v-0 mm! . 9"!‘7’ ammuw -.' .I. I.- _~ — ”CO..- -‘ .5. l "0 hf." Id] “#13." 7.- ,N .,.‘.9. ‘f "J'Y‘(-'.: "W‘" W. s . .5 31);} “:9, K; I} kW? [gag Q n‘ v , . i. as ”123'. . ~$=~a¢=$ .x- ‘ ‘ "sud-:éérldq Egg“) 0_;'¢ §%:$J...->i°~ 45?» x}, ‘I M: ‘H|PV‘Y. is) )Og‘xzisjk ”A." $ . fig» . ,‘l' u..- _ FL. .1. 7.‘~ {fig—"J ob ’. ~§.::. . ‘ _‘ _ - 1:.“- 'z.L W W3? "x'."‘.-"“."3“5 ‘. _ K ' ‘9:..tk"£'w':l' v . ‘ -. (I . «3‘7; N "o- 'vv-o‘o}..:::f ~v::“':|:‘:; . ‘ y ‘ ‘l ”kl, :a I ‘I ’ . , (é; {IA‘K'IA 0‘ " 5:43;: fii'flcmy‘. ,v/Q pt I ".n.‘ “ :‘v. .' “L". :1 c‘M! v. . ”and”, is.“ 'J_ Liz/"Cr . ”I n f 4%: - I :2" it‘}:'_ «({1} '31” P :-J ”i” tic-Q; 34 5:1,! 0 I ‘\ “‘19? V?” ",‘gf’tg’li n‘_ ’9'." 5631}: ‘ F1 7'; $39.? 11",‘I'Ihf'af \~;;A' 1.2;. “MW "4.. A» ' "Vfi‘ ”1.2 D ”‘1' ' l'I' mgr!" ":11. ,1 5'9!“ . ”'4‘, u. ‘-. "19'" 393% 53%;” .L \I’l ‘ '- 7') . WI “ I I‘ (“'1 “IV", (.05 v ,2 3 21% l (Ill! llll lllllllll'll llll l .1 L/ MILIBRARY . thigan State Mi”annuity ~—-—_ ll This is to certify that the dissertation entitled Consumption Sets as Indicators of Consuming Style presented by Kathleen Marie Best Rassuli /has been accepted towards fulfillment of the requirements for _Eh_.D_._ degree in MEL M jor professor Datew 042771 MCIIicn-n Afr o‘ A ' I 7 rr PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before due due. DATE DUE DATE DUE DATE DUE w to 5 ~ MSU Is An Aflirmdlve Action/Equal Opportunity lnflitutlon CONSUMPTION SETS AS INDICATORS OF CONSUMING STYLE By Kathleen Marie Best Rassuli A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Marketing and Transportation 1988 ABSTRACT CONSUMPTION SETS AS INDICATORS OF CONSUMING STYLE By Kathleen Marie Best Rassuli In the past a number of prominent marketing scholars have called for the incorporation of the notion of assortments into research on consumers, as well as an orientation toward the process of consumption. This dissertation develops a new conceptual model of consumer behavior which incorporates both of these factors by introducing the concepts of ”consumption sets“ and "consuming style.” It is shown that the model is able to encompass a broader set of consumer behaviors, than brand choice alone, and is able to incorporate variables, hereto- fore classified as exogeneous, such as culture. The new model has implications for learning in consumer behavior. Furthermore, the model would also facilitates the study of consumer behaviors other than purchasing. The empirical work explores the existence of status-based differences in the construction of consumption sets. Data were gathered from members of two white collar occupations and one blue collar occupation. It was hypothesized that the set of attributes chosen by the two groups should be more alike within the group than between the groups. Cluster analysis, on the attributes chosen, was used to group individuals with similar consumption sets. The findings of the study show that there is a difference between the consumptions sets created by members of different occupational groups. It was also found that younger individuals and those with less established households tended to create an ”ideal,” rather than an ”actual,” set. To my husband, Ali iii ACKNOWLEDGMENTS I owe a debt of gratitude to a number of people who assisted me in this endeavor. First, I would like to thank my chairman and my dissertation committee. For my chairman Professor Harrell, anything is possible, and this is one of the best traits that a dissertation chairperson can possess. I would like to thank the other members of my committee, Professor Stanley C. Hollander, for helping me on theoretical issues, and Professor Forrest S. Carter, for helping with the methods. To all of the members of the committee, I owe thanks for the time they spent working through difficult issues with me and keeping me on track. Second, I would like to thank my husband, Ali. His encouragement started me on this journey; his love, support and strength helped me to persevere. I would also like to thank my parents Thomas B. Best Sr. and Arlene M. Best, and my grandmother, the late Julia Langel for the encouragement they provided me. I would like to thank George W. M. Bullion, Dean of the School of Business and Management Sciences, Indiana-Purdue University at Fort Wayne, for his support and encouragement. I am indebted to several individuals for their aid in the data collection process: Dr. Homayoon Seirafi and Mr. Stanley Gibson, Thomas and Arlene Best, Joyce Miller, Joe Good, Mark Durbin, and all of the individuals who participated in this study. A special note of thanks goes to Dr. Seirafi for the amount of effort he expended in this regard. In addition, I would like to thank Dr. William W. Baden of Computing Services at Indiana-Purdue University at Fort Wayne, for his tireless patience in answering questions and providing assistance with the data analysis. Finally, I would like to thank Mrs. Dianna Henderson for her aid iv in preparation of some aspects of the manuscript and Ms. Elizabeth Johnston for her work in editing the final draft. I am grateful to all these people, however, the responsibility for the final product rests on my shoulders. TABLE OF CONTENTS INTRODUCTION CHAPTER I: LITERATURE REVIEW APPROACHES TO UNDERSTANDING AND PREDICTING CONSUMER BEHAVIOR A. INDIVIDUAL DIFFERENCE VARIABLES l. Personality/Demographics 2. Psychographics/Demographics B. PROCESS VARIABLES: MULTIATTRIBUTE MODELS C. EXOGENOUS VARIABLES CHAPTER II: THE CURRENT STATE OF AFFAIRS A. GENERAL ASSESSMENT B. CRITICISMS OF THE FOCUS ON BRANDS C. THE GRAND MODELS: DIRECTIONALITY AND DYNAMICS l. Directionality 2. Dyanmics 3. Implications CHAPTER III: NEW MODEL A. OVERVIEW OF THE PROPOSED MODEL B. DEVELOPING A MODEL FOR LONG-RUN CONSUMER BEHAVIOR C. CHOICE D. THE COMPOSITION OF CONSUMPTION SETS Goods Attributes Consumption Sets Interrelationships and Points of Clarification 95”.“? vi NW!» IO l7 l9 19 23 32 32 35 36 38 38 45 46 48 50 50 55 55 DYNAMICS WITHIN A CONSUMPTION SET CONSUMING STYLE -- Further Dynamics in Consumption Sets Acquisition Consumption Limits on Consuming Style Discussion 95”.“? DO PEOPLE CREATE MEANINGFUL CONSUMPTION SETS? CONSUMPTION SETS AND CONSUMING STYLE ARE USEFUL TO CONSUMERS CHAPTER IV: IMPLICATIONS OF THE CONSUMPTION SET MODEL FOR THE CONSUMER BEHAVIOR LITERATURE A. H. QWWPQF’ CONSUMER BEHAVIOR MULTIATTRIBUTE MODELS LEARNING BUYER BEHAVIOR STOCHASTIC BRAND CHOICE ENCOMPASSING DEVIANT BEHAVIOR INFLUENCE OF CULTURE CULTURE AND SOCIAL CLASS CHAPTER V: HYPOTHESES AND METHOD A. HYPOTHESES B. OVERVIEW OF THE METHOD Instrument Subjects Limitations of the Sample Data Analysis .AP’NT‘ CHAPTER VI: RESULTS A. PARTITIONING B. CLUSTER INTERPRETATION AND PROFILING 1. Interpretation of Clusters by Attributes vii 58 60 60 61 63 63 66 68 71 71 72 73 74 74 75 75 79 81 81 83 83 85 88 89 92 92 96 96 H. DYNAMICS WITHIN A CONSUMPTION SET CONSUMING STYLE -- Further Dynamics in Consumption Sets Acquisition Consumption Limits on Consuming Style Discussion :5pr DO PEOPLE CREATE MEANINGFUL CONSUMPTION SETS? CONSUMPTION SETS AND CONSUMING STYLE ARE USEFUL TO CONSUMERS CHAPTER IV: IMPLICATIONS OF THE CONSUMPTION SET MODEL FOR THE CONSUMER BEHAVIOR LITERATURE A. H. QWWPOP’ CONSUMER BEHAVIOR MULTIATTRIBUTE MODELS LEARNING BUYER BEHAVIOR STOCHASTIC BRAND CHOICE ENCOMPASSING DEVIANT BEHAVIOR INFLUENCE OF CULTURE CULTURE AND SOCIAL CLASS CHAPTER V: HYPOTHESES AND METHOD A. HYPOTHESES B. OVERVIEW OF THE METHOD Instrument Subjects Limitations of the Sample Data Analysis :“PPI‘ CHAPTER VI: RESULTS A. PARTITIONING B. CLUSTER INTERPRETATION AND PROFILING 1. Interpretation of Clusters by Attributes vii 58 6O 60 61 63 63 66 68 71 71 72 73 74 74 75 75 79 81 81 83 83 85 88 89 92 92 96 96 2. Profiling on Variables Not Used in the Clustering Procedure 104 C. INTERNAL VALIDATION 109 D. EXTERNAL VALIDATION AND DISCUSSION 112 E. RESULTS OF HYPOTHESIS TESTS AND DISCUSSION 112 l. Hypotheses Concerning Consistency within Sets 112 . Actual versus Ideal Set 116 3. Discussion 1 18 CHAPTER VI: CONCLUSIONS, LIMITATIONS, AND FUTURE RESEARCH 124 APPENDIX A: CONSUMER BEHAVIOR MODELS 130 APPENDIX B: QUESTIONNAIRE 137 LIST OF REFERENCES 144 General References 153 viii Table 1.1 Table 1.2 Table 2.1 Table 2.2 Table 2.3 Table 2.4 Table 5.1 Table 5.2 Table 6.1 Table 6.2 Table 6.3 Table 6.4 Table 6.5 Table 6.6 Table 6.7 Table 6.8 Table 6.9 Table 6.10 LIST OF TABLES Explanatory Value of Variables Used in Early Consumer Choice Research Some Research Results from the Multi-Attribute Literature Examples of Conditions for Desire Competitors Example Conditions for Generic Competitors Example Conditions for Product Form Competitors Example Conditions for Brand Competitors Objective and Subjective Attributes Appearing in Questionnaire Abbreviated Original List of Subjective Attributes Pearson Correlation Coefficients Between Subjective Attributes Percentage Frequencies on Demographic Characteristics by Cluster Percentage Frequencies on Living Room Characteristics by Cluster Cluster Means on Subjective Attributes and Univariate F-Tests Percent of Cluster Members Who Used Attributes At Least Once Results of MANOVA for Attributes by Groups Percentage Creation of Products by Cluster Cluster Membership By Occupational Group Membership Actual Versus Ideal Living Room Set by Cluster and by Group Standardized Discriminant Coefficients & Discriminant Functions Table 6.11 Classification Results for Discriminant Analysis ix 13 27 28 29 3O 86 87 95 97 98 99 100 103 105 106 107 1 10 ll] Table 6.12 Breakdown of Occupational Groups by Cluster Table 6.13 Cluster Means for Living Room Activities Table 6.14 Activities by Cluster Table 6.15 Cluster Means for Other Variables 113 119 120 121 Figure 1.1 Figure 1.2 Figure 3.1 Figure 3.2 Figure 3.3 Figure 3.4 Figure 3.5 Figure 3.6 Figure 3.7 Figure 3.8 Figure 3.9 LIST OF FIGURES Overview of Variables Related to Choice Forms of Multi-Attribute Models Generalized Grand Model of Consumer Behavior Dynamic Model of Consumer Behavior LIST OF VOCABULARY Consumption Sets and Subsets Choice in the Context of Sets The Process of Filling a Consumption Set Attribute Dimensions Consumption Set Vanilla Advertising Showing Attribute Combination Figure 3.10 Perishability Continuum Figure 4.1 Figure 4.2 Figure 4.3 Figure 6.1 Figure 6.2 Figure 6.3 Figure 6.4 Consumption Possibilities Set When Barrier Exists Consumption Possibilities in the Presence of New Technology Class Similarities and Differences in Consumption Sets Dendrogram for Cluster Analysis Scree Diagram for Cluster Analysis Cluster Profiles on Subjective Attributes Attribute Commonality Across Clusters Figure A-l Large-System Models xi 10 12 40 41 42 45 46 47 52 53 57 59 76 78 80 93 94 101 115 130 Figure A-2 Figure A-3 Figure A-4 Figure A-5 Figure A-6 Figure A-7 Howard and Sheth Model EKB Model An Underlying Behavioral System as Viewed from a Segmentation Perspective Bettman Information Processing Model Carman’s Closed Loop More Detail on Consumption Set xii 131 132 133 134 135 136 INTRODUCTION To date the literature has tended to treat consumer behavior as a static rather than a dynamic process. The preponderance of work has focused on the choice and/or purchase of single products by individuals for singular situations. Calls for recognition and incorporation of a process orientation have been made (Jacoby 1976; Sheth 1979). At the same time, several prominent scholars have expressed the need for additional research avenues in consumer behavior, particularly more encompassing yet theoretically sound (Kassarjian 1978, 1987; Hirschman 1985, 1986; Belk 1985; Holbrook 1985, 1986, 1987). Theorizing requires abstraction. Empirical research is made manageable by the abstraction and simplification inherent in good theory. Moreover, the principle of parsimony suggests the phenomenon under study be explained in the simplest way. The difficulty arises when theories omit relevant constructs necessary to understand complex reality. Although critics of the present approach to consumer behavior have not explicitly stated the above point, the ”theoretical never-never land” (Kassarjian 1987), as it has been referred to, describes this difficulty in relating concepts to practical applications in consumer behavior. Many relatively disparate consumer research streams have been dominated by attempts to explain choice. In this dissertation, an explanation is offered by looking at the process of consumer behavior as a phenomenon that encompasses a great deal more than brand choice. The intent is to show how the broader treatment of consumption provides a mechanism to tie together several divergent theoretical avenues. It utilizes a new conceptual basis. 1 2 The premise is that the consumer behavior discipline can research and understand the interrelatedness of consumer decisions and the processes of consumer behavior. To so demonstrate, this dissertation will disentangle several related concepts and phenomena. In addition, it will propose several process variables that can be incorporated into consumer models for useful understanding and prediction. CHAPTER I: LITERATURE REVIEW APPROACHES TO UNDERSTANDING AND PREDICTING CONSUMER BEHAVIOR Consumer research is comprised of several important streams, including demographics and the study of personality, life-style, and attitudes, which have attempted to explain consumer behavior using parsimonious theories. The pattern of work within each stream has been similar. An introduction of the theory/- method and findings is followed by a critique of the appropriateness of the theories as explanations of consumer behavior. Generally, reasons are given for the inadequacy of the research track, such as a lack of theory, a misspeci- fication of the model, missing variables and/or poor, errorful measures. Sub- sequently, the research is improved or, often, the research stream is abandoned. Choice and/or purchase are two important variables in the study of consumer behavior. Different research tracks address understanding, explaining, and predicting consumer choice. Still, much of the literature of consumer research deals directly, indirectly, or tangentially with choice. Those models that deal directly with choice include the decision-making process, communication hierarchy, and stochastic choice models, to name a few. Preference and attitude models often address choice; that is, their main purpose is to explain preference formation, and correlations with brand choice are often used to describe the validity of the models. In addition, research streams have developed to study certain facets of larger decision processes, and often they tough tangentially on choice. For instance, in the literature on information overload (Jacoby, Speller, 4 and Kohn 1974), correct product choice is used to determine overload. Here, also, choice provides a test of model validity. In fact, one taxonomy of consumer behavior models shows that all models (with the exception of pure evaluation models) touch on choice (see Lilien and Kotler 1983, p. 205). Refer to the Appendix A. Purchase decision models (choice models) can be broken down into two major categories (Pessemier and McAlister 1982). The first includes generally deterministic models that assume complex, highly involved decision making. The second includes stochastic or random choice models. In the deterministic stream, understanding and explaining the reasons for purchase/choice are weighted at least equally with prediction of choice, whereas research into stochastic choice has been limited to Mg choice alone (Sirgy 1985, p. 108). With the exception of some cross-over works (Blin and Dodson 1980, Ahtola 1975, McAli- ster 1979), little or no attention is given in the latter research stream to understanding the reasons behind choice. Often, as in the case of attitude formation models, understanding tends to be weighted more heavily than predic- tion. Within the deterministic literature, which attempts to predict as well as explain, modeling the consumer’s decision-making process has taken a variety of forms. Large-scale models attempt to show the relationship of a broad array of variables to the decision process. Others have attempted to link a limited number of variables to choice. Often the research has sought to find good segmentation bases. Examples include demographic, personality, life-style, and attitude variables as they relate to choice processes. Many of the well-docu- mented multiattribute attitude models are of that type. For purposes of describing modeling approaches, the variables linked to choice can be classified into three categories. These will be used in the next 5 section to discuss models related to this research. Individual difference variables are used to divide people into homogeneous groups (demographics, personality, and life-style). Process variables describe how choice occurs (preference and attitude formation). Broad environmental exogenous variables (culture and social class) provide a context for describing the influence of norms and societal variables.l Each of these areas will be reviewed in turn. The focus will be on the models as predictors of choice. A. INDIVIDUAL DIFFERENCE VARIABLES 1. Personality/Demographics Early research attempted to find a link between personality, demographic characteristics, and product (or brand) choice. For example, in a 1959 study by Evans of the choice between Fords versus Chevrolets, a statistically significant R2 of .04 was found for personality and R2 s .05 for demographic variables; in combination they predicted 8 percent of the variance in choice. In 1971, Fry reported a study of cigarette brand choice based on personality characteristics. He noted that previous researchers had usually found R2 values of less than 10 percent, so he was encouraged by the fact that his Rz’s ranged from 10 percent to 30 percent. Two points are worth mentioning. First, Fry’s analysis was not on individual brands but on two groups of several ”like” brands. Second, socioeconomic variables had an R2 equal to .068, and socioeconomic and per- sonality variables together predicted 29.2 percent of variability in brand choice 1Howard and Sheth (1967) p. 494 discuss what they mean by exogeneous variables. They are variables which "we do not wish to explain their formation and change...we wish to separate the effects of this environment which occurred in the past and not related to a specific decision from those which are current and directly affect the decisions that occur during the period the buyer is being observed.” 6 for the ”high self-confidence group" (only 15.7 percent for the low self-con- f idence group). (See Fry 1971, p. 301.) Another personality study (Alpert 1972) used the Edwards Personal Pre- ference Schedule (EPPS) to relate such personality characteristics as achieve- ment, affiliation, dominance, and nurturance to the choice of product attributes and products (for cars, movies, and homes). Only the results for houses (place of residence) were published. The highest correlation found was 11 percent between social activities (at place of residence) and need for change (per- sonality). The study was unable to predict product choice. Using canonical correlation (and various other mathematical manipulations) Alpert found stronger relationships (R3 between .46 and .35) between groups of personality traits and groups of attributes. However, he noted that the relationships were not readily interpretable (See Alpert 1972, pp. 89-92). Table 1.1 summarizes some of the results of this early research. Table 1.1 Explanatory Value of Variables Used in Early Consumer Choice Research Demographics -- Choice .05 to .07 Personality -- Choice .04 to .11 Demographics and Personality -- Choice .08 to .30 More recent research reports findings along the same lines. For example, Saranson et al. (1975) analyzed 102 personality studies. On average they found 7 R2 of .045 for situational variables, .03 for personality variables and .01 for demographic variables. (Also see Peterson et a1. 1985.) According to Wells (1975, p. 196), "work with personality inventories has been judged ’equivocal’.... The correlations have almost invariably been low, and the relationships uncovered have been so abstract that they could not be used with confidence in making real-world marketing decisions.” Kassarjian and Sheffet also used the same word saw when they evaluated the literature in 1975. Roscoe, LeClaire, and Schiffman (1977, p. 69) note that most often the explanation given is that low R2 will occur as long as we are dealing with discrete dependent variables and continuous independent variables (1977, p. 69). While this is correct, other explanations are in order. High on the list are personality and psychographics. 2. Psychographics/Demographics As it became apparent that efforts in the area of demographics were not highly fruitful, emphasis shifted to life-style research (see, for example, Levy 1963). Wells (1975, p. 196) called the area of psychographics a blending of personality inventories and motivation research. He pointed out that in their effort to know consumers better, researchers wanted a variable that would provide them with ”new, more comprehensive, and more exciting descriptions” (1975, p. 196). Wind and Green (1974, p. 106) listed the various ways that a person’s life-style had been measured in the literature: (1) the products and services he/she consumed, (2) activity, interest, and opinion (AIO) variables, (3) the person’s value system, (4) personality traits and self -concept, and (5) attitudes toward benefits of product classes. 8 Since there was a lack of consensus as to which of the five was best, various combinations were used in practice. While any number of confoundings could have occurred as a result, the practice does not appear to have changed in recent years. Wind and Green attribute the practice to a lack of theory in the life-style research stream. Fenwick et a1. (1983) draw a similar conclusion. According to them, the "analyst sets out to measure a many-sided, ill-defined concept ’life-style.’ The components of this ’life-style’ are both unspecified and by design unexpected” (1983, p. 71). Some recent work on life-styles assumes that certain groups constitute "a life-style" and begin analysis at that point (for example, ”housewives” and ”working wives;" see Jackson et al. 1985; also see Andreason 1984). Wells (1975) concludes that psychographic variables explain a large portion of the variance of individual behavior. Darden and Perrault (1975), in a typical study, found that life-style ac- counted for about 15 percent of the variance in behavior (vacation choice). A recent study by Andreason (1984) has found similar results by looking at changes in life-style and changes in preferences. The correlation coefficient was .14. Several reasons have been given to explain the low predictive power. Roscoe, LeClaire, and Shif f man (1977) point out that the purpose of psychograp- hics is not to predict behavior, but to help explain and describe consumers. In addition, much of the literature has pointed to the lack of reliability of psychographic variables. Wells seems to have instigated these investigations when he said that ”unsuccessful prediction can be due to unreliability in dependent variables, to lack of any ’real’ relationship between psychographics and behavior in question, or to some combination of all three. In the absence of reliability data it is impossible to determine which is the case” (1975, p. 205). Reliability is not uniform in psychographic studies. Fenwick, Schellinck, and 9 Kendall (1983) cite a range in coefficient alpha’s from 0.40 to 0.80. These authors note: "Useful lifestyle measures will be those that lie somewhere between being almost totally redundant and being entirely unrelated to the behavior being studied“ (1975, p. 59). In addition, Kinnear and Taylor (1976, p. 424) contributed to the original controversy by showing that psychographic segments, developed using factor analysis, were unstable and, therefore, not construct valid. These authors note: ”What is likely to happen is that the user will assign a ’face’ validity to the segments developed and their catchy descriptions. The results may seem all too good to those unaware of their potential instability” (1976, p. 424). The nature of the life-style measure itself has been questioned. Wind and Green (1974, p. 102) stated: "Little appears to be known at this time about the relationship between the more specific versus the more general life style items.” On this same point, Reynolds and Darden (1974, pp. 90-91) felt more empirical research was needed to determine the relevancy of the issue. They cited a study by Ziff: ”She concluded that more product specific measures yielded relatively more insight than did general measures” (Reynolds and Darden 1974, p. 90). They concluded that while product-specific measures are needed for prediction, the cost to construct such measures is often not worth the benefit. The strength of the relationships documented in the literature are sum- marized in Figure 1.1. Each arrow represents a separate stream of study; it is inappropriate simply to sum up the correlations. 