psychological influences on smoking behavior are severely limited. These traditional explanations are based on one of two ap- _ proaches: (1) marketing survey technique, which does little more than provide simple descriptive information; or (2) a univariate model that explores the cause-effect relationship among,dependent and independent variables. Most sociial behavior consists of a complex set of behaviors influenced' by ani array of variables interacting in some combination or pattern. What many re- searchers have overlooked is that cigarette smoking i~s also a multiplex social phenomenon and, therefore, shou~ld be studied'using, an approach that will allow us to explore it as such. The research approach used'for this study invol'ves. multi - variate research technique. This technique allows an examinationi of common- alities in variance among interrelated variiables to identify non-obvious connec- tions, so-called higher order constructs. To illustrate the point, consider the relationship betweem height and weight. Height and weight are not causal.ly related, but interrelated. That is, one of these variabTes does not cause the other, but they are related,. They share commonivariance components with a higher order construct, "body mass". Multivariate research technique provides a robust and reliable method for identifying these kinds of inferred, higher order, con- structs. 1000392176

I I I F I I I I I F L I L L L L The goal', of the current research study is to discover reIationships between psychosocial factors and smoking behavior. A review of the literature led to the formulation of six theorems that re- late smoking behavior to social psychological correIates. 1. Smoking behavior is related to concern over one's own health. 2. Smoking behavior is rellatable to an indivi'dual''s degree of stress and choice of stress handling mechanism. 3. Smoking behavior wiIll vary with changes ini affective or emotional state. 4. There are specifiable environmentaI variables that influence smoking behavior. 5. There are specifiable socio-cultural influences relatable to smoking behavior. 6. Role definitioni or self-conception has specifiable effects upon smoking behavior. These six theorems together constitute a general statement of the social psychology of smoking. But the statements are, indeed, genera1ities. They are the guideliines for initiating the search for more specific knowledge. METHOfD, 10003g21.7-'7 Procedure In order to obtain information about the six psychosocial areas specified by the theorems, it was decided to gather naturalistic observations. A 1!ist of variables was drawn up for each of the six theorems, the being to obtain as much information relatable purpose to the six theorems as possible ~

F F F I I I I I I I- t F I I I L L withiin time and~resource constraints. In addition, a 1 iist was prepared of a num'- ber of smoking behavior variables. From these lists were generated questions, multiple choice items, etc. to be used in a combined interview and inventory- completion procedure. The method of investigation being used ideally requires that information be collected from a very large sample of smokers, with, speciial efforts made to, get extreme cases included. Not having unil imiited~ funds, a sample was chosen from a rigorously control'1led~ population of smokers. The respondents were selected from the two ends of the smoke-intake spectrum: ultra-Tow delivery brandi smokers and regular filter smokers. Bothi groups were drawn from among >45 year-old,, white, coll.ege-educated women residing in Virgi'nia. By thiis strict del ineation! we were able to partially control the extraneous variance normally encountered in randomi sampliing. It is also no coincidence that these demographic constraints define the group identified as the most frequent consumers of ultra-low delivery cig- arettes. Our findings, then, will be immediately applicable to amost critical segment of the cigarette market. Subjects The survey was conducted on a sample of 55 ultra-low deliivery smokers and 55 regular filter smokers. Thirty-one of the women were obtaiined'in the Richmond metropoliitan area by using all available recruitment resources, e.gi. the P.M. R&D mail survey roster (POL), a marketing research field service, local organiza- tions,, community announcements. The remaining, women were POL panel,ists obtai'ned in the Northern Virginia area. sooo392178 ~

I- F I I I I I I I I F I ~ F L L L Data Analysis We first translated interview information into categories, ranging~ from simple dichotomies through! ordinal categories to true intervals. This was the quantification step requiring the development of a codiing manual whereby clerical transcribers convert the informational! unit recorded'on the interview form into a number. Once in this form the data was ready to submit for computer processing. The first phase of the analysis was to identify the hiigher order common f actors among the hundreds of measures obtained. We used the multiple factor analysis method, effectively reducing the -391 independent measures to -55 retained factors. In the same manner we identified a number of smoking behavior f actors, from, among which we have sel~ected four for further consideration., In the second phase of the analysiis, the underlying relationships between the psychosociall factors identified in the first stage and the selected measures of smoking behavior were studied by either multiple regressioni or discriminant analysis (dependiing on whether the smoking behavior measures were categorical or continuous). These analyses provide a quantitative statement about the psycho- social variables that influence selected~smoking behaviors. RESULTS • 1000392179 Overview The factor analyses produced a set of factors for each, of the theorem- related psychosocial' variable sets. The factor analysiis also revealed four smoking behavior factors that should be studied: smoking; control, salience of delivery level, cigarette consumption, and brand switching. The sets of factors derived from the six theorem-related psychosocial vari- able sets were used in multi~plle regression analyses to explain the variation in each of the smoking behavior measures.

I- F F I I I I I I I i ~ I ( I L L Factor Analysis A factor analysi s was performed on each of the theorem-rel ated psychosoci al variable sets and the smoking behavior measures. Factors were selected from each variable set for further study by multiple regression analysis. Three criteria for retention, of a factor were used~: the factor is interpretable, the factor represents >2% of the variance in the original variable set, and the factor is likely to have predictive value. Presented below are the factors retained in each variable set, the factor number, a short phrase that describes the type of variables reflected in the f actor, and' the percent of variance in the original variable set that is carried in the factor. Smoking Behavior Measures Smoking Behavior Factor Anallysis: Four of the smoking,behavior factors were chosen as criterion variables to be expliained by the psychosocial factors. These four measures of smoking behavior are Iiisted below. See Table 1-A in Appendix A for asummary of questions in each factor. TABLE 1 Smoking Behavior Measures % of Total Variance Factor # After in Variable Set That is Rotation Description Represented by Factor 1 Smoking Control 11.4 2 Salience of Delivery Level 8.4 4 Cigarette Consumption, 5.2 8 Brand Switching 4.3 ~ O TOTAL 29.3 p w ~ N ~ O L