10 Figure 1.1 Overview of Variables Related to Choice Demographics W %‘ Psychographics, \z Choice .15 / Personality/ .04 ~ .11 In summary, demographics probably predict life-style better than brand choice. Life-style (psychographics) predicts choice somewhat better than does personality or demographics. Unfortunately, one cannot make a statement as to whether this difference is statistically significant, since different studies are involved, as well as different techniques. B. PROCESS VARIABLES: MULTIATTRIBUTE MODELS Research in consumer behavior, particularly during the 19705, has been overwhelmingly dominated by the multiattribute attitude models (Sheth 1979). The early work on attitude began in social psychology with the research of Peak (1955), Rosenberg (1956), Fishbein (1966), and others (see, for example, Jones 1955). Frustrated by attempts to predict behavior from socioeconomic and per- sonality characteristics, Bass and Talarzyk (1972, p. 93) hoped that measures "specific to preference alternatives" would improve prediction of consumer behavior. 11 The essence of the multiattribute model is that attitude can be broken into (in the case of the decompositional approach) or is made up of (in the case of the compositional approach) two components: beliefs about attributes and an evaluative component. The nature of the evaluative component depends on the respective version of the multiattribute model with which one is dealing. It is either the likelihood of possessing an attribute, the importance of the attribute, or an evaluation of the outcome of an attribute. Members of the multiattribute school believe that once attitude is known, then it can be used to predict and understand preference (that is, affect), purchase intention, and choice. In the information processing literature, choice is viewed as the result of a linear, compensatory heuristic. In other words, given a number of attributes, the consumer chooses by weighting and summing up the values of attributes across brands; the brand with the highest weighted average score is chosen. A generalized form of these models is given in Figure 1.2. Some of the results from the literature are presented in Table 1.2. Again in this literature, as in the other stream, the strength of the relationship between attitudes and behavior is varied (Harrell and Bennett 1974). In the psychology literature, Ajzen and Fishbein (1977) report correlations of between 0.02 and 0.49. The relationships found in marketing have not been as high, 0.20 to 0.30. Furthermore, combined attitude components tend to predict behavior (choice) better than evaluations of attributes ()3 b) alone. Nakanishi and Bettman (1974, p. 18) conclude: ”The X aibij model is slightly, but discernibly more descriptive than the Z bij model in this comparison." Also, the relationship between )3 ba and attitude is relatively stronger than that between attitude and choice. Early debates in the literature concerned the form of the model, that is, expectancy-value versus attribute adequacy (see Bass and Talarzyk 1972; Sheth 12 .5323 2802 2352...: a 25$ 222 . m 22.8.5 .8328 u a N4 05mm msgae8fieom u om 2335 32.8 n 5252 8.85358 88835 n :m a agnémesaa 2. .88 “ES 1 J _m 82 Pm u mean .8258 .83 85358 38m - 2.32 29.8.5328”. 22:95 uncommon. 83:3 1 _ .2 8358.20 £830 13 Table 1.2 Some Research Results from the Multi-Attribute Literature Ba ----- > Aact .40 ~ .46 Ba + 2 NB ----- > BI .08 ~ .38 Ryan and Bonfield (1980) Aact + SC ----- > BI .24 ~ .58 Ba ----- > Aact .40 Lutz (1977) Aact ----- > BI .69 Aact ----- > B .20 ~ .30 Aact ----- > B .02 ~ .49 Ajzen & Fishbein Aact ----- > BI .16 ~ .67 Review of Literature (1977) BI ----- > B .21 ~ .97 14 and Talarzyk 1972; Cohen, Fishbein and Ahtola 1972; Bass 1972; Sheth 1972; Talarzyk 1972). Later emphasis focused on improving measures of various model components. The multiattribute models not only found their way into studies of communication strategy but also into marketing strategy. Multiattribute f ormula- tions have filtered into benefit segmentation and positioning and, therefore, into strategic marketing planning (Wilkie 1986, pp. 483-90). Moreover, a number of strategic decision models are founded upon multiattribute assumptions, such as ASSESSOR (Urban and Katz 1983). Notwithstanding early hopes, the strength of these models appears to be that of understanding consumer behavior, that is, diagnostic capabilities. Wilkie and Pessemier wrote: While one might be able to choose better predictors of these depen- dent variables [purehase or purchase predisposition]..., diagnostic benefits often are not offered by the better predictors... The basic purpose of the multi- attribute models is to gain an understanding of purchase predisposition (1973, p. 429, emphasis added). Although these models appear to have continued in their function as a diag- onostic device, they have been questioned as a model of choice. For example, Reibstein (1978), who uses the probabilistic approach to model the multiattribute problem, concluded that ”the probabilities of brand choice based on the multi-- attribute attitude model are not closely related to the actual choice exhibited in the experiment” (p. 166; also see Blin and Dodson 1980). Reibstein (1978, p. 166) shows that behavioral intention has the highest correlation with actual relative frequency choice vectors for 65 percent of the subjects, preference measures for 29 percent and the multiattribute model for 6 percent (1978, p. 166). Disappointing results have been variously attributed to a number of factors. Ajzen and Fishbein (1977) felt that incongruence between the content of attitude questions (which focused on general attitudes) and the type of behaviors or actions studied (specific) could be blamed for weak relationships. 15 Studies that measured the attitude-behavior relationship in the context of specific situations did, in fact, attain higher correlations. They also noted that poor measures were at least partially responsible for the lack of correspon- dence. Laroche and Brisoux (1981) found a slight improvement in R’, as a result of also including attitude toward competitors’ brands with attitude toward the brand under study. Fishbein and Ajzen (1974, p. 63) found better prediction when attempting to predict a multiplicity of behaviors as opposed to just one. Bagozzi et a1. (1979) noted that lack of consideration for the "tripartite” nature of attitudes, (cognitive, affective and behavioral intention components) may be the reason for poor results. They found all three components to be construct valid. Bagozzi et al. (1979, p. 93) concluded: "Most previous studies in marketing have relied on a single measure of attitude and have focused only on cognitions or beliefs. Very few studies have examined the affective dimension of attitudes, and virtually none have investigated the tripartite model.” These authors also raised another important point: "To the extent that products represent bundles of attributes, one might expect that attitudinal responses of consumers would be heterogeneous and complex at least during the early stages of information processing” (1979, p. 93). Another critique of the multiattribute models is related to models of information processing. This point was elaborated upon by Bettman (1979, chapter 7). As previously stated, the multiattribute model is a compensatory model, and it has been shown that consumers use other choice heuristics as well. For example, consumers may use noncompensatory models, such as lexico- graphic models. In this case, the person compares all products using brand processing and attribute processing, Bettman (1979, p. 185) concludes: "Thus the notion of choice rules as yielding attitudes appears to be inadequate, at least as far as the idea that choice rules provide an attitude measure as a direct 16 output.” In fact, ”choice can be made without forming such an overall evalua- tion (e.g. lexicographic)” (Bettman 1979, p. 209). The work of Bagozzi (1982) tends to support Bettman’s belief. Using a causal framework, Bagozzi showed the amount of variance explained by expectancy-value judgments (22 ba) for affect (attitude, Am), intention, and behavior. The results, on average, were 0.56, 0.24 and 0.14, respectively. The one result which he found to be fascinating was a direct effect from expectancy-value to intentions (1982, p. 581), which implies that an attitude is not formed. Moreover, it appears that the conclusion drawn by Reibstein in 1978 with regard to the multiattribute literature, still holds true. He said: Many efforts have been directed to predicting choice behavior from the multi-attribute attitude model.... Results have been mixed, probably because many factors may intercede before the actual choice is made. How much information is contained in attitudinal measures about actual choice behavior for an individual over several purchases remains to be ascertained" (1978, p. 164). Hence, we may not be considering all the relevant factors affecting choice. In summary, multiattribute models have been useful in deciphering various aspects of information processing, the components of attitudes, and relationships of attitude formation to other variables. They have been helpful in understand- ing and predicting choice. From the perspective of predicting choice perhaps a more significant critique would stem from the researcher’s assumption about attributes. Basically, these models are founded on the assumption that attributes are important. Researchers then ask ”which" attributes are important, ”how" people use attributes in decision making (the evaluations component), and "how” people attach significance to attributes. However, all models fail to include antecedents that describe why attributes are important. Therefore, research in this area has not yet addressed important opportunities that explain and predict behavior as opposed to information processing. 17 C. EXOGENOUS VARIABLES Attempts have been made sporadically to show that variables such as culture and social class are related to consumer decision making and choice. In the large-scale models, these variables are usually viewed as exogenous to the decision maker or consumer (Sirgy 1985) and for the most part are taken as given. However, there have been studies that correlate exogenous variables with choice. The social class literature provides an illustrative case. It has been argued that social class ought to be related to consumption behavior (for example, Coleman 1960). Research in the late 19405 discovered different consumption goals for upper-middle-class versus lower-middle-class Americans (Coleman 1983, p. 269). This was true for home furnishings, appliances, clothing, and food. Social class came to some prominence in marketing in the 19503, but there was little improvement in the research beyond 1960 (Bettman, Kassarjian and Lutz 1978). Coleman (1983) believes that social class, as a legitimate area of study in consumer behavior, was hampered by the fact that sociologists disagree about the "value and validity" of the concept. Moreover, measurement problems and expense as well as questions about when and how to apply social class also led to problems. Finally, Coleman (1983) believes that the controversy over income versus social class led researchers to shift focus to other variables. In their 1978 review, Bettman, Kassarjian, and Lutz concluded that social class was in the "decline stage” as an area for research. Coleman (1983, p. 269) attributes the decline to the emergence of alternative research traditions, especially life-style. Authors such as Levy (1966) and Myers and Guttman (1974) consider social class variations to be variations in life-style. 18 An ongoing debate in this literature is the question of whether social class is a better predictor of purchases than income, generally in terms of market segmentation (Curtis 1972; Myers, Stanton, and Hang 1971; Schaninger 1981). Wind (1978) concluded that both income and social class are needed for segmen- tation. Schaninger (1981) concluded that the answer depends upon the product class. Social class is superior for consumer expenditures that reflect underlying life-style differences (Schaninger 1981, p. 206). Furthermore, income is a better predictor for major appliances, and a combination of income and class are necessary for symbolic products (ibid., pp. 206-207). More recently, Coleman (1983) believes that research should not focus on whether income or class is a better predictor. Rather the new question should be class affect the use of income. Coleman pleads the case that social class is still relevant to the study of consumer behavior. Fisher (1986, p. 492) has argued that social class is relevant to consumer behavior research, but several problems exist in the present literature. Accord- ing to him, marketing researchers have paid little attention to reliability and validity issues. Moreover, research is plagued by naive conceptualization. Finally, marketers have relied on Warner’s notion of discrete class membership (ibid., pp. 492-93). Fisher refers researchers to Max Weber’s work, which he says offers an explanation of why social class is related to consumption. Renewed interest in social class has come in the form of the ”sociology of consumption” (see Mochis 1981 for a review of consumer socialization). This tradition examines institutions, cultural values, and role structures to determine their impact on consumption (Bettman, Kassarjian, and Lutz 1978). Social class merges with topics such as consumption symbolism (Belk, Bahn, and Mayer 1986; Solomon 1983) and consumption’s cultural context (Hirschman 1985). CHAPTER II: THE CURRENT STATE OF AFFAIRS A. GENERAL ASSESSMENT While some authors may disagree, the consumer behavior discipline seems to be in a state of flux (Holbrook 1987). The search for a new approach appears to be at least as intense as at any other time in the history of the discipline, perhaps more intense. In fact, Kernan (1987, p. 133) says: ”[Holbrook’s] plea is not the ranting of a lunatic fringe, but rather a position that represents a growing number of consumer researchers who, for whatever reason, have been unable or unwilling to express it." In the minds of many leaders of the discipline, it appears that conventional theories and techniques have taken consumer researchers to a certain point in understanding the consumer but have reached their limit. Evidence of discontent came as early as the mid-19703. Mittelstaedt (1971) and Sheth (1979) criticize what the former refers to as the "eclectic borrowing" on the part of consumer behaviorists. Some of the main criticisms are summarized by Sheth (1979) and Jacoby (1976), who provide a convenient framework for discussing suggested new research directions. Both authors point to problems with consumer research, and both offer numerous alternative solutions. Other writers have at various times joined Sheth and Jacoby in calling for new theories and new approaches. According to Sheth (1979), three main difficulties plague research in consumer behavior. The first is the implicit assumption of a rational problem- solving process, as exemplified by multiattribute information processing and brand choice models (ibid., p. 573). Olashavsky and Granbois (1979) also question 19 20 the assumption of rational decision making, and conclude that perhaps no decision making at all occurs. As a remedy, Sheth suggests that researchers turn their attention to habit and conditioning, situationalism, novelty-curiosity, deviant, and obsessive consumer behavior. Along the same lines, Zajonc (1980) and Zajonc and Marcus (1982) suggest a shift from cognitions to affect. Gardner (1985) summarizes research to date on affective dimensions of mood states. Rock (1985) explores ritual behavior. Foxall (1983) suggests a shift to be- haviorism implicit in such notions as habit and conditioning. Sheth points to the overemphasis on the individual and the lack of focus on groups as the second and third problems in consumer research (1979, p. 573). According to him, focus has been on the individual shopper, buyer, decision maker, and user (ibid.). Even such groups as market segments, social classes, and ethnic groups are studied as "aggregates of individual consumers rather than distinct group entities” (ibid.). As a remedy to the exclusive focus on in- dividuals, Sheth suggests research should examine dyads, small groups, families, and organizations (ibid.). Sirgy (1985, p. 112) also believes that consumer research has not adequately answered the question of how family decision making can be distinguished from individual or organizational decision making. Sheth makes several suggestions regarding to the study of groups. Re- search is needed into nonproblem-solving group behavior. Moreover, "research should be directed at the m (group) rather than at the r_n_ic;Q (individual) level" (1979, p. 514). For instance, household decision making, organizational buyer behavior, the sociology of consumption, and cross-national buyer behavior would be t0pics to include (ibid., p. 517). Hirschman (1985), Belk (1985), and Uusitalo and Uusitalo (1981) have similarly called for more macro research. Alderson (1957) developed the rudiments of a consumer theory with the house- hold as the unit of analysis. Glock and Nicosia (1964), Nicosia and Mayer 21 (1976), Foxall (1976), and Zielinski and Robertson (1981) all have called for the interjection of sociology into the study of consumer behavior. Others have proposed the study of history both as an alternative technique and as a method for infusing a macro perspective (Savitt 1980; Kirkpatrick 1983; Hirschman 1985; Rassuli and Hollander 1986). Sheth further suggests that consumer behaviorists must develop their own constructs rather than rely on either psychology or sociology. For example, typologies of consumption needs/wants and consumption life-styles, and a consumption life style must be developed (1979, pp. 516-17). Belk (1985) develops a more macro approach in his materialism scale, but the scale is again based on an aggregation of individual data. Turning to a partial list of Jacoby’s criticisms, additonal problems with consumer research are illuminated. He points out that the discipline uses static methods to understand dynamic processes. While ”99% of consumer behaviorists” believe that consumer behavior is dynamic, all researchers measure after the fact (1976, p. 3). According to the author, such methods lose the richness of the processes involved. He recommends the use of dynamic methods. Jacoby also suggests that the discipline should focus on consumption behavior as opposed to just buyer behavior (1976, p. 10). Alderson (1957) was among the first to call for such a shift, noting that there is a difference between buying behavior and consumption behavior. Theories of both types of behavior are needed (Alderson 1957, p. 166). Similarly, Nicosia and Mayer (1976) define consumption behavior as buying, use, and disposal. Jacoby’s also criticizes researchers’ exclusive focus on market exchange (although he does not use those words). He suggests a need to look at varieties of acquisition behavior, such as gifts, loans and trade (1976, p. 10). Belk’s work (1976) on gift giving is an example of such a shift in focus. In addition, the humanistic research of ”The Odyssey 1986” (Kassarjian, 1987) exemplifies of 22 research into such acquisition phenomena as flea markets. Those who perceive a need to understand other types of acquisition behavior often believe that a change in metatheory is needed. Anderson (1986), Hirschman (1986), and Uusitalo and Uusitalo (1981) propose ”humanistic” research methods, such as the "Odyssey.” According to Uusitalo and Uusitalo (1981, p. 559), “progress is not attainable merely by applying traditional models of consumption to new areas of phenomena.” What is needed, in other words, is a shift from the logical empiricist (positivist) tradition of thought to the historical-institutional (hu- manistic) tradition. Finally, Jacoby suggests that the heretofore unrelated domains of consump- tion and production are in actuality ”integrally related” (1976, p. 10). Resear- chers should ”examine this inter-relationship by considering both domains simultaneously” (ibid.). Pollay (1986)1 also alludes to consumption and production activity. In conclusion, the call has gone out for new theories, new methods, and new ways of approaching consumer behavior questions. Scholars have proposed a variety of alternatives, some of them are radical. Solutions range from minor changes in theoretical focus within brain functions, to changes in metatheory. Often a change in emphasis is viewed as being accompanied by a change in method, or preceded by a change in metatheory. Included in the list of sugges- tions are the following non-mutually exclusive categories: 1. change in focus from cognitions to affect; 2. change in school of psychological thought from cognitive learning to behaviorism; lUnpublished lecture materials presented at Association for Consumer Research Meeting, Toronto, Canada, October 1986. 23 3. shift in focus from psychological theories to theories from other related disciplines, such as sociology, and anthropo- logy; 4. change in methodology or technique of analysis, for example, from questionnaires to projective techniques, and/or from respondent answers to researcher observation, and/or from correlational research to causal modeling; 5. change in metatheory and the techniques of generating theories, for example, from logical positivism to humanism. The list serves to reinforce the point, made at the beginning of this section, that consumer researchers are seeking a new approach. It would not be unrealistic to conclude that the ggestigns of consumer researchers have not changed; members of the discipline simply seem to be dissatisfied with the answers they have derived to date. It is sufficient to note that interest still lies in understanding why consumers behave in certain ways, in explaining why they behave, and in predicting how they will behave. However, many of the proposed solutions have been rejected by a large portion of consumer researchers (for example, see Kernan 1987; Cooper 1987). It is argued here that consumer research critics may have, in part, misidentified the problem. The section that follows presents another plausibe explanantion of the discontent in the literature. B. CRITICISMS OF THE FOCUS ON BRANDS The consumer behavior discipline came to prominence in the late 19503, grew through the 19603, and gained a strong foothold in the 19703. Marketing professors and researchers developed consumer behavior as a separate sub- 24 discipline in the 19503 at a time when the term marketing concept was becoming popular. Increased competitiveness and the development of new technology fed upon one another. Hirschman (1985) believes that the latter influenced research in consumer behavior. The marketing concept’s emphasis on profit through the satisfaction of consumer wants/needs resulted in an urgent need to know the consumer. Marketing managers had to learn of consumer needs and design marketing mixes to meet them. Moreover, marketing expenditures could be made more efficient and marketing strategies more effective if they were tailored to the needs of segments of consumers. Marketing management led researchers to develop models and theories to provide at least two types of data. First, researchers had to search for the variables mated to choice (also, perception and preference "because of the loss of information when one is considering only choice," in the words of Lilien and Kotler 1983, p. 364). Second, researchers needed to discover the variables that W between groups (segments) on the basis of choice. According to Wilkie (1986, p. 7), early research focused on the descriptive question of ”who” bought, while later emphasis shifted to inferential research into the questions of 'why" and ”how.” From the beginning, scholars realized they were dealing with multidimensional, multicansal phenomenon. Speculating on the answer to questions of how and why consumers behave, Cassady wrote (1940, p. 119): Indeed -- to know why a person reacts the way he does, one would have to know that person’s entire life and his parents’ lives, and the lives of every one he’d ever met. One would have to know every word ever said to him, every word he’d ever said, what he’d seen, read, thought about, done, even what kind of food he’d eaten. Clearly, Cassady’s statement foreshadowed the subject matter of the discipline. The table of contents of any contemporary consumer behavior text provides 25 additional evidence (for example, Harrell 1986; Wilkie 1986; Engel, Blackwell and Miniard 1986; Assael 1987). Factors affecting why people buy range from culture, to reference groups, to family, to individual influences. The number and complexity of the variables and relationships have been demonstrated in such grand theories of consumer behavior as those of Engle, Kollat, and Blackwell (1986) and Howard and Sheth (1967) (see Harrell 1986, pp. 557, 555). In attempting to answer these questions, however, one could argue that research has followed a rather specific and narrow path. In the prelude to their section on consumer behavior, Sheth and Garrett (1986, p. 460) note the over- emphasis on brand choice. ”The consumer behavior school has attempted to understand and theorize the buying behavior of consumers, especially with respect to brand loyalty and brand choice behavior. The overemphasis on brand choice has been noted elsewhere (Wind, 1977). In discussing their model of consumer behavior, Howard and Sheth (1967, p. 477) make their assumption explicit: ”Much of buying behavior is more or less repetitive brand choice decisions.” There appears to be tacit agreement among consumer researchers about the significance of brand choice. However, the extent to which the focus on brand choice has influenced consumer research has gone undetected. Consider, for example, the diversity of models in essence founded upon a brand assumption. The Howard and Sheth model is a brand choice model. While Engle, Kollat, and Blackwell do not make such an assumption explicit, their model also deals mainly with brand choice. This is particularly true if one considers the modifications of their decision-making process. Extensive, limited, and routinized decision-making behavior differs from each other, in respect to the extent of prior knowledge of product category attributes and brand attributes. Clearly, the multiattribute attitude models, as they are used in marketing, deal with brand attitude, pre- 26 ference, and choice. Attitude toward the object is generally attitude toward the brand, as denoted by the subscript j, (Am), conventionally used to denote the brand. Moreover, the demographic/personality literature began by trying to distinguish Ford owners from Chevy owners (Evans 1959), although in this literature, more so than others, product classes are studied. Finally, while research into exogenous variables as they relate to choice is often at the product form level (Westbrook and Fornell, 1979), here, too, brand choice plays a significant role. The implications of this brand f ocns have not been explored. Specifically, the nature of the difficulty of examining brand choice alone can be clarified by considering the implicit conditions behind the classification of competitors into different "levels of competition." Discussions of the topic are usually phrased in terms of the products viewed by consumers as possible competitors. What conditions must consumers meet in order to be in a position to make a decision in each of the four competitor categories --de3ire, generic, product form, and brand?2 The material that follows explores the answer to that question. Qgsirg cgmpgtitgrs are considered the "immediate desires the consumer might want to satisfy” (Kotler 1986). In order to be in a position to make a choice between or among desire competitors, the consumer must be planning to spend some money. That is, the person should have a given amount of discre- tionary dollars. Moreover, one would also have to assume that some type of internal need (desire) has been aroused or that some stimulus can arouse the need. The consumer is motivated to make some type of purchase. Table 2.1 lists examples of several conditions. 2Definitions for this discussion are taken from Kotler (1986, p. 129). 27 Table 2.1 Examples of Conditions for Desire Competitors 1. Some internal need has been aroused or some stimulus is required. 2. The consumer is motivated to purchase something. 3. The consumer is planning to spend money. 4. The consumer has some given amount of discretionary money to spend. aneric competitors are ”other ways in which the buyer can satisfy a par- ticular desire” (Kotler 1986). Here one assumes, in addition to the conditions for desire competitors, that the consumer knows exactly what desire needs to be satisfied, or if the person can be shown a series of alternatives he/she will recognize the need. In the case of generic competitors, disparate alternatives (in the transportation market, for example, a cruise ship, a car, a moped, a bicycle, and a pair of shoes) can be tradedof f against one another under certain circumstances. One such possibility occurs when the product in question is complemented (or accompanied by) other products. See Table 2.2. 28 Table 2.2 Example Conditions for Generic Competitors l. The consumer knows what specific desire needs to be satisfied. 2. If shown a series of alternatives, the consumer can recog- nize the need. 3. Disparate alternatives can be traded off against one another. 4. Products can be complementary to one another. Brdddgt fdrm competitors are "other product forms that can satisfy the buyer’s particular desire" (Kotler 1986). The consumer has recognized the need, chosen among generic alternatives, and now must decide on the form of the product. It is assumed that the consumer is i_11 the market (willing to buy one of the alternatives). Moreover, in the mind of the consumer products with different sets of attributes (such as station wagons, sedans, and convertibles) are viewed as equivalent substitutes. See Table 2.3. 29 Table 2.3 Example Conditions for Product Form Competitors 1. The consumer has recognized the need, has chosen among generic alternatives, and now must choose the product form. 2. The consumer is 13 the market -- is willing to buy one of the alternatives. 3. Products with different sets of attributes are viewed as equivalent or substitutes. Finally, at the level of b11031 competition, a consumer’s needs or desires are well defined and explicit. Products that are nominally alike in nearly all respects, except minor brand attribute differences and brand name, are viewed by the consumer as equal trade-offs (part of an evoked set) or substitutes. Generally, the consumer knows both the important and determinant attributes and is familiar with the attributes of various brands. It is useful to note here that routine, limited, and complex decision-making processes are formulated in terms which would place decision making at either the product form or brand stages. Wynn [1982, p. 32] notes that the Howard-Sheth model is not applicable where a person is not aware of the product class. The same is true to a greater or lesser extent for the EKB model. At the brand level, the brand name itself or the attributes of the brand should evoke a desire that is 39 well defined in the consumer’s mind, that the person makes an immediate inference about the pro- duct’s relation to his/her needs and the product’s solution capabilities (or without too much difficulty can be shown the relation). See Table 2.4. 30 Table 2.4 Example Conditions for Brand Competitors 1. Needs/desires are well defined and explicit. 2. The consumer knows the important and determinant at- tributes. 3. The consumer knows the brand attributes. 4. Products, nominally alike except for their brand name, are viewed as near trade-of f s (substitutes). 5. Product attributes are closely related to needs. While brand decision-making may represent a large portion of consumer decision making, it only explains part of consumer behavior. In sum, the decision-making that occurs at the brand level may differ greatly from decision-making at other levels. The discussion which just preceded, leads to a straightforward point. That is, perhaps the literature reviewed in Chapter 1 is limited to the extent that it carries with it the implicit conditions of brand choice. The final condition under brand choice, in Table 2.4, "Product attributes are closely related to needs," aids in demonstrating this point. Consumers making brand choices must be 39 aware of their needs that brand attributes become synonymous with needs, benefits, and consequences. Attributes, then, 93.11 be defined as "outcomes, consequences or benefits people obtain from a product" (Wilkie 1986, p. 483) and can, be used interchangeably with values and consequences, as some of the multiattribute literature uses attributes (see Cohen, Fishbein, and Ahtola 1972, p. 456). Then, 31 brands can be viewed as the means and attributes as the ends of consumer behavior (Wilkie 1986, p. 460). The ramifications of this simple brand assumption become enormous when one recognizes that modeling consumer behavior in this way narrows the focus of research. If attributes are synonymous with needs, benefits and consequences, then understanding of consumer behavior becomes a simple matter. The reason- ing might be as follows: If a consumer researcher understands the attribute consumers’ prefer in a brand and their attitudes toward attributes, then con- sumer behavior can be understood, and it should be possible to predict behavior (choice or purchase)" However, this scenario probably accurately describes the theoretical "never-never land" against which Kassarjian (1987) rails. The conditions for the modeling of brand choice help explain a great deal of the discontent in the consumer literature. All research streams appear to conclude that one theory cannot fully explain consumer behavior. Consumers buy on the basis of attitudes toward the sum of the attributes of a brand, which are simply equivalent to the person’s cultural values and life-style. But the sum of the evaluations of attributes only partially explains choice behavior. And while cultural values and life-styles are important, they do not add to our ability to predict brand choice because they simply may, the information contained in attributes. So in attempting to advance the literature, researchers have concluded that we must be missing any number of other variables: the situation (Belk 1975), reference group pressures, the family influence, emotions (131; "Call for Research on Emotions, 1985), feelings or affect (Zajonc 1980, Zajonc and Markus 1982), hedonistic tendencies (Holbrook 1981), irrationality (Sheth 1979), collecting behavior and having behavior (Belk 1982), and materialistic tendencies (Belk 1985). Yet, including the situation as a variable did not help predict brand choice, and the jury is still out on more 32 recent variables. Or perhaps, it is reasoned, brand choice is simply a stochastic process, or random to such an extent that it can be modeled as stochastic, as the literature on stochastic choice has always argued (Bass 1972; see McAlister and Pessemier 1982 for a review). Returning to an earlier discussion, Howard-Sheth is a brand model, as is EKB. Multiattribute models are brand models and unstated brand assumptions have been covertly carried into other areas. Marketing decision models which are based upon multiattribute models have carried along brand assumptions. These implicit conditions for brand modeling also spill over into routine, limited, and extensive choice models in both consumer and organizational markets. The assumptions are carried over into the segmentation literature and into position- ing and preference maps. The information processing literature grew out of the multiattribute stream, and the stochastic brand models explicitly attempt to explain brand choice. All of these models comprise the bulk of consumer behavior texts, while culture, social class, values, and lifestyle add little to our understanding and prediction of ”consumer behavior.” In terms of the explanation offered here, one would not expect values, life- styles, and so forth, to help predict brand choice. Indeed, in the world of brand competitors, factors such as values, culture, and lifestyles, are actually gi_ve_n or exogenous, or can be assumed to be beyond change or control -- just as most consumer behavior texts treat them. C. THE GRAND MODELS: DIRECTIONALITY AND DYNAMICS 1. Directionality What may be missing in consumer research is an understanding of how a product becomes part of a consumer’s life. This is implicit in Jacoby’s call for 33 a look at processes and "consumption” and in Sheth’s suggestions about consum- ing life-styles. In other words, part of the explanation for why people buy (also, why they form beliefs, attitudes, evaluative criteria, and so forth) might be found in how the product will fit into a consumer’s life. In essence, then, perhaps what is required for understanding and predicting consumer behavior is a change in the underlying questions asked. Rather than ask why people buy or on what basis people make their choice of products, one might inquire how the choice of product contributes to a person’s life. The work of consumer behaviorists in the early 19803 has begun to reflect the use of products (as opposed to simply the purchase of products). Studies by Belk on possessing and owning, by Holbrook and Moore (1984) on features, and by Solomon (1983) on products as props for consumer roles are representative examples. But the research of these scholars has tended to be combined with a call for more investigation of affect. Only recently has a small body of literature sprung up concerning what consumers do with products after purchase (see Belk, 1982). This literature, which focuses on what Belk refers to as "having behavior” (1982, p. 185), emphasizes that other fundamental consumer activities occur after purchase. As shown in the previous section, the majority of consumer behavior models and theories focus on some aspect of choice: preference, acceptance, intention, selection, or purchase. Examples include the decision-making process, which focuses on purchase decisions; the hierarchy of affects and communication models, which focus on acceptance of the message and purchase of the product; information processing models which end in consumption; and the diffusion of innovations models, which focus on adoption. In addition, the various research streams that have arisen to study certain facets of different decision processes often use choice as a test of model validity. For instance, in the information 34 overload literature, Jacoby, Speller, and Kohn (1974) used correct choice to determine the extent of overload. In addition to being focused on choice, all conceptual schemes assume that consumers move toward choice of a brand or product. Appendix A contains the grand models of Howard and Sheth (1967) and the EKB model (Engel, Blackwell and Miniard 1986). Also included are Bettman’s information processing model (1979) and Wilkie’s schema for segmentation (1986). It is noteworthy that in all of these the consumer arrows move in the direction of choice. While all models include feedback loops, these generally deal with feedbacks to evoked sets, decision rules, satisfactions, and perceptions. Furthermore, as shown in the previous chapter, many major streams of research dealt with the prediction of choice from other variables. Included here are choice as a function of: personality variables (Evans 1959; Fry 1971; Alpert 1972; see Saranson 1975 for a review of personality studies); demographic variables; psychographic variables (such as Darden and Perreault 1975; Andreasen 1984); attitudes (the multiat- tribute literature discussed previously); and a variety of stochastic choice models. Likewise, research in the area of social class has generally been focused on choice/purchase as a function of social class (Mochis and Moore 1981); see Mochis (1981) for a review of this literature.) More often, however, grand conceptual schemes treat variables such as class and culture as exogenous variables (see Sirgy 1985). In other words, consumer research assumes a single direction from recogni- tion to purchase/choice. Research into consumer behavior rarely, with the exception of those mentioned in a previous paragraph, considers what occurs after the purchase of a product. The focus of most grand conceptual schemes (EKB 1986 and Howard-Sheth 1967) and most intermediate stage theories (Bet- tman 1979) is the internal and external processes that Lead to the purchase of a 35 product. Thus, choice is the output of one or more consumer processes. With the possible exceptions of the satisfaction literature and the dissonance litera- ture, little consideration is given to what occurs after the purchase. Moreover, generally the only effect discussed in the satisfaction and dissonance literature is the change that occurs in attitudes/beliefs and affective reactions of the consumer toward the product or its attributes. Notably conspicuous by their absence in the models in Appendix A are any feedback loops to lifestyle, social class/status or cultural values. Potential feedback loops are shown by cross- hatched arrows. 2. Dyanmics The authors of grand conceptual models would argue that their notion of consumer behavior is dyanmic and that they have modeled dynamism. Upon a closer look, the nature of the dynamism becomes apparent. The Howard-Sheth model is considered dynamic in the sense that it is a learning model; thus, exogenous variables change over time (Wynn 1982, p. 32). Likewise, the EKB model is dynamic in the sense that buyers learn over time of changes in environmental variables and incorporate these changes into future decision making (ibid.). In addition, EKB is modeled around the dynamics of the decision making process. Even so, Bettman (1979, p. 345) believes the EKB model is static with respect to the information processing aspects of decision making. Leaving aside any pejorative comments about static versus dynamic models (according to Machlup [1963, p. 33] dynamics equals “my own theory,” statics equals "the theory of my opponent"), both grand conceptual models, lose their dynamism in application. That is, for both models, since exogenous variables are largely taken as given for the type of consumer decisions under study, then change is not relevant to the research. At the extreme, one moves from one 36 given state of the environment to another given state of the environment. The dynamics are captured, as it were, in a series of "snapshots“ rather than a motion picture. Rather than modeling and understanding the process of change, the level of analysis is reduced to what Frankenberg (1967, p. 83) would call exogenous comparative statics. As a result, it becomes difficult for scholars to argue the case that exogenous variables are relevant to consumer theory. An even more difficult problem for dynamics, however, stems from the fact that consumer behavior models are for the most part brand models (see previous section). Since these models generally deal with one brand and one purchase decision by one consumer, not only are they static but also they are abstrac- tions. There is no interaction between products or product classes, and no interaction between decisions made for different products. It is fair to say that all modeling requires abstraction from reality. However, from a consumer theory perspective, consumer decision making is now focused on a narrow range of choice processes. Perhaps, as Howard and Sheth (1967) posit, most of consumer behavior consists of repetitive brand decisions. However, if one of the goals of theory building is to understand consumer behavior, brand-based models unnecessarily abstract from the process. 3. Implications Taken as a whole, the problems outlined -- the lack of feedback and dynamism, along with the brand perspective -- may account for the inability of consumer research to answer questions about macro consumer behavior. Lacking an understanding of how the product fits into the consumer’s life, we are unable to: 37 state with any degree of accuracy the effect of consumption on society; state what the effect of marketing activities has been; form a meaningful response to the critics of marketing and of the high consumption society; formulate a technique for teaching consumer education because nothing happens after purchase; purhasing occurs in a vacuum; decide whether and how to aid people in industrializing, because we are not sure whether consumption is good/bad; and/or develop appropriate marketing strategies that mutually benefit the consumer and the seller over the long run. CHAPTER III: NEW MODEL A. OVERVIEW OF THE PROPOSED MODEL This section attempts to construct a new conceptual model founded upon past research and the suggestions of the many consumer researchers cited in the foregoing chapters. The goal is to provide the foundations of a system that can enmesh the influences of macro phenomena or exogenous variables into a model of consumer behavior. To do so, this section attempts to ease the restrictions imposed by brand research. The crux of the solution is to posit that underlying single occasion consumer behavior is a dynamic, circular, long-term process. Brand decision making is a special case of that more general model. The purchases of consumers are related to one another in nontrivial ways, which do not depend on the typical assumption of a rational consumer. The material that follows attempts to explore and explicate the foundations of the model. This dissertation proposes several changes in the modeling of consumer behavior that may lead to a new perspective on the consumer behavior process. The suggested changes are, first, a shift in focus to those processes that occur ELLE]: purchase; second, an attempt to incorporate assortments into consumer behavior models; and, third, a more detailed look at the information contained in the attributes of products. New insights into consumer behavior can be gained by considering the part of the process that occurs after purchase (beyond brand/product satisfaction, dissatisfaction, or dissonance); in other words, by considering how choice feeds back into the consumer behavior process. As noted earlier, current models of 38 39 consumer behavior assume a flow from culture to values to life-styles to beliefs to attitudes to purchase and its outcomes of satisfaction or dissatisfaction. Moreover, as shown in the literature review, many streams of research look for a simple relationship --demographics to purchase; life-style to purchase; attitude to purchase, and so forth. Marketers are interested in purchase and consumer theorists have tended to seek the reasons for purchase, but, in their search they have asked what factors 119511.! in purchase. By posing the question in that form, the search tends to focus on the antecedents of purchase. In the past, authors have suggested a change in focus from buying behavior to consuming behavior (Alderson 1957; Jacoby 1976; Nicosia and Mayer 1976). Alderson (1957; p. 166), for example, posits that buyer behavior is a derivative of consuming behavior. However, note that "derivative” implies that buying behavior mm [mm consuming behavior. Here, again, one might be led to search for the amededems to purchase (as in many conventional consumer behavior studies), albeit probably different antecedents. Clearly, it is correct to assume that the anticipation of consumption may affect choice or purchase. While it may appear to be a fine distinction, this dissertation posits that choice and purchase should be viewed as jams into a process, 9.2!. merely the end of consumer decision-making efforts. In order to specify a model with choice as the beginning, one must recognize the feedback from choice to other consumer behavior variables. Carman (1978) has suggested such a feedback loop. A generalized grand model of consumer behavior appears in Figure 3.1 (An abbreviated version of Carman 1978, p. 405; for the original figure, see Appendix A). Carman (1978) has proposed that values, subculture and consumption actually form a "closed loop.” Therefore, purchase feeds back to the initiating variables -- values, and so forth. While his insight is significant to the model proposed in the next 40 section, Carman makes no suggestions as to the nature of the feedback. In their call for a sociology of consumption, Nicosia and Mayer (1976) also outline a circular process whereby consumption activities feed back to cultural values and institutions. Family (Belief s» Attitudes-Intentions) Society Values Life-style :7 Purchase Culture Genes [l IL———~_———~——-——o——————————o L....__...._.._. _______ ..._...__. _____ .4 D-.— ‘--~ Figure 3.1 Generalized Grand Model of Consumer Behavior Much of [the consumer behavior literature is concerned with the nature, directionality, and strength of the various arrows in the upper half of the process depicted in Figure 3.1. The multiattribute attitude literature, which is by far the largest research stream in consumer behavior, focuses on the last half of the top part of the model. The model to be developed herein will emphasize the the lower half of the consumer process in Figure 3.1, enclosed within the dashed lines). 41 Broadly, the outline of the proposed model of consumer behavior might appear as shown in Figure 3.2. Note that it is a reversal of Figure 3.1.1 Choice/Purchase :p Consumption Set ¢ Consuming Style 41 1 Culture, Values, A. L V fi Life-styles, Attitudes Figure 3.2 Dynamic Model of Consumer Behavior Figure 3.2 shows, in abbreviated form, the circular process posited in this research. However, the model is only a tentative representation because it does not capture the dynamics properly, nor does it address the long-term nature of the process. The remainder of this chapter will develop and explain the details of the model. A list of vocabulary to be used in this study follows. 1It should be noted that this conceptualization is consistent with present research in consumer behavior. Current research is correlational in nature as opposed to causal. That is, consumer behavior variables are correlated with choice, but this research posits that causality runs in the opposite direction. 42 Figure 3.3 LIST OF VOCABULARY Universal Consumption Possibilities Matrix: all potential attributes and all possible combinations of attributes -- discovered and undiscovered-- available to all individuals and households to consume. Consumption Set: the entire assortment or portfolio of complementary and substitute attributes and attribute combinations that a consuming entity holds at a particular point in time. Consumption Subset: a subgrouping of products and attributes in a consumption set which are related because they are used together as a system. Consuming entity: any consuming individual, household, group, or society. Consuming Style: the manner in which a consuming entity furnishes the requirements of consuming behavior and other aspects of life. Operational Def.: the manner in which a consuming entity acquires and consumes a portfolio of attributes (consumption set). Acquisition: the accumulation of an attribute(s). (Acquire: (W)"' to come to have a new or additional characteristic, trait, or ability.) Consumption: the depletion of an attribute(s) (W) the act or process of consuming; the utilization of economic goods in the satisfaction of wants or in the process of production resulting chiefly in their destruction, deterioration, or transformation. Attribute: (W) an inherent characteristic; a word ascribing a quality Concrete Perceived Attribute: intangible characteristic with universally agreed- upon units and/or measurement scales, perceived through one or more of the senses. Subjective Perceived Attribute: intangible characteristic which tend to be charged with feelings and varying from individual to individual. For instance, "homey" to one person may differ dramatically from that of another person, depending upon the individual’s experiences with home. Elementary Physical Attribute: tangible constituent characteristic of more complex attributes or products in reduced form, generally recognized by experts but often not by nonexpert consumers. Physical Component Attributes: tangible characteristic conventionally recognized by both experts and nonexpert consumers. 43 Figure 3.3--Continued. Potential Attribute: physical and perceived attribute that exists in the product but remains undetected either because it exists in insufficient quantity to be recognized or because it is unimportant to consumers for present known use contexts. Good: a marketable combination of attributes ”a bundle of attributes” (W) something that has economic utility or satisfies an economic want; personal property having intrinsic value but usu. excluding money, securities, and negotiable instruments; wares, commodities, merchandise. Consumer Behavior: the activities and processes necessary to develop and meet the requirements of a consuming style. Activities and processes include, but are not limited to, discovery, evaluation, and choice. Wdrds as defined in In; markgting literature: Assortment: ”a collection of two or more types of goods which either complement each other directly or in total possess some degree of potency for meeting future contingencies” (Alderson 1957, pp. 198-99). a collection of assorted things or persons (assorted: suited by nature, character, or design: matched.) Consumption System: "The way a purchaser of a product performs the total task of whatever it is that he or she is trying to accomplish when using the product" (Boyd and Levy 1963, pp. 129-30). Product: ”the need satisfying offering of a firm” (McCarthy and Perreault 1987), ”anything that can be offered to a market for attention, acquisition, use or consumption that might satisfy a want or need” (Kotler 1986, p. 758). 11 win finiin 'fr frm h in in s r quoted: Activity: the quality or state of being active; vigorous or energetic action; a natural or normal function; a process (as digestion) that an organism carries on or participates in by virtue of being alive; similar process actually or potentially involving a mental function; a pursuit in which a person is active. Consumer: one who utilizes economic goods. Inherit: to come into possession of or receive. Produce: to give being, form, or shape to, MAKE esp. manufacture. Barter: Trade: Purchase: Steal: Have: Possess: Collect: Store: Display: Use: Deplete: Consume: Give: Dispose: ’ (W) stands for definitions for ngster’s Ngw gdllggiat; Digtidgary. 44 Figure 3.3--Continued. to trade or exchange one commodity for another. to give one thing in exchange for another. to acquire by means other than descent or inheritance; to obtain by paying money or its equivalent. to take away by force or unjust means. to hold in possession as property; to consist of; to acquire or get possession of: obtain; to exhibit or show. to instate in as owner; to have and hold as property; to enter into and control firmly. to bring together into one body or place; to come together in a band, group, or mass. to lay away: accumulate; to place or leave in a location (as a warehouse, library, or computer memory) for presentation or later use or disposal. to exhibit ostentatiously. to put into action or service: avail oneself of: EMPLOY to consume or take in regularly; to expend or consume by putting to use. to empty of a principal substance; to lessen markedly in quantity, content, power, or value. to do away with completely: destroy; to use up; to eat or drink in great quantity. to grant or bestow by formal action; to put into the possession of another for his use. to place, distribute, or arrange, esp. in an orderly way; to transfer control to another; to get rid of; to deal with conclusively. 1977, Springfield, Mass: G. and C. Merriam Company. 45 B. DEVELOPING A MODEL FOR LONG-RUN CONSUMER BEHAVIOR The long-term nature of the consumer behavior process can be made explicit by considering the individual prior to his/her first choice, and then following the individual through time. Every individual is born into a world filled with products. These surrounding products are labeled a household consumption set.2 The household did not choose those products arbitrarily; rather, they are a subset of the products sanctioned by society -- a societal consumption set. Furthermore, the set of products in a society is itself a subset of all possible products that ever existed -- a universal consumption possibilities set. See Figure 3.4. Universal Consumption Possibilities Set P Societal Consumption Set Household Consumption Set Figure 3.4 Consumption Sets and Subsets Gradually, over time, the consumer learns about the products that compose the 2This section will begin by using the broad term "products," while definitions found in the vocabulary list are in terms of ”attributes.” The difference will be explained shortly. 46 household’s consumption set. Likewise, the consumer becomes aware of some of the potentialities that exist in society’s consumption set. As a result of the household consumption set experienced as children, consumers do not begin to make consumption choices or develop consumption sets in a vacuum. New consumers have pre-existing biases toward certain attributes, goods, and styles of consuming. C. CHOICE Choice is not solely the output of a discovery and evaluation process as depicted by many current consumer behavior theories. Rather, choice might be viewed as an input. Thus, choice is a m £91 an end. If one were able to focus on the individual’s first choice and purchase of a good, the scenario in Figure 3.5 would depict that process. Note that the person begins with an r- - Universal C.P. Matrix r- "1 Societal C.P. Matrix [Household Consumption Set] 1.- 1.- " '1 New Consumer’s 3 Choice Empty Consumption A Set I- .1 Figure 3.5 Choice in the Context of Sets 47 empty consumption set. Familiarity with other sets provides the context for the decision. In most cases, he/she is limited to the consumption possibilities set for the society. Moreover, consumption sets encountered during his/her child- hood act as a reference point for the current purchase. In the scenario depicted in Figure 3.5, culture is shown to have a direct influence on the individual’s purchase decision. Over time, the individual fills in the empty consumption set (see Figure 3.6). Consequently, as depicted in the present model, choice is not merely the Time T_1 Figure 3.6 The Process of Filling a Consumption Set 121112111 of a process but rather an i_nmi_t. Choice is not an e_n_d; choice is a was toward the end of accumulating a workable consumption set. The first product chosen contributes by inaugurating the individual’s consumption set. Each good with which an individual comes into contact has the potential to renew the process of developing a consumption set. Later choices are partially predicated upon the individual’s now-established consumption set. Therefore. the 48 established consumption set now becomes the context from which the consumer makes decisions. The next section provides a discussion of the elements that compose the consumption set. Briefly, choice results in goods which in turn contribute their attributes to the construction of a consumption set and a consuming style. Both terms will be discussed and defined next. D. THE COMPOSITION OF CONSUMPTION SETS The model outlined in the previous section calls for the incorporation of the notion of assortment, defined here as a consumption set. Alderson (1957, pp. 198-99) observes that products are not useful in themselves; utility arises in an assortment of complementary goods. Economists also define utility as the satisfaction a person derives from consuming a market basket. Boyd and Levy (1963, p. 130) refer to the use of multiple products for various tasks as a ”consumption system" (also defined in the vocabulary list). In essence, the implication is that in order to understand the utility of an object for consumer, one must understand that object’s contribution to a person’s assortment. The basic notion of complementarity of products has not gone unrecognized. Economists have long known that calculations of the cross-price elasticity of demand account for the influence of the change in the price of one good on the quantity demanded of a complementary or substitute good. Illustrations often involve obvious complement or substitute goods (such as tea and sugar or lemon, and loose tea and tea bags, respectively). However, economists use calculations of cross-elasticity to define a relationship between goods and do not explore the nature of complementarity (the reason behind it). Marketers also recognize complementarity and substitution of goods, but the grand consumer behavior 49 models overlook, to a great extent, the fact that products are not used in isolation. This would imply that by considering a single purchase, in isolation, as brand research often does, one cannot hope to understand fully the rationale for that purchase and cannot accurately predict purchase outcomes. Other scholars have called for a change in the focus of marketing from brands to assortments (Wind 1977; Nicosia and Mayer 1976), but thus far the change has not been forthcoming. Green, Wind, and Jain (1972) discuss measure- ment of item collections, but they look at a single purchase occasion (a meal) and implicitly assume that each item contributes one characteristic (attribute) to that occasion. Within the framework of the multiattribute models, McAlister (1979; 1982) developed a technique for looking at assortments of substitute brands in an inventory (different brands of soft drinks), but she did not extend her model to assortments of apparently disparate items (that is, an assortment). The challenge, then, is to develop a framework for incorporating consumption sets into consumer behavior research. Insight into the nature of assortments can be found in the good x attribute matrix of the economy described by Lancaster (1966a and b).3 The present study is founded upon such a good x attribute matrix or consumption set. Douglas and Isherwood (1978), two anthropologists, believe that "goods assembled together in ownership make physical, visible statements” about their owners; ”the problem is to crack the code" (1978, p. 5). Referring back to Cassady’s (1940) statement, perhaps much information about a person’s life experiences, that is relevant to the study of consumer behavior is contained in the information locked in the goods in a household consumption set. The discussion that follows, provides details of the composition of consumption sets and the relation between components. 3Alderson (1957) has a discussion which would be construed as describing a good x attribute matrix. 50 l. Goods Thus far, the discussion has been couched in terms of products or goods. By convention in marketing, goods are defined as ”bundles of attributes.” (for example, Kotler 1988, p. 187) As such, people do not purchase products per se; rather, they purchase the need-satisfying attributes. Lancaster believes that preference maps do not rank products but the attributes of products; they rank products indirectly (1966a, p. 153). Furthermore, according to Lancaster, goods produce attributes. For the purposes here, a good can be decomposed into or built out of a number of attributes. 2. Attributes Attributes can be thought of as descriptive characteristics sought by consuming entities. The terms feature, aspect, property, and element are alternatives for the word 3113111111.:- Attributes are usually incomplete in themselves, from a consumer behavior perspective. This does not imply that attributes are fragments. They are more properly thought of as building blocks or ingredients.‘ An attempt is made here to get away from the definition of attributes used in the multiattribute literature, where they are equated with needs and outcomes. Here, an attempt is made to outline a model of attributes that separates the actual attributes of the product from the meanings people inject into the product. Lancaster (19663), in his new theory of consumer behavior, has ‘The term attribute is used for conceptual and modeling convenience. Since many marketers believe that consumers buy bundles of attributes, the concept has face validity. As used in the multiattribute models, attributes appear to have construct validity; i.e., discriminant and predictive validity. The validity of attributes as a conceptual tool is l'_1_0_[_ founded on their predictive validity. That is, it is not necessary for consumers to actually use attributes in consuming decisions for attributes to be useful in a theoretical sense. Nevertheless, it is hoped that eventually they will aid in efforts at prediction. 51 suggested that there is information contained in goods which is separate from the preferences of individuals.‘ What is the nature of an attribute? Are attributes objective or subjective characteristics? For the purposes of generality, one must include both objective and subjective attributes. Wilkie and Pessemier (1973) point out that Sheth (1969), Hansen (1969), and Pessemier (1972a and b)agree that attributes must reflect perceptual dimensions, while Heeler et a1. (1973) use objective product attributes. Lilien and Kotler (1983, p. 357) echo the same sentiment: ”physical features of the product as well as its psychological features and promises. . .are believed to influence the consumer-choice process at all levels.” Attribute lists are often product specific. Consider the list of attributes from an automobile study. Mazis, Ahtola, and Kippel (1975, p. 42) report that the list included ”safety, gas mileage, styling, repair record, acceleration, quality, of workmanship, and price." Attributes such as safety, quality and styling are, for the most part, subjective (perceptual). Gas mileage is more closely related to an objective dimension. Notice that perceptual dimensions such as safety result from even more elementary perceived dimensions, such as solidity, weight, and size. Other research includes similar lists of attributes: for instance, taste, carbonation, and sweetness for cola studies; color, price, flavor, and effectiveness for toothpaste; flavor, tar and nicotine content, and ease of draw for cigarettes (see Mazis et a1. [1975, p. 42] for the latter two sets). In the examples mentioned, the lists include objective-physical attributes and subjective- perccptual attributes. However, it is suggested here that attributes be viewed in a more global sense. That is, the material that follows discusses attributes without reference to any particular product. 5However, Lancaster’s model is specified only in terms of objective attributes. 52 For purposes of this study, the two main dimensions of attributes in a consumption sets are (l) objective and (2) perceptual. These dimensions appear in Figure 3.7. The objective dimension deals with tangible attributes, and the perceptual dimension deals with intangible attributes. Objective Attributes Perceptual Attributes Figure 3.7 Attribute Dimensions Both dimensions can be further broken down into more elementary dimensions. Appendix A gives an example listing of components of the various dimensions; the lists are not exhaustive.6 Figure 3.8 provides an abbreviated example of a consumption set. The attributes listed there on first level of each dimension -- elementary physical attributes and elementary concrete perceived attributes -- generally have universally stable definitions. Consider, for example, the perceived dimension 6The listings provided are meant to be suggestive. Even these lists could be further reduced, e.g., chemicals are composed of atoms, and reduced again to sub-atomic particles. An exhaustive and mutually exclusive list would require scientific training. And since most consumers do not use these dimensions, they are not necessary for purposes of this research. Consumption Subset Living Room Coods Floor coverings Furniture Ta bles Chairs Shelves Electronics Furnishings Books Statues Paintings Physical Elementary Component Physical Attributes~ Attributes Wood Chemicals Stone Elements Ceramic Compounds Brass Minerals Glass Fabrics Figure 3.8 Consumption Set ainixal suoguodord Sugxnpng qloouig sofiuv auaias [entrain] suuog P103 alqeimeog 355321;) unpow saun sainqglnv sainqginv $31618 ajarauog pasgaaiad newsman paspsnd unsofqng 54 "forms.” Cubes are widely recognized and distinguished from cones. The term cube has a generally accepted definition. The same holds true for the second level of the physical dimension. For example, stone is a term whose meaning remains the same from context to context. Usually there is a process of combination of first level attributes necessary to form attributes at the second level. The second level of the perceptual dimension, subjective perceived attributes, differs from the other dimension, just described, in that attribute definitions are generally colored by each individual’s feelings. For example, comfortable is both context-specific and person-specific in meaning. The range of variability on the definition of comfort is not as wide as for some subjective perceived attributes (”homey"). By specifying attributes in this manner, it is hoped that the model will lead to a categorization of attributes that is both unidimensional in interpretation and general enough to encompass more than just brand choices. Wilkie and Pes- semier (1973, p. 429) note that few marketing articles explicitly deal with the desired characteristics of attributes (p. 429). They state: "Basic criteria for specification of attribute lists require that they be exhaustive, semantically meaningful, subject to unidimensional interpretation, and reflect possible variations in choice or use contexts” (1973, p. 429). The first three criteria they specify are laudible from a measurement perspective. The last criterion points to the particular nature of attribute specification under brand choice models. By specifying the domain of choice/use contexts, the researcher only has the capability to deal with known or given attributes of an existing brand or product -- not all ”potential attributes.” As a point of clarification, it is important to note that the specified dimen- sions do not correspond to the cognitive, affective, and behavioral dimensions of 55 attitude. This model is meant to be neutral in the sense of how one learns or ”knows" the dimensions. Any attribute can be ”known" in either a cognitive or affective sense. 3. Consumption Sets A consumption set can be formally defined as the entire assortment or portfolio of complementary and substitute attributes and attribute combinations that a consuming entity holds at a particular point in time. Note that a consumption set can be decomposed into consumption subsets or subgroupings of goods and attributes used together as a system, such as a pantry, a wardrobe, a laundry, and so forth. Figure 3.8 is an example of the terms relating to consumption sets for the example of the consumption subset of living room furnishings. Consumption sets are purposeful, in the sense that, the attributes and goods are used by consumers. However, consumption sets can contain purely aesthetic goods and attributes. 4. Interrelationships and Points of Clarification The attribute list has been specified as two dimensional because it would be incorrect to depict goods as simply a combination of several objective attributes and perceptual attributes. There are relationships between perceived and objective attributes as well as levels of attributes. Consider several alternatives. Combinations of elementary physical attributes can lead to certain perceived attributes. Chemicals may produce patterns or forms -- quartz crystal. Feller (1988), Artiat’s Eigmgnts; A Handbddk df Thgir Hi§tdry and Qharagtgriatigs, speaks about the synthetic and natural organic chemicals used in developing pigments found in paint. He gives the example of the yellow found in Persian miniature paintings of the 16th to 20th century. It was made from the urine of 56 cows fed only mango leaves, that is, it consists mainly of clacium and magnesium salts from the anthraquinone of the leaves. Also, an action applied to a physical attribute can produce a perceived attribute (for example, combining certain chemicals leads to combustion, which produces temperature and the feeling of warmth). Furthermore, consider the example provided in Figure 3.9. This is clearly an advertisement for the firm involved, but, it provides pertinent information. Here, one can see that a number of activities, both natural and human are used to produce the good (vanilla). These activities are performed at the manufacturing and wholesaling. The resulting flavoring is composed of what experts know as 26 essences. The result for the ultimate consumer would be the physical consumer good --vanilla-- and the subjective perceived attribute-- vanilla fragrance/taste. Moreover, note that by chemically extracting liquid from wood pulp, one can produce vanilla fragrance and taste. The discussion could be taken one step farther. A marketable product for consumers who prize "natural" products (natural is a subjective perceived attribute) would be the beans alone, not those ”carefully selected” and processed into liquid by Nielsen- Massey. 57 ”Grown mainly in Madagascar and Indonesia, the vanilla bean or pod comes from the only orchid to bear a fruit. Each flower, open only one day a year, must be hand polinated to produce a bean. ”Pure Vanilla is more expensive than imitation vanillas because the bean requires a labor-intensive 3-6 month curing process to develop full flavor. About 5 pounds of uncured beans produce 1 pound of cured beans. ”No one has ever been able to duplicate the flavor of Pure Vanilla, produced by more than 26 natural flavoring essences. The main ingredient of imitation vanillas is vanillin, an inexpensive, artificial by-product of the paper industry, chemically extracted from wood pulp. Vanillin only partially simulates the complex flavor of Pure Vanilla. "Our exclusive cold process method extracts the essence from selected choice vanilla beans, the finest obtainable, producing a rich and delicious flavor that is unsurpassed." Nielsen-Massey Vanillas, Lake Forest, 11. Figure 3.9 Vanilla Advertising Showing Attribute Combination At a broader level, examples of the interrelationships among the elementary components of a consumption set abound. A good may be considered equivalent to a perceived attribute if the good is considered synonymous with the attribute (for example, sugar and sweet). Several attributes together can sum up to one good. Also, several goods combined can produce another good, or alternatively they can combine to produce a single attribute. Lancaster (1966a, p. 134) states that "Goods in combination may possess characteristics different from those pertaining to goods separately.” Tangible and intangible attributes in a consumption set can W111 one another. Goods and their attributes can act as Was for one another. It should be emphasized that widely varying alternative goods can act as substitutes for one another if they contain substitute attributes. Holbrook and Moore (1981) 58 has done research, using the Lens model, to decompose attributes of a product (a musical score) and determine whether certain combinations of attributes are better complements than other combinations. E. DYNAMICS WITHIN A CONSUMPTION SET The consumption set is dynamic, in the sense that it changes over time and it continues to change through time.7 Part of the dynamics of a consumption set result from the dynamics within the goods and attributes. Part of the dynamics result from the collection and use of the set. Goods have a dynamic component. They have the quality of being ”stores” of attributes or reservoirs for attributes. McAlister and Pessemier (1982) speak of an ”inventory retention factor” of products.8 As stores of attributes, goods have certain characteristics. These include, but are not limited to, the following: (1) the retention rate of an attribute, (2) the rate of emission of an attribute, (3) the perishability of an attribute. A continuum based on the speed of perishability might be constructed. Mukerji (1978, p. 352) notes that ”most commodities are expected to depreciate 7There is a significant difference between the latter two possiblities. Changing over time would imply that the changes can be captured in measurements taken at different points in time. This perspective would suggest measurements that might be labeled "comparative statics” (Frankenberg, 1967). Continuous changes through time would suggest the need to measure continuously over a period of time, in order to capture variation. This is what Frankenberg refers to as dynamics. 8Although they use the term with respect to an individual and situation, do; with respect to the product itself. 59 in value over time."9 In a consumer context, some clear examples are the loss of carbonation from an open bottle of soft drink; salt water losing its salinity through settling; and dresses and suits becoming unfashionable. An example of a product perishability continuum (in order from more rapid to less rapid depletion) are shown in Figure 3.10. Fads Fashions Durables Antiques Buildings Art/ masterpieces Rapid Slow Figure 3.10 Perishability Continuum Such a continuum might suggest a new classification of goods.10 °Although the point the author is trying to make has to do with the fact that some commodities do not depreciate but in fact appreciate. loFurther, one might conceive of a new definition of risk (in a consuming sense). Risk might be thought of as the perceived perishability of an attribute. 60 F. CONSUMING STYLE -- FURTHER DYNAMICS IN CONSUMPTION SETS The notion of consuming brings a level of dynamism to the conceptual framework specified in this chapter.11 People consume portfolios of attributes. Consuming is predicated on acquiring those attributes. Cgasdming fly]; is the manner in which a consuming entity furnishes the requirements for consuming behavior and other aspects of life. The term can be operationally defined as the manner in which an individual or household acquires and consumes a portfolio of attributes (consumption set). Since consumption sets indicate a consuming entity’s pattern of consuming at one point in time, the consumption set can provide an indication of the tangible aspect of consuming style. However, a static conceptualization can be inadequate and misleading because consuming style should be thought of as a process. In a dynamic sense, then, one might define consuming style as the alteration or dynamic changes that occur in an accumulation of attributes over time. It is apparent that consuming style has two closely related components-- acquisition and consumption. Aggaisitidn is the accumulation of an attribute(s). and W is the depletion of an attribute(s). Both acquisition and consumption are transformations of a consumption set. In reality, these transformations are an on-going process, but for purposes of clarification they can be separated. 1. Acquisition Consider the various aspects of acquisition. A form of provisioning behavior, acquisition includes all activities required of individuals to provide for 11"Consuming" is not specific with respect to whga the consuming ends. That is, consuming can be construed to imply an ongoing process. 61 themselves and/or their family. It is the process of building a consumption set, and it would include all transformations of a consumption set to make the set ready to be consumed. Acquisition can be accomplished by different means, including: inheriting, producing, bartering for, trading for, purchasing (buying), or stealing. Each of these are techniques by which a consumer could initiate or replenish a consumption set. One may simply be born into an accumulation of attributes. Alternatively, in modern society one might purchase an attribute. One might barter or trade certain attributes for others. If none of these methods are available, or if a person is so-inclined, he/she could produce the needed attribute. Finally, stealing it is an option. 2. Consumption Consumption is "the act or process of consuming, the utilization of economic goods in the satisfaction of wants or in the process of production resulting chiefly in their destruction, deterioration, or transformation” (Webster). As in acquisition, one finds a transformation of the consumption set. Attributes can be acquired for (1) instantaneous use, (2) use at some time in the near future, or (3) use over an extended time period.12 Consumption can be accom- plished by a number of means: having, possessing, storing, collecting, displaying, using, depleting, consuming, giving or disposing. It should be emphasized that all of the aforementioned methods of consumption have several common characteristics -- the amount, the rate of speed, and the timing. All are ways of using up, which is how the dictionary 12Belk (1982) has called for a redefinition of consuming behavior as "having behavior." He says, ”consumption is only one of several reasons for acquiring an item. Possessing and collecting are two prominent alternate goals” (1982). 62 defines the verb ”to consume.” Consider, for instance, the methods -- collecting, using and consuming. ‘Halidlg and 995mm indicate the existence of an attribute; the latter may indicate the legal status of an attribute. Neither term indicates when, how, or at what speed the attribute will be consumed. ’Qoflggfing indicates a rather slow consumption. Nevertheless, a time element is clearly present. Collecting takes place over time. The term also indicates the potential for an addition of other attributes that have some commonality or connection with the first. ‘Stdrjng and displaying also indicate relatively slow consumption of an attribute. They imply intention -- a store of something has the potential for future consumption, and displays are meant as signs. A display may be short- lived; for example, a lavish party. ‘yaiag, dgplatiag, and gdnsdming are all relatively specific with respect to the time element. Depletion must occur over time, as must use. Consuming may be almost instantaneous, but, most consumption requires time. ‘gijviag or Mag most likely occurs at moment. It indicates intention in that the individual does not intend to have any more to do with the attribute. One exception is an explicit or implicit condition imposed on a gift. The various of methods of acquiring and consuming all have in common an indication of the individual’s disposition toward the consumption set. They indicate: (1) the time element (point in time or length of time); (2) the type of activity the individual will use, apply, or favor; (3) the individual’s intended action; and (4) the relations a person perceives among goods/ attributes. 63 3. Limits on Consuming Style Consuming style depends on the availability and relative scarcity or abundance of attributes. The person’s ability to acquire raw materials (attributes) or goods can influence consuming style, and this could be a function of disposable personal financial resources. Consuming style also depends on (is limited or enhanced by) an individual’s ability to use raw materials and goods to produce attributes and on the activities that he/she is willing/able to perform to acquire attributes. Futhermore, consuming style is modified by social conventions and norms as dictated by classes, status groups, families, and reference groups. In addition, the dictates of "good taste” as defined by taste- makers and status groups can influence consuming style. However, the aforementioned modifications are likely to depend upon a person’s preference trade-of f between individuality and conformity. 4. Discussion Note that as part of a consuming style an individual may exhibit a pre- ference for one or several of the various methods of acquisition and consumption. Alternatively, an individual may be mandated by circumstances to use one or several of the methods. Examples abound. Consider the situation where an abundance of products exist from which people may choose. Individuals may favor purchasing as a method of acquiring the needed or desired attributes and may express a distaste for bartering. Moreover, they may consume more rapidly, than in other situations, because of the abundance and ease of acquisition. A contrasting scenario can be developed for a situation where there is an absence of finished goods. Individuals may be forced by circumstances to produce attributes or combinations of attributes that they desire. Recognizing 64 the amount of effort required to acquire the attribute and produce goods, the individual may consume more slowly and have a tendency to collect, hoard, and repair attributes. Inheritance may become a favored method of acquiring. One can continue to speculate on consuming styles. The process of consuming style has a number of important features. The first aspects are the 121.1151. and amddnt of attributes acquired and consumed. Another dimension is the rate of speed of acquisition and consumption. Included here is the manner in which an individual uses up attributes, but this does not preclude the manner in which an individual perceives attributes to lose their value. Third, there are also a variety of relationships among attributes which (a) actually exist, (b) are perceived to exist, (c) can be created, or (d) potentially exist (potential attributes). Fourth, consuming style would include the manner in which goods are perceived as being the embodiment of an attribute. Fifth, also included would be an individual’s focus or emphasis on a certain portion of the consumption possibilities matrix. For instance, a relative focus on certain attributes may be viewed as either materialistic or idealistic; a focus on good to the exclusion of ideas might be labeled materialistic. Sixth, individuals may differ in their like or dislike of the process of developing a consuming style. Alternatively, persons may differ in their conscious awareness of developing a consuming style. For example, utopians are conscientious about developing their notion of an ”ideal” consuming style. Elaborating on the last feature, the model of consuming style does not depend on the individual’s recognition of his/her consuming style. In other words, individuals may not recognize that they have or are developing a consuming style. Indeed, they may be appalled at the suggestion. Stated plainly, 65 ”I don’t care about products” may be a consuming style. Likewise, "I am an aesthetic" may be a consuming style, or even "I am an ascetic.""’ Moreover, this model of consuming style does not depend on a conscious goal orientation in the acquisition and/or consumption of attributes. That is, it does not depend on what economists might call a rational consumer. Alderson (1957) implies that consumers purchase goods that maximize the number of potential future contingencies they can handle given an inventory of products. As a result of the line of reasoning given in the present research, one does not need to make the rather strict assumption that consumers maximizg the number of contingencies. It is constructive to think that products are often purchased that will be used for a number of purposes. If one good, with its contingent attributes, will be used for numerous purposes, then a consumer may purchase a good with an attribute that can be relied on to serve those purposes. Bandura (1977) believes that people "create and activate environments.” According to Clyde Kluckhohn (1951, p. 406), "acts, as has been said, are always compromises among motives, means, situations, and values.” Given the variety of attributes and the infinite number of combinations, given a multiplicity of uses and a multiplicity of situations, there is likely to be a great deal of ”ad Lila" in consuming style. In a consumer behavior sense, people may not have, or may not be able to articulate, a reason for acquiring an attribute (for example, purchasing a product) because they simply plan to store it for consumption at some future date. This concept provides a significant departure from present lines of consumer research. 13Further research is needed to identify style groups. 66 G. DO PEOPLE CREATE MEANINGFUL CONSUMPTION SETS? Do people have consumption sets, and do people create consumption sets? The answer to the first question is clearly yes. Almost any individual, regardless of age or culture, has a set of attributes. Howard and Sheth (1967) are correct to assume that 90 percent of consumer decisions are repetitive brand choices. But consider what this implies. Those repeat purchases imply that, for the period of the repetition, provides the person considers the product to be a relatively permanent part of the consumption set necessary to facilitate living. In other words, these individuals have empty spaces in their consumption set labeled for those attributes. Since the brand of product purchased has an emission rate for its attributes, the space must be refilled on a repetitive basis. An answer to whether people create sets can be framed as follows. Organizational consumers of all types create inventories. Manufacturers create stockpiles of raw materials, semifinished goods, and supplies. One rationale for this is to avoid stockouts or production down time. These organizations also produce in advance of demand, so that they create inventories of finished goods. To the extent they produce complementary products, manufacturers also create assortments. Retail organizations create both inventories and assortments. The retailer’s assortments are based on complementary goods which consumers use together or like to purchase together. An article by Burton and Dorner (1941) describes ”The Wishmaker’s House." For the 1941 season, retailers were introducing this comprehensive plan for merchandising which included matching colors and styles for all parts of home furnishings to replace the "haphazard buying and selling of home f urnishings" (p. 41). 67 But does the notion of creating a "market basket" (consumption set) have validity for the ultimate consumer? Theories such as the family life cycle implicitly assume that people build inventories of goods (mainly durables) over their lifetime. David Riesman (1950) referred to the manner in which the standard package of consumer goods and services has changed over time. Fashion theory suggests that people create meaningful assortments. That is, people develop wardrobes of clothing necessary for the activities in which they are involved. Within each subgroup of clothing, people also create assortments of goods that complement one another. Moreover, popular experts on dress such as Molloy (1976), base their advice on certain asumptions, not only about an item of clothing, but also about the set of goods worn together (outfit) and the attributes of those goods and sets. Lauer and Lauer (1979) also note that the popular media have a long tradition of suggesting appropriate apparel for various occasions (such as, clothing for working women) and of drawing inferences about character based on a person’s apparel. They include advice from didy’s Ladigs dek in the mid-18003. Anthropologists and archaeologists also make assumptions and posit hypotheses based on people’s collections or assortments of goods. In Warld 91f Q0911: (1978), Douglas and Isherwood discuss, at length, assortment building. A- rchaeologists draw inferences about past populations from goods and collections of goods found at dig sites (Stewart-Abernathy 1986, p.4). A number of situations can be suggested where consuming entities build consumption sets. If the consuming entity is an individual, expectant parents, brides-to-be, newly employed persons, new sports enthusiasts all build sets. If the consuming entity is a society, strategic stockpiles, strategic arsenals, libraries, and museums all are examples of the building of consumption sets. 68 H. CONSUMPTION SETS AND CONSUMING STYLE ARE USEFUL TO CONSUMERS Consumption sets and consuming style produce utility for consumers because, in the process of developing a style, consumers take a meaningless variety of isolated goods and turn them into a meaningful consumption set. The proposed model suggests that, consumer behavior is ultimately purposeful to the extent that through the process of consuming style people develop workable consumption sets. Thus, an attempt is not made here to explain why people consume, only why people act like consumers. Moreover, consuming need not be considered an end in itself; clearly it is only one possible end. Consuming may facilitate other goals, including learning, creating, resting/sleeping, exploring, nurturing, educating, and worshiping. If I want to be an artist, I must also consume «paint, brushes, canvas, clay and other materials are necessary. It should be emphasized that discussion here does not revolve around the single goal of consuming. In this vein, an attempt is not made here to explain why people consume, only why people act like consumers.“ In the present model consumer behavior would be viewed, in the main, as instrumental to the development of a consumption set and a consuming style, which in turn enable the consumer to 012019;. desirable outcomes. This production theme has found its way into the works of various authors. Lancaster (1966; 1971) uses a production analogy: goods produce characteristics l‘Certain authors adopt a very specific stand as to what consumers are trying to accomplish. Alderson (1957) speaks of Freud’s ”drive for perpetual bliss," (p. 170) and "clearly visualized life plans." He believes that consumer behavior is the study of ”hedonomics" (p. 288). Holbrook (1987) has also referred to consumer behavior as hedonic. 69 (attributes). Pollay (1986) and First (1986)“5 both alluded to consumption as a production activity. According to the former, consumption produces "self, ideal self, identity, relationships and roles.“ This view is consistent with the anthropological view of Douglas and Isherwood (1978), who discuss the use of goods to produce social identities. Rainwater (1974) says that consumption is important as a signal of membership in society. Stewart-Abernathy (1986) presents an archaeological/anthropological view in the reporting of a dig site in the Ozark Mountains. His implied; hypothesis is that the consumption choices "reflected directly in the archeological record of the Moser occupation, should be patterned to support a shared version of social reality” (1986, p. 4). Returning to the statement made at the beginning of this chapter, if choice is an jam into a creative process, then what is the output? Any answer to this question by consumer researchers would include: 1. creating a life-style (perhaps harmony, order/disorder), 2. creating a personality/self, 3. playing role(s), 4. surviving (in a physiological sense), 5. being an individual, 6. belonging to a group, 7. expressing oneself, 8. signaling membership, 9. entertaining oneself, and so forth. l‘5Richard Pollay and Fuat Firat, unpublished lecture material, presented at Association for Consumer Research Meeting, Toronto, Ontario, Canada, October 1986. 70 Ultimately, in creating a consumption set and developing a consuming style, the consumer creates possibilities to deal with multiple demands.“ 16Some people are good at it, others are not. CHAPTER IV: IMPLICATIONS OF THE CONSUMPTION SET MODEL FOR THE CONSUMER BEHAVIOR LITERATURE A. CONSUMER BEHAVIOR As defined above, consuming style and consumption sets have implications for consumer behavior. Consumer behavior according to most textbooks, usually includes ”actions and decision processes...involved in discovering, evaluating, acquiring, consuming, and disposing of products and services” (Harrell 1986; also see Wilkie 1986; Berkman and Gilson 1986). If, as defined here, the domain of "consuming style" encompasses acquisition, consumption, and disposal, then one can redefine the term. ansdmgr bghavidr, as used here, is "the activities and processes necessary to develop and meet the requirements of a consuming style." Using this definition, ”activities and processes" would include, although not be limited to, discovery, evaluation, and choice. The difference in definitions is not merely semantic. The new version distinguishes two separate, but interrelated domains -- consumer behavior and consuming style. Thus, consumer behavior will be partially derivative of a person’s pre-existing consuming style. More important, the behaviors undertaken by consumers (such as, choice) will act as an input to consuming style. The latter points can be expounded upon in the following way. A pre- existing style will limit (or expand upon) "discovery.” That is, if I have developed a style or a consumption set to which I am committed, I may not even perceive the attributes in a new good. Therefore, I will not ”discover” them. Evaluation may occur in terms of the good’s "fit" into the present mix. In other 71 72 words, above and beyond an evaluation of a brand’s attributes, I may be evaluating the brand’s fit with a present mix. Alternatively, I might be evaluating the brand’s creative potential. That is, I may value the good for its potential in completing a set of attributes, not simply for its possession or lack of possession of an attribute. B. MU LTIATTRIBUTE MODELS Goods themselves can be used as raw materials in the production of other goods or attributes that the consumer wants. This facet provides an important extension of the typical multiattribute approach. A good possesses an attribute, but it is not readily apparent (visible) or is not apparent to everyone. Alternatively, the attribute might not exist in sufficient quantity in a particular good to be considered important or determinant. It is possible that the attribute might become Mali—zed when one good is combined with another. Or the attribute may become significant as part of a set of goods. For example, a navy blue jacket, by itself, does not produce enough of an "Ivy League look” without the grey flannel slacks and the white oxford shirt. Or, consider a Jeep Blazer truck; ownership of the product without certain other products to accompany it could be considered "gauche." Under those cirumstances, according to a typical multiattribute approach, the consumer may say that he/she does not like the attributes of the Jeep, does not consider them important, or that the consumer does not believe the Jeep possesses any desirable attributes. 73 C. LEARNING In the present consumer behavior literature, learning is merely acknowledged. Howard and Sheth’s model is based in Hullian learning theory, and learning occurs with the system (consumer) under study. However, models and texts focus much of the learning discussion on the learning of brand attributes. This treatment has not been satisfactory according to some critics (Nord and Peter 1980). Under the present model, consumers would be viewed as learning a number of varied consumer-related phenomena. To begin with, they would learn what is an acceptable consumption set within their family. They would then learn the boundaries of society’s consumption set. They would learn a consuming style. Over time they would refine and redefine their style (one example would be learning to become parsimonious). Furthermore, they would learn to become efficient at learning and developing their style (that is, learn more quickly or at a lower cost). There is some support for this process occurring at the brand level. Woodson, Childers, and Winn (1976) showed that 65 percent of men in their 203 held the same auto insurance as their father did; the number drops to 55 percent of men in their 303 and to 25 percent over age 50. Consumers would also learn skills at consumer and buyer behavior. That is, they would become adept at discovering, evaluating, and choosing products. Brand loyalty or inertia, on the one hand, and variety seeking or stochastic choice, on the other, may be part of learning at this level. In addition, consumers would learn to shop and to buy. Thus, under the present model, consumers would learn a variety of consumer behaviors and phenomena. 74 D. BUYER BEHAVIOR In the present model, M Maxim; is one form of acquisition behavior. Buyer behavior deals with the purchase of attributes or goods. Some consideration must be given to the fact that buyer behavior is only and possible means of acquisition. Moreover, buyer behavior may be a phenomenon that occurs in varied amounts in different societies. Thus, consuming style, as defined, provides an avenue for the study of the behavior of different cultures and countries, at different stages of economic development. In either case, people composing the culture may be willing and able to exhibit one or another acquisition and/or consumption method. E. STOCHASTIC BRAND CHOICE Based on the present model, new light may be shed on brand choice. Part of research in that area has been modeled as stochastic. McAlister and Pessemier (1984) refer to various attempts to explain the nature of the randomness. Brand choice may be related to that portion of consuming style which is so well defined that it has become habit, loyalty, or variety-seeking. An alternative explanation is that, in part, the reasons for purchase lie outside the small brand portion of the consumption set. The consumer sees additional possibilities for application to a different problem. Considering the size of the consumption possibilities set, in some cases the power to detect the reason for choice would be low. 75 F. ENCOMPASSING DEVIANT BEHAVIOR By defining consuming style in the manner described above, one can also encompass so-called deviant behaviors such as collecting unusual items (see, for example Belk (1982) and ACR Special Session on Fanatic Consumer Behavior, 1987). Also, pathological behavior, such as stealing, is encompassed within the boundaries of the model. G. INFLUENCE OF CULTURE The universal consumption possibilities set provides a comprehensive list of the types of attributes and goods available to the consuming individual. Using the present model, culture would act either to limit or enhance the selection possibilities in the consumption possibilities matrix. The location and society in which one lives will influence a person’s knowledge of the goods and attributes. Consuming is defined (limited or enhanced) by physical, physiological, psychological, and geographical barriers and resources. For example, a range of mountains may inhibit a hunter-gatherer society from tracking prey. The societal matrix of possibilities would have a geographic limit. If a researcher asked a person w_ha_t he/she would consume, he/she would define the set of possibilities within those limits. In other words, a knowledgeable person working to consume in that situation would be unaware of the possibilities over the mountain. (See Figure 4.1.) 76 A1 A2 A3 A4 A5 Figure 4.1 Consumption Possibilities Set When Barrier Exists Beyond the physical barriers, however, the consuming individual is limited (or encouraged) by religious, cultural, and social mores and conventions. Judeo- Christian religion defines gluttony as a sin, which implies a restriction on the amount of an attribute consumed. The US. social system allows one to consume as long as consumption does not infringe on the rights of others. Current legal prohibitions (for example, on liquor or cigarettes) and requirements (helmets for motorcyclists) and other types of sumptuary laws (Hollander 1984) are examples of limitations on consumption possibilities. Even in less formal situations restrictions apply -- chewing gum audibly in the classroom provides a case in point. Cultural inhibitions are plentiful. Various cultures restrict certain types of consumption and the consumption of certain goods and attributes. International marketing discussions regularly note interesting prohibitions in some African cultures, such as a on the consumption of eggs (because of the belief that baldness will result). One does not have to rely on anecdotes of culture to note 77 the effect on the consumption possibilities matrix. Consider the prohibition on the use of marijuana. Recently, researchers have pointed out that the prohibition on the use of the product has limited the recognition that one of its elementary physical attributes, THC, can aid in relieving some disorders. The point is that the culture acts to restrict the choice/purchase or use of the product. In so doing, cultural inhibitions act to inhibit the acquisition of the prohibited good’s attributes. The prohibition on a good also prohibits (1) its constituent attributes, (2) any potential combinations of those attributes with other attributes, and (3) any activities dependent upon that good or its attributes. Also, activities which might have resulted in other potentially useful combinations are precluded. Even in the instance of a cultural aversion, such as that of Texas farmers to tumbleweeds, the result might be to delimit the attribute combination possibilities. A recent article by Fincher (1988) describes the potential that tumbleweeds have for water retention, in the badly parched Texas soil (1988); the researcher points out his difficulty in even raising funds for the research project. Conversely, the culture may act to elicit possibilities that might otherwise go unrecognized. Western cultures, which emphasize and reward progress, may allow the borders of the matrix to remain relatively open. The individuals within the culture search for new products and attributes. Activities that lead to progress are encouraged. Eastern cultures, which encourage cooperation, may also enhance consumption possibilities, although perhaps in different ways. Societal technology modifies all of the previously acknowledged barriers. Returning to the example of mountainous terrain as a prohibition, a society may have or create the technology to surmount the barrier. Activities can modify the possibilities for consuming, in the sense that innovation may add products and/or their constituent attributes. Likewise, each of the above-mentioned 78 factors can serve to limit d; c;pa_nd the matrix of potential consuming alternatives, as shown in Figure 4.2. Individual capabilities contribute to societal technology and interact with the consumption possibilities. A1 A2 A3 A4 A5 A6 Figure 4.2 Consumption Possibilities in the Presence of New Technology In this model, then, culture is no longer an exogenous variable in individual decisions. Culture may act as a boundary for persons making consuming decisions; therefore, culture has immediate relevance. If the person wishes to choose a product beyond cultural sanction, then the person must be willing to be a path-breaker (in diffusion terms, an innovator) or must be oblivious to the social consequences (perhaps because socially isolated or a rugged individualist). This would be related to the person’s preference for individuality versus conformity. With regard to the vast majority of culturally sanctioned goods/attributes, culture may have an even more subtle influence on choice and consumer behavior. For instance, culture may affect the attributes that we recognize in a tangible good. That is, culture may act as a blinder to the possibilities. In western cultures, given the fact that many goods are available 79 in finished form in the marketplace, a person confronted with only the physical component attributes may be unable to construct a useful or desired product. Alternatively, culture may act to build connections between physical and subjective attributes. Therefore, as stated, the model of consuming style moves the study of consumer behavior one step closer to integrating the influence of culture. While the present model lacks specificity as to the exact nature of the influence of culture, it is more specific than the current grand models of consumer behavior. H. CULTURE AND SOCIAL CLASS The influence of social class can be analyzed in a manner similar to the previous discussion of culture. Within a culture there may be certain attributes and products common to all classes. The notion of a common basket of consumer goods in industrial societies is illustrative. There may be certain attributes upon which all classes agree; for example, in the United States these might be functionality and independence; in Japan, perhaps beauty. But the difference between the classes may lie in their emphasis on certain attributes, products, their mix of attributes (consumption set), or the choice of goods used to acquire attributes (consuming style). One is born into a group (family, neighborhood, class, subculture, culture), and the importance one attaches to the acquisition and consumption of certain attributes is partially determined by birth group. Dorothy Lee (1959) speaks of the fact that needs are societally deter- mined. According to Douglas and Isherwood (1978, p. 67), ”consumption is the very arena in which film is fought over and licked into shape.” In Figure 4.3, attributes 2 and 3 are common to all classes. Attribute 4 is common only to members of Class 3. Attribute l is common only to Class 1. 80 Class 1 Class 2 Class 3 A1 A2 A3 A4 A5 Figure 4.3 Class Similarities and Differences in Consumption Sets Hirschman (1986) refers to a similar phenomenon (although she uses the word value) when she describes her experience as a participant-observer in the WASP culture: Suddenly (it seemed), the author grasped that these same core values (e.g., practicality, conservatism, individuality, self-control) were expressed in virtually all aspects of the consumers’ lifestyle -- from clothing preferences to automobiles to leisure activities. Once this value ”code” was comprehended, its manifestation throughout every area of consumption could be discerned and the nature of the subculture became evident. The research of Laumann and House (1970) also tends to support differences in consumption sets between classes. Although they were looking at possession of goods (not attributes), they found differences in the composition of living rooms by socioeconomic status. CHAPTER V: HYPOTHESES AND METHOD A. HYPOTHESES Chapters III and IV sketch the underpinnings of a new model of consumer behavior based upon the notion of consumption sets. The framework developed provides a structure for studying the influence of consumption sets on consumer behavior. The latter was described as a long-term process of acquiring and consuming a set of attributes sanctioned by membership groups, and replacing as well as replenishing the attributes in the set on a periodic basis using a number of methods sanctioned by society. ‘ The process involves learning, comparing, gauging, and fine-tuning the set. " Sets act to facilitate consumption and other aspects of life. Sets act as a context for acquisition and consumption decisions. "‘ Sets of attributes, defined by society, delimit or enhance possible acquisition and consumption. Evidence supporting the model was provided in the form of examples drawn from a variety of literature, including the work of consumer researchers indi- rectly supporting the model. However, it remains to be shown that consumption sets can be readily referenced by consumers. If, in fact, consumption sets exist, then it is possible that they have an effect on consumer behavior. Therefore, evidence of the existence of consumption sets is a crucial foundation for further research on the conceptual model. The purpose of this section is to report on an empirical study of consumption sets. 81 82 If the hypothesized conceptual framework is correct, then one would expect to find that consumers are able to construct a set. Moreover, relying on Hirschman (1986) and Laumann and House (1970), the sets that individuals create should differ by social strata. Also, to the extent that all individuals are members of the same culture, some cultural similarities should be evident among groups (strata). At the same time, differences in the activities for which a set is used should also lead to observed differences in sets. Solomon (1981) discusses products as 'props" for social interactions, and Belk (1975) has included activities in his work on situational influences. Relying on these earlier works, it is expected that the sets developed by individuals within the same stratum will be more similar than the sets observed between one stratum and another. The following hypotheses, which result from the conceptual framework in Chapters IV and V, will be tested. The first hypothesis is stated in its alternative form. Hypothesis 1: The consistency of sets of attributes developed by in- dividuals within occupational groups will be greater than that between one group and another. Second, it is expected that, when faced with the task of creating a set, most individuals will create sets that are similar to those they already have, that is, their actual sets. Since, this result should become stronger as the individual ages, presuming that the individual acquires the actual set over time, younger people will have a greater tendency to create their "ideal set.” Therefore, 83 Hypothesis 2a: The development of an ”actual set" of attributes by an individual will be strongly and positively correlated with that person’s age. The family life cycle concept states as individuals advance in age, get married, and have children and as the occupational status of the husband and wife change, the family will purchase goods (particularly durables) accordingly. One might also expect that as the family life cycle variables change, a subject would be likely to have a more established set and, therefore, would say that he/she created an ”actual set.” Hypothesis 2b: The subjects that develop ”actual set" rather than an "ideal set” can be distinguished from one another based on family life cycle variables -- age, marital status, and age of children. Returning to the model, it is important to state the boundaries of the empirical test. Both consumption sets and consuming style were hypothesized as dynamic processes. The present test is static and, in fact, does act directly ask for a description of the subject’s own (actual) set. Respondents are given a good deal of latitude in terms of the sets they create. B. OVERVIEW OF THE METHOD 1. Instrument A living room set was chosen, as opposed to some other subset of a consumption set (such as a wardrobe), for two reasons. First, past research has 84 dealt with living rooms and provides some a M expectations about the products and attributes that might be included in sets. (Laumann and House 1970; Csikszentmihalyi and Rochberg-Halton 1981). Findings have been noted earlier. Second, living rooms are used for display purposes, more so than other rooms (Laumann and House 1970). Therefore, if, as Douglas and Isherwood (1978) state, people use goods to "signal membership" in groups, then living rooms should provide evidence of status differentiation. All subjects responded to an identical questionnaire. It was divided into five sections. In the first, respondents were instructed to create a living room set. The second section asked questions related to the set the respondent had created. Questions in the third section related to the respondent’s own enter- taining practices. A constant sum scale question regarding the types of ac- tivities for which the respondent’s living room was used was included. Respon- dents were asked to allocate ten points among activities. Section four con- tained demographic questions. Section five asked a set of life-style questions (Wells and Tigert 1971), to be rated on a five-point scale, ranging from "de- finitely disagree" (1) to "definitely agree" (5). The main manipulation instructed subjects to construct a living room set. As it appears in the questionnaire, seven blank boxes and a list of attributes was provided. Subjects were asked to fill in the boxes with any products he/she desired and check off the attributes that they felt corresponded to that living room product. While somewhat ambiguous, most subjects easily solved the task of developing a living room set. The restrictions were that the set must be within his/her present income, he/she must make at least five of the seven products, and there should only be seven attributes per product. The latter two restrictions were placed in order to 85 avoid an excessive number of possible permutations. [See questionnaire in Appendix B.] The attribute list included both objective and subjective attributes (see Table 5.1). To reduce respondent fatigue, subjects were instructed not to feel compelled to construct the product in infinite detail, that is, ”not down to the last string or nail.” The list of subjective living room attributes was developed on the basis of a content scan of home-oriented magazines for descriptive attributes. An attempt was made to draw on magazines covering a broad range, including ngntry Hdmg, Arghitggtural Digest, Battgr Hdmas and Qardgns Ham; I§§dg, Traditidnal Hdmg, Fina Hdmgs, and dethgrn Hdmg. Attributes were chosen to appeal to a broad range of individuals yet have potential to differentiate group memberships (see Table 5.2). Attributes such as ”irreverent," ”opulent," roman- tie," and ”grandure" were thought to have somewhat limited appeal; thus, they were not included in the questionnaire. The questionnaire was pre-tested on 30 individuals to determine whether clusters would emerge and whether these attributes could be used to construct sets. After the pretest, the word "genteel” was dropped from the final list of attributes. 2. Subjects This study hypothesized that intergroup consistency on sets of attributes should be greater than intragroup consistency. The subjects were four groups of individuals chosen on occupation a 11193.1 Given the large number of possible groups across society and the probable range of diversity of sets within strata. it was desirable to limit the potential variability. 2 and 1Members of a pridri groups will be referred to as, ”group 1, group ‘ b6 so forth." Memberships in the groups that result from the analysis wil referred to as "cluster 1, cluster 2...." 86 Table 5.1 Objective and Subjective Attributes Appearing in Questionnaire Qbiagtivg Attribdtgs Subjggtiva Attrihtttfi Wood Simple Stone Futuristic Fabric Cozy Glass Natural Brass Authentic Chrome Understated Leather/Suede Harmony Wicker Classic Charming Dramatic Rustic Tradition Comfort Practical Gracious Distinctive Pretty Impeccable 87 Table 5.2 Abbreviated Original List of Subjective Attributes Gracious Cozy Serenity Simplicity Understated Authenticity Harmony Romantic Stately Charming Distinction Practical Formality Pretty Genteel Friendly Tranquility Nostalgic Futuristic Elegant Irreverent Classic Opulence Relaxed Grandure Rustic Sumptuous Original Subtle Functional Ostentation Family Authority Comfort Natural 88 Occupation was chosen to segment groups because research shows that living room items differ by socioeconomic status (Laumann and House 1970). Occupation (source of income) also plays an important role in most social status scales (of Hollingshead’s two item index -- occupation and education -- the first weighted more heavily). Segmentation by occupational groups reduced the range in socioeconomic status. The first group of subjects consisted of college professors; the random sample of 200 was drawn from a population of 307 individuals teaching in a medium-sized, midwestern city. Questionnaires were mailed to the sample, and 82 usable ones were returned (two returned blank and two with incomplete answers). The second group consisted of doctors practicing in one hospital in a medium sized, southwestern city. Of the 270 physicians on the staf, 110 were randomly selected; 42 questionnaires were returned. Members of the third group consisted of a sample of 100 individuals randomly chosen from 400 blue-collar workers (mostly firefighters) attending a training conference; all were residents of a midwestern state. The last group was a convenience sample of 43 individuals working at three different occupations, two white-collar and one blue-collar. Questionnaire distribution was on the basis of subject willingness. 3. Limitations of the Sample Descriptive statistics for the four a m groups show a broad range of incomes. There was an overrepresentation of Ph.D’s and M.D.’s. Men were also overrepresented, although this was more true for group 2 and 3. While these characteristics of the sample may limit generalizability, the purpose of the study was to determine the existence of differences among groups. The study does not attempt to cover all possible groups in society. These limitations do not affect the tests of the hypotheses. 89 4. Data Analysis Cluster analysis was chosen because of the nature of the problem and its appropriateness for theory construction. Cluster analysis represents a family of empirical techniques for identifying homogeneous subgroups from a heterogeneous sample (Speece et a1. 1985). The individuals that comprise a culture are het- erogeneous with respect to the contruction of living rooms. Across the popula- tion one should witness a great deal of heterogeneity regarding attributes, while at the same time witnessing homogeneity within groups. In other words, while individuals may use different products to fill their living room and those products may differ in outward appearance, individuals within a group should be attempting to make similar expressions with the items in their living room. Hierarchical cluster analysis, an agglomerative method, was used to classify individuals into groups based on the set of attributes chosen. Hierarchical cluster analysis begins by assigning each individual to a given cluster. Then, subjects/clusters are successively added, based on similarity of response, until all individuals are merged into one cluster. Cluster analysis involves problems in determining the appropriate number of clusters (Green, Frank and Robinson 1967; Frank and Green 1968; Inglis and Johnson 1970; and Morrison 1967). In their review article on cluster analysis, Punj and Stewart (1983) note that part of the problem arises from different researchers using different cluster techniques, which limits comparison of results. The authors indicate that cluster analysis suffers from the same weaknesses as multivariate techniques in general (1983, p. 134). They also note that while cluster analysis has rarely been used for theory construction, it is a legitimate facet of the technique. Researchers must make several decisions when using cluster analysis. First, the choice of similarity measure defines the criterion on which clusters are 90 joined. Squared Euclidean distance was chosen for the similarity measure. Punj and Stewart (1983) note that choice of similarity measure does "not appear to be critical.... Choice of correlation coefficient, a Euclidean or city block metric does not seem to produce much difference in the final outcome." The second decision concerns the clustering method. Ward’s minimum variance method was used. ”Ward’s minimum variance method, average linkage, and several variants of the iterative partitioning methods appear to outperform all other methods," according to Punj and Stewart (1983). They add that Ward’s outperforms the average linkage method except when outliers are present. The data, in this study, were standardized before analysis to reduce the outlier problem. The iterative methods, which include Ward’s minimum variance method, work best when one can identify a nonrandom starting point. However, in the present study, no such nonrandom starting point could be stated. That is, a maid expectations on a profile of attributes for each group could not be specified. Hair et a1. (1987, p. 304) also note that Ward’s has a tendency to "combine clusters with a small number of observations" and is biased toward ”production of clusters with approximately the same number of observations.” As will be reported in the findings, neither problem appears to have occurred with the current data set. The third issue with which a researcher must deal is the number of clusters, which affects the stability and reliability of the cluster solution. In the present study, a number of methods were employed to ensure the correct number of clusters were chosen. The "mixture model approach" was used; that is, the results of cluster analysis using Ward’s method and the complete linkage method were compared. The cluster analysis was run using 75 percent of the sample, and then the remaining 25 percent were reclustered. This provides a 91 test of the stability of the cluster solution, or evidence of convergence. Finally, discriminant functions were derived, and then the observations were reclustered on the basis of those functions. The final issue to be addressed was that of variable selection. The rationale may be based on theory or hypothesis. In this case, the design of the study was based on theory. The choice of attributes was based on the hypo- thesis that descriptive adjectives, relating to living room design, should be used by groups of individuals to construct the living room of their choice. CHAPTER VI: RESULTS A. PARTITIONIN G Ward’s minimum variance method was used to partition the sample into meaningful clusters. Figure 6.1 shows the dendrogram, a visualization, of the clustering process. A plot of the coefficients against the number of clusters yields four clusters. (See Figure 6.2.) Furthermore, on an a pr_iQt_i_ basis, one would expect to find four, since four occupational groups were sampled. As a final check, the cluster analysis was performed specifying that only three groups be formed. In that solution, clusters 1 and 4 (from the final four- cluster solution) remained, for the most part, the same. Clusters 2 and 3 (from the four-cluster solution) were grouped together. The implication is that, in the sense of external validation, the four-group solution can be better interpreted. The design of the measurement instrument provided relatively independent attributes. The correlation matrix, in Table 6.1, shows that most attributes are not significantly related to one another. The highest significant correlation, 0.35, is between 81 (simple) and S6 (understated). Hair et a1. (1987) indicate that attributes which fail to differentiate between or among groups will diminish the quality of the cluster solution. Objective attributes Tl (wood), T3 (fabric), and T5 (brass), as expected, were used by 99 percent, 95 percent and 74 percent, respectively, of all groups. These attributes were removed, and the cluster analysis was rerun. While the 92 u . .- . +uunniii+ +iiu+iniuuii +iii+ +1-11: +iuiii+iiiui *' ¢i+ +inu+i +uiu: _ +- +iii+ +1 _ +iii+u + +111 *S--"+-'- 4h *0---‘ +iii+iiiiu +iui+ +1-11: _ +iii+iiuii +8'-l-+ +8---- _ +-------+----- +""'+' +I'l'-'l'- .M. +iuuiinn+ +iiiii + +aui+ +11: +a+iii +iiit34 +I +uiiii+u _ I +i+ +- _ +iuiii+i +iiniuin+ +siu +iuuii+ +n +i+i +l'- +I'-"*--- +i-uiuuiiiii+ +1 +ininiuu+n +8 +- +111: +1 +iui+n +i+ +- _ +iiu+u +uiuui+i +iuiiiii+ +iuu _ +iiiii+ain +-"-- +unniiiini+ +siiiuinui+ +1 _ _ +8 +i+ +unniiiaon _ _ I +niiia+ +iiu+n +uii+ +1 +iin+i +- D 4 m~ NF me n o acme-co eouuadu ouceuama todeuaoa vogue: nee: ueyu: Innoceueoo 4h - -P an. own AN. Pom as. n~n an, won as as. o- n~¢ a~r ~on +n~ o~+ o- F~+ OPN mos mm. aon np~ nee ca. can op~ are now 3- omN m~s no~ sen ~o~ can o~. an~ New can “N? ps~ ~o. eon Fm. m~n an. com a? New a so. n3~ an. +~F os~ on. e~n n- o~¢ no ..~ Po~ “an no. -n NF~ ~94 oo~ «an oo~ so. mo~ ox» so, pan or o—. so~ man are 3n~ 80. man on. ~nn 38. men so. mm" mm. o~n we. Nan no. oom cor son on~ .n3 Ps~ one no. can an~ nns ~n~ -3 ~n. com com Loans m m < u F.m oa=m_u 93 E6 Figure 6.1 Dendrogram using Hard Method Rescaled Distance Cluster Combine __-+. 135 3: ---+-+ 181 a ..... + +___,. 10 15 20 25 + + 1 + Iii: #117 +88888J +8888'8888* V +88888888888888888 *8-888+88888888888888888 + iiiii + . .-------+--------------- +8 88888 888888888 +8-888888-88-888-8 . _ +1.. +oiiainainii+ +iiiiuinii+iui _ _ _ .- +iia+ +i+i _ .--- +tiiinuuui+iii *888 +uui+ +1 _ +i+u +iiuuuii+ +1 +iiuii+n *888 + i + +uiuiiiiiu+ +niuuiiuii _ + i + +88888888888 +iuuiiiunuiiit+ +iniii+ +iio +i+iii AY'88888888888888-888888-'- +8888- .8888... _ I +iiiui+ +i+i _ +aii+ +- +u+n +iuiuu+ +11: _ +iiu+ +t _ +u+i +iiiii+ +1 +iiiin+u *8888888 A188888AT +ioiiu+ +8-888+8888888 _ _ .--- + iiiii + +iii+uin _ .--- +iiiiiinii+iii ’88- +1 + +888 D II II- I‘ 1‘ *8888888888888 +1 + +i+ +iiuinuiui _ +iiiiiiiii+ +11: *88888+888 +88888888888+ _ _ *888888888 +8888- 94 Difference in % Change Between Coefficients 3.0 -- —_ O 2.0 ~— _— O O 1.0 --— O . . O . O 1 1 1 t l l l 1 1 I 1 I I l l l ' ' Number of 1 2 3 4 5 6 7 8 9 Clusters Figure 6.2 Scree Diagram For Cluster Analysis 95 Table 6.1 -------- Pearson Correlation Coefficients - - - - - - - - Between Subjective Attributes 51 $2 53 S4 55 $6 $7 $8 $9 $10 51 1.00 52 -.10 1.00 S3 .07 -.17' 1.00 54 29" -.16 .18' 1 00 $5 -.01 -.02 -.02 .14 1.00 56 .36" -.12 -.08 .24“ .26'* 1.00 S7 .17‘ -.10 .11 .21H .25" .31" 1.00 58 -.03 -.09 -.08 .0571 .41" .21' .09 1.00 59 -.08 -.07 .17' .01 .27“ .07 .20' .26“ 1.00 510 -.09 .28“ -.14 -.03 .11 -.03 .01 .13 .10 1.00 511 .08 -.01 .13 .27*‘ -.O4 -.03 -.08 -.09 .05 -.13 $12 .05 -.14 -.01 -.01 .23** .08 .21“ .33“ .18It -.13 $13 .14 -.07 .14 .11 -.09 .14 .21‘ -.08 .06 -.03 $14 .23‘* -.12 .12 .12 -.04 .14 .17 —.06 .08 -.03 $15 -.15 -.13 .01 -.05 .18' .12 .23‘"I .26" .30.1. .20* 516 -.23'* .23" -.18* -.01 .15 -.07 .12 .15 .10 .36.. 517 -.18* -.12 .12 -.15 -.03 -.13 .14 .04 .301’. .02 518 -.05 .13 -.14 -.09 .10 .21" -.02 .14 .04 .07 $11 $12 $13 $14 515 S16 $17 518 S11 1.00 512 -.12 1.00 513 .02 .02 1.00 514 .01 .01 .31.1. 1.00 515 -.20* .20* .11 -.04 1.00 516 -.17* .03 -.01 -.03 .20' 1.00 517 -.12 .17' .04 -.01 .30“ .03 1.00 518 -.11 .03 -.04 -.01 .05 .04 .01 1.00 ‘ - SIGNIF. LE .01 *‘ - SIGNIF. LE .001 (Z-TAILED) 96 solution did not change dramatically, the groups were more clearly defined. This analysis was used as the final solution. B. CLUSTER INTERPRETATION AND PROFILING Tables 6.2 and 6.3 contain percentage frequencies on the descriptive statistics for the four clusters that emerged. Table 6.2 contains the descriptive statistics for D7 (income), D10 (education of the household head) and D8 (occupation of the household head). Table 6.3 contains information on V2 (Actual versus Ideal set), V3 (living room style), and V4 (in which room the respondent entertains most). Profiles of the four clusters, on product attributes, are given in Tables 6.4 and 6.5. The first table shows the means for the groups and the univariate F- tests for each attribute across clusters. This output resulted from a MANOVA routine performed on the attributes (dependent variables) by cluster membership (independent variables). The table shows that, for objective attributes, only T8 (wicker) was significantly different between groups. All of the subjective attributes were significantly different across clusters, except Sl8 (impeccable). Figure 6.3 provides a bar chart of the attribute means for the four clsuters. 1. Interpretation of Clusters by Attributes The first cluster is a small one containing 10 percent of the entire sample. Living room sets of individuals in Cluster 1 were characterized by five attributes: simple, harmony, natural, practical, and understated. These attributes provide anchor points for interpretation purposes.l It should be noted that while none 1The assignment of names to clusters is an arbitrary process. However, the a comparison of the mean profile on the clustering variable gives a good indication of the anchor points. 97 Table 6.2 Percentage Frequencies on Demographic Characteristics by Cluster ' Cluster Cluster Cluster Cluster 1 2 3 4 D7 Income 1 Under $ 5,000 2 $ 5,000 > $15,000 4 7 3 $15,000 > $25,000 8 12 19 6 4 $25,000 > $35,000 12 12 18 10 5 $35,000 > $45,000 20 5 11 18 6 $45,000 > $55,000 16 12 13 15 7 $55,000 > $65,000 20 10 11 10 8 $65,000 > $75,000 8 12 6 4 9 $75,000 > $85,000 8 5 3 1 10 $85,000 > $95,000 5 2 4 11 $95,000 > $105,000 4 19 1 14 12 $105,000 > $125,000 2 2 3 13 $125,000 > $150,000 2 14 $150,000 > $175,000 4 15 $175,000 > $200,000 16 $200,000 > $225,000 4 17 $225,000 > $250,000 1 18 $250,000 and up 2 1 Missing Values 7 2 6 D10 Education of HH Head 1 Some high school 4 3 1 2 High school grad 12 12 30 6 3 Some college 12 12 26 13 4 College graduate 16 12 10 ll 5 Masters Degree 4 12 2 24 6 Ph.D. or MD. 52 52 30 46 D8 Occupation 1 Professional 56 62 32 63 2 Semi-professional 24 5 6 19 3 Skilled Non-Prof. 4 14 16 4 4 Non-Skilled Non-Prof. 16 12 40 13 Missing Values 7 6 1 ’Percentages on some variables, for some clusters, may sum to more than 100 due to rounding error. 98 Table 6.3 Percentage Frequencies on Living Room Characteristics by Cluster " Cluster Cluster Cluster Cluster 1 2 3 4 V2 Actual vs Ideal Set? 1 Actual 68 43 48 62 2 Ideal 20 45 40 3O 3 Both 12 10 3 6 4 Other 2 8 3 Missing Values 1 1 V3 Living Room Style 1 Colonial 2 5 6 2 Traditional 36 5 25 40 3 Modern/Contemporary 24 55 23 17 4 Victorian 7 5 French Provincial l 3 6 Country 8 12 25 7 7 Eclectic 28 14 12 14 8 Modern/Classic 9 Period 2 10 International 2 11 Asian 12 Mix 4 5 2 4 13 Other 2 7 3 Missing Values 1 V4 Entertain Most? 1 living room 40 62 61 56 2 family room 32 14 24 21 3 den/library 5 l 8 4 dining room 16 12 5 7 5 kitchen 12 2 4 7 6 other 5 6 l ‘Percentages on some variables, for some clusters, may sum due to rounding error. to more than 100 Cluster Means on Subjective Attributes 99 Table 6.4 and Univariate F -Tests Mean Mean Mean Mean Signif. Cluster Cluster Cluster Cluster F of Variable 1 2 3 4 Value F T2 Stone .056 .096 .073 .066 .98618 .400 T4 Glass .220 .289 .315 .280 1.85782 .137 T6 Chrome .072 .105 .080 .039 1.84535 .140 T7 Leather .053 .099 .099 .07 2 1.30183 .274 T8 Wicker .020 .035 .074 .012 4.28791 .006 81 Simple .674 .292 .264 .169 26.37399 .000 82 Futurist .022 .202 .083 .034 14.00777 .000 S3 Cozy .364 .150 .263 .243 6.32027 .000 S4 Natural .523 .321 .252 .251 7.62315 .000 85 Authentic .186 .160 .092 .250 7.74669 .000 S6 Understate .341 .101 .052 .096 13.73401 .000 S7 Harmony .614 .307 .090 .347 36.91209 .000 S8 Classic .151 .061 .099 .361 21.34003 .000 S9 Charming .158 .061 .073 .313 21.15803 .000 $10 Dramatic .056 .308 .067 .132 18.93769 .000 811 Rustic .128 .045 .153 .037 6.46137 .000 812 Tradition .193 .144 .091 .41 1 27.80504 .000 813 Comfort .496 .346 .266 .291 1 1.00471 .000 S14 Practical .670 .468 .352 .301 10.85447 .000 S15 Gracious .152 .120 .043 .225 13.09804 .000 S16 Distinct .090 .566 .126 .357 51.76398 .000 8” Pretty .086 .071 .171 .338 13.37113 .000 $18 Impeccable .063 .041 .035 .066 .7 7441 .509 100 Table 6.5 Percent of Cluster Members Who Used Attributes At Least Once Stone (31ass Chrome Leather/Suede Wicker Simple Futuristic Cozy PJatural Authentic Understated Harmony Classic Charming Dramatic Rustic Tradition Comfort Practical Gracious Distinctive Pretty Impeccable Cluster Cluster Cluster Cluster 1 2 3 4 32 48 38 38 76 88 88 86 20 36 28 14 20 43 39 33 8 10 26 7 96 71 68 56 16 55 35 17 76 62 76 72 88 71 67 67 52 50 34 57 6O 33 17 25 96 67 32 76 36 41 35 74 44 26 31 65 28 74 28 43 28 17 46 15 56 45 31 80 100 95 78 83 96 88 78 83 40 45 18 57 36 100 51 81 32 29 49 71 16 14 10 24 101 Mean $ h-- ——r — —————— b ————— Clusterl l- 31 $2 33 $4 $5 so 57 sa so 810 sn’sm 813 su $15 sre sn Attrbutas Mean w r— ———————————————————— I-n— — Cluster 2 3 .7 . 7 . _. ""-J 81 $2 83 S4 85 $6 87 $8 89 510 $11 812 813 $14 815 $16 817 Altrbulaa Mean .50 _. ———————————————————————— Cluster3 51 52 53 ’54 ss 55 S7 53 59 510 511 $12 513’s“ 515 515 517 Attrbutes Mean '50 — ———————————————————————— Cluster4 51752 S3 S4 55 $557 $5 59 $10 $11 512 511514 515 516517 Anrbutos Figure 6.3 Cluster Profiles on Subjective Attributes 102 of the attributes were highly correlated (as mentioned earlier), simple and understated were significantly, positively correlated with each other. However, this fact merely places additional emphasis on the the use of simplicity by this cluster. While three of four clusters included the attribute comfort in their sets, Cluster 1 had the highest mean use of this attribute. Cluster 2 contained 16.9 percent of the entire sample. The sets of individuals in this cluster were characterized by three attributes: distinctive, dramatic, and futuristic. Table 6.5 shows that all individuals in this group used the attribute distinctive at least once in their sets. This cluster was also high on the use of the attribute comfort. Cluster 3 contained the largest portion of the sample, 43.95 percent. Living room sets constructed individuals in this cluster had the highest mean values, within group, on three attributes: cozy, practical, and comfortable. In Table 6.5, note that, compared to other clusters, this cluster used fewer attributes per product and also developed fewer products. However, the effect of the heavy use of attributes, by other clusters, would have been minimized by the process of standardization discussed in an earlier section. Thus, this cluster did not have a larger mean inclusion on any attribute with the exception of rustic. Perhaps it is better to note that this cluster had the lowest use of the attributes gracious and harmony. The final cluster, Cluster 4, comprised 29.03 percent of the entire sample. These subjects developed sets composed of three attributes: tradition, classic, and authentic. In general, members of this cluster tended to be more expressive when describing the products they created. Table 6.5 shows high values on all attributes relative to other clusters. These four clusters were found to be significantly different from one another based on the results of a MANOVA performed on the original clustering 103 Table 6.6 Results of MANOVA for Attributes by Groups Multivariate Tests of Significance (S - 3, M = 9 1/2, N = 110) Test Name Value Approx. F Hypoth. DF Error DF Sig. of F Pillais 1.58997 10.98205 69.00 672.00 .000 Hotellings 3.54238 11.32878 69.00 662.00 .000 Wilks .09998 1 1.17881 69.00 664.07 .000 Roys .60767 104 attributes. Results are found in Table 6.6. Wilks’ Lambda was 0.09998, with p = .001 or better. Since clusters were formed using Ward’s minimum variance method, the MANOVA simply provides evidence of a significant difference in the between- versus within-group variances [W - A]. 2. Profiling on Variables Not Used in the Clustering Procedure The profiling portion of analyzing cluster results is extended to comparing clusters on (a) products, (b) demographics, and (c) a £12111 group memberships, which were not used for the purpose of clustering. a. In general, when asked to create a living room set, the products created most by individuals were sofa/couch (P1), chair (P2), coffee table (P4), end table (P5), and lamp (P12). (Refer to Table 6.7 for the discussion to follow.) Distinct clusters did have different proclivities to create products. Cluster 1 created the largest percentage of chairs and second chairs (usually rockers, re- cliners, or easy chairs). These persons included lamps and bookcases in their rooms. Cluster 2 added fireplaces and paintings, as well as stereos. Cluster 3, by far, had the highest percentage of televisions. In the category of large accessories, this cluster tended to include items such as foot stools, grandfather clocks, and, on occasion, woodstoves. Cluster 4 had the lowest percentage of televisions and stereos. b. Profiling would not be complete without considering how well the cluster analysis classified the a m groups. Table 6.8 addresses this issue. Since the original occupational groups contained white-collar and blue-collar individuals, the table has classified cluster members by occupational status. The discussion here should be considered in light of the demographic and living room characteristics of each group presented in Tables 6.2 and 6.3. Of Cluster 1, a relatively small group of individuals, 80 percent came from white-collar occupation groups. Individuals in this cluster had relatively high 105 Table 6.7 Percentage Creation of Products by Cluster Cluster Cluster Cluster Cluster Var“ Product 1 2 3 4 Pl Sofa/Couch 96 93 95 93 P2 Chair 1 84 67 73 74 P3 Chair 2 32 19 25 22 P4 Coffee Table 60 67 46 49 P5 End Table 40 33 41 42 P6 Desk 12 0 5 11 P7 Entertain Center 8 19 17 18 P8 Piano 12 12 6 32 P9 Fireplace 32 43 29 24 P10 Television 28 31 53 21 Pl 1 Stereo” 16 29 26 1 1 P12 Lamp 68 45 65 58 P13 Drapes 12 14 6 11 P14 Carpet 44 43 33 46 P15 Plant 8 7 7 10 P16 Painting/Picture 20 45 26 49 P17 Folding Screen 0 5 1 6 P18 Art Objects 0 5 6 10 P19 Small Accessories” 8 12 15 24 P20 Large Accessories” 52 29 39 33 ’ "Var” = Variable " The category ”stereo” also included some VCRs. Small Accessories and large categories included miscellaneous items. In the case of clusters 1 and 4 large accessories included bookshelves; for group 3 this category often contained foot stools. 106 Table 6.8 Cluster Membership By Occupational Group Membership“ Blue Collar Group 2 White Collar Group 1 Cluster 1 80% Cluster 2 69% Cluster 3 38% Cluster 4 83% 20% 31% 62% 17% The table should be read as in the following example. "80% of the members of Cluster 1 originated in the white collar occupations.” Table 6.9 Actual Versus Ideal Living Room Set By Cluster and by Group Cluster 1 White Collar Blue Collar Cluster 2 White Collar Blue Collar Cluster 3 White Collar Blue Collar Cluster 4 White Collar Blue Collar Actual Ideal Both Other Missing Sum 17 5 3 0 0 25 (68%) (20%) (12%) 12 5 3 0 0 20 5 0 0 0 0 5 18 19 4 1 O 42 (42.9%) (45.2%) (9.5%) (2.4%) 14 12 3 0 0 29 4 7 1 1 0 13 52 44 3 9 1 109 (47.7%) (40.4%) (2.8%) (8.2%) (.09%) 22 12 2 6 0 42 30 32 1 3 1 67 44 21 4 2 1 72 (61%) (29%) (5.5%) (2.7%) (1.4%) 39 17 2 1 l 60 5 4 2 1 0 12 108 incomes, and 60 percent of the heads of household were employed as professionals. More than two-thirds of these people said they had created their own (actual) living room set. (See Table 6.9 for comparison of Actual versus Ideal). Many said they use their living room for "reading" and 60 percent entertain in rooms other than the living room. Members of Cluster 2 were predominantly white collar also. This group is united by the fact that a relatively large portion were describing their "ideal set" or a combination of their actual and ideal sets (54.7 percent). 69.1 percent described the living room style they created as ”modern/contemporary.” This group had the highest mean use of the attribute futuristic. Membership in Cluster 3 was 62 percent blue-collar. The other 38 percent cluster were individuals originating in other groups. In a comparison of the ac- tivities for which the living room is used, variables Al-A7, the white-collar in- dividuals, in Cluster 3, tended to read and watch television in their living rooms as opposed to entertain. In this respect, they are similar to the blue-collar members of this cluster. Of the individuals in Cluster 3, 53.9 percent said they had created their ”ideal set.” (Discussion of ideal sets follows in section E-2.) More than four-fifths of Cluster 4 worked in white-collar occupations. Sixty-one percent of the cluster said they were describing their ”actual" set, and the styles they described were split between ”traditional” and ”modern/contem- porary." Twenty-eight percent of the cluster originated in Group 4; while they were not unusually high on the education variable, their incomes were high relative to the average for the a m Group 4. These individuals were similar to the rest of Cluster 4, in that they were describing their actual set and said they entertained most in their living room. It is interesting to note that the 10 percent of people who originated in Group 3 said that they were describing their ideal set. 109 C. INTERNAL VALIDATION Internal validation was carried out by the split sample validation technique. Reliability of the solution can be demonstrated by cross-validation (Punj and Stewart 1983). Seventy-five percent of individuals were reclustered, and the same clusters emerged. Then the classification of individuals into clusters was checked to determine whether they were clustered into the same groups as they had been. The analysis showed that 21.7 percent, or 43 out of 198 individuals, were misclassified. While this rate is high, 28 of the 43 (65 percent) were original members of cluster 3 who were misclassified into cluster 2. Earlier discussion showed that if one were to use a three-cluster, instead of a four- cluster solution, cluster 2 and 3 would be grouped together. Therefore, the fact that the 75 percent solution misclassified individuals in clusters 2 and 3 is not surprising. The finding reinforces the close relationship between these two clusters. Internal cohesion was tested using MANOVA as described earlier. As part of the output of the MANOVA, discriminant weights for the functions that differentiate clusters are produced. (See Table 6.10 for the standardized discriminant weights and discriminant functions.) Using these weights, 3 discriminant analysis was performed on a random sample of 90 percent of the total observations. The grouped cases correctly classified was 88.9 percent. Results appear in Table 6.11. All of the holdout sample of 10 percent was correctly reclassified. 110 Table 6.10 Standardized Discriminant Coefficients and Discriminant Functions Function 1 Function 2 Function 3 S12 -0.41450‘ 0.16525 0.15812 S9 -0.36554" 0.07370 0.21381 S8 -0.361 18‘ 0.07326 0.18699 81 7 -0.30223"I -0.08820 0.02441 815 -0.22372’ 0.21151 0.17913 85 -0.l8695* 0.15884 0.12501 T6 0.1 1386’ 0.02948 -0.06341 S16 -0.17541 0.61081“I -0.39134 SlO -0.00701 0.35222“ -0.29995 81 1 0.141 13 -0.20620‘ 0.06607 T8 0.11009 -0.13l73’ -0.11153 S7 -0.08073 0.38926 0.53362‘ S1 0.30622 0.11561 0.43574’ S6 0.08550 0.16039 0.39812“ S2 0.16658 0.17422 -0.34428‘ S3 0.02662 -0.1 1617 0.28293“ S13 0.09555 0.19826 0.26951‘ S4 0.14358 0.10986 0.22790‘I S14 0.18200 0.12964 0.18414‘ T7 0.05035 -0.02275 -0.l6505"' T4 0.00273 -0.03477 -0. l 4850"I S18 -0.07694 0.09044 0.10914“ T2 0.01275 0.0271 1 -0.04698‘ 111 Table 6.11 Classification Results for Discriminant Analysis No. of Predicted Group Membership Actual Group“ Cases 1 2 3 4 Group 1 22 21 0 l 0 95.5% 0.0% 4.5% 0.0% Group 2 38 l 35 l l 2.6% 92.1% 2.6% 2.6% Group 3 97 2 5 88 2 2.1% 5.2% 90.7% 2.1% Group 4 68 l 3 8 56 1.5% 4.4% 11.8% 82.4% Ungrouped Cases 23 3 4 13 3 13.0% 17.4% 56.5% 13.0% Percent of "Grouped” Cases Correctly Classified: 88.89% *Note: The term group is as used in discriminant analysis. It should not be confused with the original groups as defined in this study. The groups in the discriminant analysis are actually clusters. 112 D. EXTERNAL VALIDATION AND DISCUSSION The issue of external validation for cluster analysis is usually answered by asking whether the solution is useful (Punj and Stewart 1983, p. 146): ”Classification is only useful if it assists in furthering an understanding of the phenomena of interest." The efficacy of the solution lies in its ”ability to discriminate between...subpopulations" (ibid., p. 141). The present cluster analysis has shown that people are able to create a set of attributes and products. More important, different groups of individuals would create different living room sets from a relatively parsimonious list of objective and subjective attributes. While each individual created his/her own idiosyncratic version of a living room set, the analysis demonstrated that the consumption sets created were related to group membership. Further analysis implies that sets are related to variables in addition to cluster membership. E. RESULTS OF HYPOTHESIS TESTS AND DISCUSSION 1. Hypotheses Concerning Consistency within Sets Hypothesis 1 stated that there should be consistency within sets of attributes created by status groups of individuals. To the extent that the clusters resulting from the cluster analysis were dominated by one occupational group, the cluster analysis lends support to this hypothesis. Table 6.12 shows the percentage breakdown of the occupational groups by cluster. However, in order to test this hypothesis, a randomized block design MANOVA was conducted, with the objective and subjective attributes as dependent variables and the first three original occupational groups as the independent variable; two blocks of 50 percent of the subjects were chosen at random from the 112 D. EXTERNAL VALIDATION AND DISCUSSION The issue of external validation for cluster analysis is usually answered by asking whether the solution is useful (Punj and Stewart 1983, p. 146): "Classification is only useful if it assists in furthering an understanding of the phenomena of interest." The efficacy of the solution lies in its ”ability to discriminate between...subpopulations" (ibid., p. 141). The present cluster analysis has shown that people are able to create a set of attributes and products. More important, different groups of individuals would create different living room sets from a relatively parsimonious list of objective and subjective attributes. While each individual created his/her own idiosyncratic version of a living room set, the analysis demonstrated that the consumption sets created were related to group membership. Further analysis implies that sets are related to variables in addition to cluster membership. E. RESULTS OF HYPOTHESIS TESTS AND DISCUSSION 1. Hypotheses Concerning Consistency within Sets Hypothesis 1 stated that there should be consistency within sets of attributes created by status groups of individuals. To the extent that the clusters resulting from the cluster analysis were dominated by one occupational group, the cluster analysis lends support to this hypothesis. Table 6.12 shows the percentage breakdown of the occupational groups by cluster. However, in order to test this hypothesis, a randomized block design MANOVA was conducted, with the objective and subjective attributes as dependent variables and the first three original occupational groups as the independent variable; two blocks of 50 percent of the subjects were chosen at random from the 113 Table 6.12 Breakdown of Occupational Groups by Cluster Cluster 1 Cluster 2 Cluster 3 Cluster 4 White Collar 13% 19% 28% 40% Blue Collar 5% 13% 69% 12% 113 Table 6.12 Breakdown of Occupational Groups by Cluster Cluster 1 Cluster 2 Cluster 3 Cluster 4 White Collar 13% 19% 28% 40% Blue Collar 5% 13% 69% 12% 114 occupational subsamples. The purpose of the blocks was to test for the presence of any effects due to sample stratification.’ The results of the test show that Wilks’ Lambda for the occupational treatment was 0.46544, which was significant at the level of p = 0.001 or better. The blocking variable was insignificant; Wilks’ Lambda was .89752, with p a 0.627. Therefore, the results of the test support the hypothesis that, given the opportunity to create a living room set of their choice, individuals within occupational groups will create sets of attributes that were more alike than those M510. groups. In addition, the results of the cluster analysis show that some attributes cut across clusters; that is, attributes such as practical, comfort, and cozy were used by all groups at about the same levels. It would appear that these are more universally desired attributes than others. Moreover, in their discussion of the meaning of home, Csikszentmihalyi and Rochberg-Halton (1981, p. 127) state: "The single characteristic of the home most often mentioned was ’eomfortable,’ ’cozy,’ or ’relaxing.’ Of the respondents, 41 percent noted this quality.” The present study also found, with relation to living rooms, that cozy and comfortable were used by a large proportion of all groups. However, these attributes were found to be distinct, that is, not significantly correlated (see Table 6.1). Figure 6.4 is a pictorial version of Table 6.5, with the exception that the attributes have been reordered to depict those attributes that cut across clusters. The clusters have also been reordered to demonstrate the effect. Figure 6.4 is similar to Figure 4.3 in the discussion of the conceptual model in Chapter III. The results of the empirical analysis in this section support the cultural- subcultural portion of the conceptual discussion. 2The fourth a priori group was a combination of three different occupa- tional groups; sample sizes within each group were too small to be included in the analysis. 115 eiqeooadtul onsnu onsumna 11011110811 paletsmpun meld 811010819 011910810 3133213 5U1LUJBLK) 91113111;ng ogtuaqinv AuouueH [(203 eidtugs lBJHlBN 110111103 leonoetd l l l Cluster 4 Cluster 2 Cluster 1 Cluster 3 , Figure 6.4 Attribute Commonality Across Cluster 116 The results that appear in Table 6.7 demonstrate consistency in the creation of products within a cluster. Again, as with attributes, products such as sofa, chair, coffee table, end table, and lamp cut across clusters. To the extent these products are abundantly available in the marketplace and possession of these products is dictated by tradition, this outcome is expected. It can be concluded that the findings support the notion that groups of in- dividuals will create similar consumption sets. 2. Actual versus Ideal Set The second hypothesis dealt with the issue of actual versus ideal set. The respondents were asked two questions in this regard. The first was whether they had a whole picture in mind as they went through the exercise. Eighty- eight percent of responded in the affirmative. The subjects were then asked whether that picture was their "actual" set, their ”ideal set," or something else (”other"). (Refer to Table 6.9 and to the questionnaire in Appendix B). Only those three options were provided. Respondents used the "other” option to describe what they had in mind. Often their description was a combination of both. Although the option ”both" was not provided, content analysis of the other category was used to judge whether that response should be coded for a particular respondent. To test the hypothesis that development of ”actual” instead of "ideal” is dependent on age, a Chi-square test was performed. The value of Chi-square with 15 degrees of freedom was 20.99294. The significance level for the test was 0.1371. Therefore, the hypothesis that actual/ideal sets are independent on age cannot be rejected. Cramer’s V, a correlation coefficient for categorical variables, was 0.16866. In an effort to further explore the the nature of individuals who created actual versus ideal sets a discriminant analysis was performed. It was 117 hypothesized that whether a person created an actual or an ideal set would be related to family life cycle variables. The dependent variable, for the discriminant analysis, was the categorical variable, "actual or ideal,” with independent variables -- age, marital status, and presence of children 6 years old and below, 6-17 years of age, and 18 years of age and older. All variables had a positive contribution in discriminating between actual sets and ideal sets, except the presence of children under the age of six. The signs of all the variables were as expected. The function discriminated between actual and ideal set 60 percent of the time. Wilks’ Lambda was 0.9378, which was significant at the 0.05 level. The prior probability of being classified in a group was 50 percent. Thus, the model improved on chance to some extent. Across all clusters, 55 percent of respondents created their own "actual" set. Slightly more than one-third said they had created their ideal. Respondents in the category of "both” often said ”everything was mine except...," and the exception was ”the big screen television” or ”the baby grand piano.” These averages can be broken down by cluster. In Cluster 1, 68 percent created their actual set, 20 percent their ideal. In Cluster 2, 43 percent created their actual set and 45 percent their ideal. In Cluster 3, 48 percent created their actual set and 40 percent their ideal. Finally, in Cluster 4, 61 percent created their actual set and 29 percent their ideal. (Refer to Table 6.9.) When asked whether they had a picture in mind, most people replied yes. To the extent that consumers carry these pictures of sets into the buying situation, one must wonder what role the set plays in buying decisions. The actual or ideal set, that a consumer has in mind may provide the context into which all competing, alternative products are placed. The fact that most subjects in this study had a whole picture in mind lends support the conceptual framework discussed in Chapter III. 118 3. Discussion In the case of a living room set, the components of the set (products and attributes) appear to be related to the activities for which the living room is used. (See Table 6.13.) Univariate F-tests on the difference between the mean activities for which the living room was used show that a significant difference exists between means for entertaining, watching television, eating, listening to music, and other. While the living room is clearly a multipurpose room for all clusters they differ by the purpose for which they use their living room most. Table 6.14 shows that Cluster 3 watches television and eats in the living room more of the time than the other clusters. Cluster 1 uses the living room for other activities more than do clusters 2, 3, and 4. While the value was not significantly different, members of Cluster 1 also do more reading in the living rooms (see Table 6.13). Cluster 2 listens to music and entertains more, on average, than do members of other clusters. Similarly, members of Cluster 4 also entertain and listen to music more than do Clusters 1 and 3. Table 6.15 gives the group means on demographic characteristics and on the life-style scales. Cluster 2 is slightly younger than Clusters 3 and 1. There were more singles, and members had relatively high education and income. These characteristics may account for members having checked ”ideal" set as opposed to actual. The individuals in this cluster also had the lowest desire to be part of a group (L11). The fact that they described their sets as distinctive may be related to this trait. Cluster 3 members were the youngest of the four clusters. On average, they had ”some college,” which makes them less educated compared to the other groups. This group also had the lowest mean income of the four groups, and a somewhat larger portion of singles. This cluster, like Cluster 2, said they described their ”ideal” set. As noted earlier, this cluster used fewer attributes. 119 Table 6.13 Cluster Means for Living Room Activities Var Description Al A2 A3 A4 A5 A6 A7 Entertaining“ Reading Watch Television“ Eating“ Playing Games Listening/Playing Music“ Other“ Cluster Cluster Cluster Cluster 1 2 3 4 .260 .299 .209 .282 .240 .202 .201 .231 .148 .171 .277 .163 .068 .045 .082 .038 .048 .081 .052 .046 .140 .220 .146 .200 .100 .024 .034 .036 “ Significant at the 0.05 level or better. Activities by Cluster 120 Table 6.14 Watching Television Eating Other Listening to Music Entertaining Cluster Cluster Cluster Cluster 3 1 2 4 .277 .148 .171 .163 .082 .068 .045 .038 .034 .100 .020 .034 .146 .140 .220 .200 .209 .260 .299 .282 Cluster Means for Other Variables Variable Description V1 V2 V3 V4 D1 D2 D3 D4 D5 D6 D7 D8 D9 PIC ACT? STYLE EMOST SEX AGE MARTL DEP-6 DEP6-17 DEP18+ INC OCCl OCC2 D10 EDUCl D11 EDUC2 L1 L2 L3 L4 L5 L6 L7 L8 L9 Latest Style Child’s Convenience Year in Paris Don’t Like Chores Art Gallery More Self-confident Quiet Evening Service Org. Member ”Once Over Lightly" L10 Dress for Fashion L11 Group Members L12 Enjoy Concerts L13 More Independent L14 I Like Parties L15 Personal Ability L16 Read Sports Page L17 Teach my Children L18 Political Campaign L19 Leader L20 Like Ballet L21 Like to Play Golf L22 Cleaning Unpleasant L23 Rather go to Sports Event L24 Trip Around World Cluster 1 1.000 1.440 4.360 2.280 .640 3.520 2.000 .240 .680 .200 5.960 .800 1.313 4.600 4.333 2.680 2.947 3.040 3.200 3.520 3.720 3.760 3.000 2.720 1.920 3.880 3.920 4.320 2.800 4.360 2.760 4.222 2.480 4.120 2.640 1.440 3.400 3.560 4.080 121 Table 6.15 Cluster 2 1.167 1.714 4.810 2.071 .786 3.230 1.762 .190 .595 .095 7.205 .744 1.063 4.810 4.000 2.500 3.296 3.071 3.357 3.357 3.595 3.476 3.357 2.857 2.310 3.275 3.667 4.262 3.357 4.310 2.714 4.192 2.190 3.690 2.524 2.025 3.1 19 3.548 3.881 Cluster 3 1.176 1.713 4.815 1.862 .817 2.514 1.917 .385 .578 .064 5.623 1.667 1.964 3.697 3.01 1 2.780 3.145 2.413 2.587 2.587 3.817 3.385 3.339 2.734 2.009 3.716 3.444 4.202 3.312 4.303 2.807 4.217 2.037 3.743 1.752 1.963 2.771 3.596 3.771 Cluster 4 1.085 1.493 4.028 2.014 .653 3.528 2.056 .264 .708 .208 7.721 .662 .949 4.875 4.069 2.930 3.057 3.282 3.806 3.806 3.887 3.714 3.314 2.652 2.250 3.629 3.803 4.310 3.414 4.352 2.845 4.235 2.338 3.817 2.806 2.200 2.986 3.493 4.264 122 Life-style scales show that these individuals are not "arts” oriented (L5 and L20). They are also the least venturesome (L3 and L24). However, in line with their choice of the attribute ”practicality," members of this cluster were the least averse to cleaning and housework (L4 and L22). While the preponderance of members of Cluster 1 originated in Group 1, members of this cluster had a lower mean income than Group 1 members who were clustered into Cluster 4. Members of this group enjoy going to concerts (L12) but do not like to go to lively parties (L14). Individuals in this cluster were more oriented to comfort in their living rooms than the other clusters (discussed earlier). Interestingly, more than other clusters they also dress for comfort, more than fashion (L10). Cluster 4 contained the most professionals. This cluster had the highest mean income (note in Table 6.3 that the distribution of income in the group was bimodal), and the highest education. Individuals in this cluster dress for fashion (L10) and have outfits in the latest style (Ll). They are more venturesome than members of Cluster 2 and 3 (L3 and L4). Concerts, art galleries, the ballet, parties, and golf are all on their agenda. As discussed earlier, this cluster was the most descriptive of their living room. They used more attributes than other clusters. Their venturesomeness may expose them to richer environments. In discussing the results of the analyses, it is important to note that attribute descriptions given by clusters could be related to a wide variety of living room "styles.” Referring to Table 6.3, the "simple, natural, harmony” of Cluster 1 was found in living room styles ranging from traditional (36 percent), to modern/contemporary (24 percent) to eclectic (28 percent). The ”classic- traditionals" in Cluster 4 had living room styles that crossed the whole range. Most described their living room as traditional (40 percent), but 17 percent were modern/contemporary, 14 percent eclectic, 7 percent each for country and vic- 123 torian, and so forth. Cluster 3’s ”comfortable-practicality" included 25 percent traditional and 25 percent country. Styles for cluster 3 were also widely dispersed. Cluster 2, the ”distinctive-dramatic” individuals, had the highest single percentage for a style -- 55 percent modern/contemporary. However, this cluster also contained eclectic (14 percent), country (12 percent) and small percentage of all other styles. Members of Cluster 4 and Cluster 2 are related in the sense that they both use their living room for entertaining. While Cluster 2 differs from Cluster 4 in that members like items that are modern and futuristic, as opposed to classic, they are similar in that both used the attributes ”dramatic” and ”distinctive” more often than the other clusters. Both of these groups included paintings in their sets; Clusters 1 and 3 included these at a lower rate. (See Table 6.15 for cluster means). However, where Cluster 4 included a piano, Cluster 2 included a fireplace. Clusters l and 3 are related in that they do not entertain as much as the other clusters. They use their living room for more sedientary activities. The sets created by these individuals contained more chairs and lamps than those of members of Clusters 2 and 4. They appear to take a more pragmatic approach to furnishing their sets. CHAPTER VI: CONCLUSIONS, LIMITATIONS, AND FUTURE RESEARCH The conceptual model presented in Chapters III and IV represents a departure from conventional avenues of consumer behavior research. The model suggests that, for some research purposes, it is advantageous to study choice as the beginning of the consumer behavior process. It was also suggested that some information about a consumer’s life is captured in the sets of attributes and products a person creates and part by the process the consumer uses to create them. First, the conceptual material presented illustrates one means of using the information contained in sets of attributes. The commonality in attributes and attribute combinations provides a link between individuals. As developed, the model would assist researchers in inspecting and incorporating the influence of groups -- culture, social class, and reference groups -- on an individual’s consumer behavior. In addition, a great deal can be learned from the process by which sets of attributes are constructed. The model proposes that people build consumption sets to facilitate their lives. However, those sets are not constructed in a vacuum. It is recognized that each person is an individual, and given the variety of attribute combinations available in the marketplace, each individual will develop an idiosyncratic set. However, several factors should lead individuals to develop sets with common factors. Similar technologies should lead individuals within a group to seek similar sets of attributes. Moreover, the need and/or desire to be recognized as 124 125 a member of a group should lead to commonality in observed attribute combinations. Given the conceptual model, one should be able to observe differences in the sets created by different status groups. This study used occupational groups as a proxy for status groups. This choice of subjects should have provided a conservative test of the model. That is, use of occupational groups should, to some extent, limit the ability to detect the presence of intragroup differences. Nevertheless, it was shown that the identified clusters tended to be comprised of individuals of similar occupational groups. Moreover, the results of the study showed differences in the attribute sets created by members of the original occupation groups (prior to the homogenizing process of clustering). The cluster analysis resulted in four clusters which revolved around the attributes, simple, distinctive, practical and traditional, respectively. An important extention of this research would be to determine whether a member of, for example, Cluster 2 whose living room was contemporary would recognize the similarity in another cluster members’ country style living room. Between the clusters, individuals differed to some extent in general life- style characteristics and on the activities for which they used their living rooms. There is some evidence to conclude that white-collar individual, who clustered into the predominantly blue-collar cluster used their living room for activities similar to those of the blue collar subjects. The study also found that some attributes, such as comfort and practicality, were similar across all clusters. This finding suggests the influence of culture. While the attributes mentioned may seem like obvious choices, it should be noted that in a historical sense, comfort and practicality were not always considered absolutely necessary. The history of dress provides many examples such as tight laced shoes, restrictive garments for women, and military apparel. Moreover, the 126 subjects in this study chose ”comfort" as defined by a culture that has been trained to sit. Their idea of comfort may differ from individuals in other cultures who are trained to lie. Finally, this study found a difference in the mean use of the attribute comfort and practical by the identified clusters. The results are not conclusive, but Cluster 4 had the lowest mean use of these two attributes and they were also the individuals who did a great deal of entertaining in their living rooms. These individuals may trade off some comfort and practicality in order to achieve a mix of attributes desirable for their type of entertaining. Commonality in the construction of products, such as couch and coffee table, may also suggest evidence of culture; or alternatively, it may be an artifact of the availability of products in the marketplace. Further research is needed to explore the latter phenomenon. There are several limitations on the generalizability of the study. First, the instrument was designed to study only one part of a consumption set, the living room subset. Consequently, findings cannot be generalized to other parts of the consumption set. Other subsets may be less prone to status group differences than the living room. However, other subsets which are used to signal membership should demonstrate a similar effect. For the most part, only two status groups were included in the study. Moreover, generalizability is somewhat limited by the fact that the sample was over-representative of persons with higher education and of males. This limitation is not serious, since the research was exploratory. It was not the intent of the study to provide a representation of the living room sets of all members of society. Still, one must wonder whether a sample that consisted mostly of females would have provided a different set of living room attributes. Gender differences must be explored further. 127 From a conceptual perspective, the test was static. It was not meant to capture the dynamics of the process of consuming style but, rather, to test for the existence of intragroup differences in consumption sets. The test asked respondents to construct any living room set and not specifically their own living room. Nor were respondents asked how they developed their set or how they planned to change their set in the future. Nevertheless, when given the opportunity to create any living room set, slightly more than half the subjects created "their own" set. Others created their ideal set. In some cases, individuals who created their ideal set were found in clusters outside the cluster to which the majority of their occupational group belonged. Further exploration showed that the creators of ideal sets also tended to have less than well-established homes; that is, they were younger, often single, and/or had children under the age of six. While age alone did not classify individuals into an actual or ideal category, the combination of age, marital status, and age of children correctly classified people. However, even the combination of these factors does not entirely explain the differences, and thus this represents an opportunity for further research. The study concludes that upon being given the task of building a set, subjects were readily able to access and construct sets. In doing so most people had a ”whole” set in mind. To the extent that people are able to access these sets in purchasing situations, the implications for practitioners are extensive. While this study did not specifically test the relation between sets and purchasing, it is fruitful to speculate on the significance of sets in buying decisions. In choosing a new product, do people select a product whose attributes fit into a pre-existing set? Or, in developing a set, how does the notion of a pre-existing "ideal" set affect the potential purchaser’s choice? Perhaps more important, what happens when a consumer’s notion of an ideal set 128 dictates the purchase of a products whose attributes are not perceived to complement each other after they are combined? The result may be dissatis- faction. In such a case, dissatisfaction would not be a result of disliking the product’s attributes per se. Rather, dissatisfaction may be a result of the attribute’s combination possibilities. The issue of sets in various stages of transition may prove to be useful in the study of satisfaction. In sum, one question that remains for future research is whether sets provide the context for buying decisions. Future research on process must include the variety of facets of learning that occur in the context of consumer behavior as defined in this model. Consumers learn as they combine products to obtain desired attributes. In this process, they must learn which attributes result in the combination desired. Moreover, consumers must learn which attributes are consistent with the sets of the various groups to which they have an affinity or are affiliated. Information processing issues also arise from the model conceptualization. Processing of information about sets may be more complex. On the other hand, the existence of a set, either actual or ideal, may act to reduce the amount of information with which a consumer must deal. A related issue is involvement. Are purchases of attributes and products -- which will become part of a set-- more involved? These purchases may be among the purchases for which consumers ask assistance, or turn to shopping professionals (Hollander 1971). The model facilitates longitudinal analyses of the process of building sets. How do people acquire and build sets? What role do marketers play in the learning process of consumers? Under what circumstances do sets change? A fascinating avenue for future research involves the study of the effect of one product, with its constituent attributes, on the set. Can the addition of one product alter the entire set? Recognition and incorporation of sets may enable 129 marketers better to formulate strategies, for the long run, that are mutually beneficial to both themselves and to consumers. By including process and incorporating consumption sets, the model facilitates inclusion of an expanded notion of learning and the inclusion of exogenous variables. It has been shown, in this study, that some attributes cross over status groups. These attributes may be evidence of the influence of broader groups, such as society or culture. The conceptual model provides the means to explore the influence of society. However, the question remains open as to the mechanism by which this influence is transferred. Moreover, marketers, entrepreneurs, and innovators are all societal entities that may influence the development of sets. To the extent that marketers preselect the assortments of products and attributes from which consumers choose, marketers may partially help to define individual sets. The model allows for research into the influence of marketing, at a macro level, on consumer behavior. Members of other cultures form sets and develop particular consuming styles. The conceptual model developed should eventually enhance our ability to make cross-cultural comparisons. Inclusion of a large proportion of objective attributes in a set may be considered materialistic, when compared to the exclusion of those attributes, or to the inclusion of subjective attributes. The latter might be labeled idealistic. Cross-cultural comparisons of the development of sets and of the manner by which sets are constructed (consuming style) may eventually provide a means to address the concerns of the critics of the mass consumption society. APPENDIX A Consumer Behavior Models Large-System Models Stages of Consumer Behavior Process Modeled Information Perception Preference I'mthasc l'mlpuu'lusc Need Arousal Search l‘omutioo Formation Decision llcluavnm Complete ] Miemdctail “ l Large-System Model * ll 2 Perceptual] Evaluation Models l’artial Microdetail . . . 3 Attitude-l'ormatmn . Models llreadtlt 01 Coverage .__4 Rational-(1min Models _ 5 Stochastic- (1min: Models 1 Macrodctall ‘1 Market-Resixiiisc Models EXHIBIT 7-3 A taxonomy 01 consumer behavlor models: level of deloll versus the consumer processes modeled. Source: Lillien, Gary and Philip Kotler (1983) Marketing Decision Making, New York. Harper & Row p. 205. Figure A-l 130 .NONIMmN .Qa .cofl.mfioomm< wcfluoxumz cmofluwe< ..Uo .uoxoz comm .mEoumxm begun: wCchmzo :fl SECTION I :.uoH>m£wm poxsm mo zuoogfi m mmumaoe: «Amooav zuozm .2 .fi use Uuozo: .< :zow H 005:5 Consumer Bill/('1‘ Bt'lllli'lt’l' School 2.32:; 2.65090 .0 3:02.... 23.5 xumoouou «30.5 new c9555.... .0 30.». 5930.39: .coEco:>cw .2oow 33.6w .n 3:52.23 .e 320335.20 .n 3.5 .~ 5:230 .— o..ooExm 333m . 3:52.»; . _ 5.33390 23m c2565.... 53.383 0. 35.595 1 :0. on new _ r lllll L. l l J . _ a . a L _ u . .om 905622 33.0.2 32.02 3:: ..< . 33.96 3.330 2.63m 2:33:02 " . _ _ « . . 7A.. .2235: _ L. < . _ .anoo . .. _ . . . _ . _ . . . W — . c n V . u 1 . ~11 «ooze; A co...noam_o8a - _ l a a 1L _ . * . q . . . I i l. . . cozcoE. _ _ c2555.... _ . . _ In - co. :8an _ . * . . A p — . . coincom L . u . . _ . _ . A . 20.55:. 1 . . . _ . . 3235a t t 11.. . . 1 a .. fillllilklt IL m::::o . . Pl '- h 1 ' I. '0' --l .11 u. 1.1 J J . o . tl'r""""'f"" ff”’r'yf ’[l/ 44441111544411!!!de A A delta .44/ . . 23on . 93:30 . 3.02:5 . u 05.... . >...mc0m.wa . _ _ . 3.2m . 320 3383a .2255“. .2qu .0 02.259»: mote . x530 . 0:305:05 nocmam n v onu:~>..oc..n.o .n m . 2:9: 8.23% .93.. .o 302... < 4 $50.... Figure A-Z nd Sheth Model How rd a 131 8.833336 . neocoaze. .mcozmazm :o..:u.:. been“. 965 3:923. 9330 «352.5 .e.oom b.3596“. 25.2... 32:) 3302 3.5.0.0220 .3222. =30... :2».on 3353:. 332.2, $063.20 r 5.95 :3.wa 1.. .. .. :o..:m.mm ”masseuse . _ . ._. > 35.385863; 5.5226 3.352? * e 3.38.0: \cofieozoeosoo . w .050 oo.m:_Eoo 5.8.8.2 « toEoE :o..:o..< .....E..m _ 5.30m _ * 539m 352:. W 05390 L. 52.30qu . 8030.“. 1 3000 9:330... an :o.u.ooo :o..~::o.:. .:Q:. .267. 330.5 co.m_ooQ EBA—ECU 2:. S as»... Engel, Blackwell, and Miniard (1986), Consumer Behavior, Source : 5th edition. ’2 .2 e A— Model 1.. EKB F 132 FIGURE I I-4 AN UNDERLYING BEHAVIORAL SYSTEM AS VIEWED FROM A SECMENTATION PERSPECTIVE Luau Person as a person mall Person as a consumer MM Person and the product Lgvgl 4 Person and the brand 1. I" I I I I I I r———-——.~ Source: -———- r l l I I I I I I I John Wiley . . . -. ~ . .'. . ‘u - ,fi -.‘_ v..." . v' .. , ('7‘. . (’3 ‘fi ‘- . ..;‘ ’7 - 'b'TII' afingfxg .':£‘:"v-EI.G’Q; é’le “ f ”"1- IQ: ‘r’ ;‘ 1:: \’ {-1 ‘ '. I ~=;.«I.L-.~,¢;y}‘,.rg. .. "- ~'I- _- " h" A; 3'. . - . _ . . . \« . ~.‘.-~ _ .>' A Demographic f?‘3‘-"I':rlt’,$3595”). -‘.4 - - ‘ ‘- . ,‘ i ;. 1A and '- :-' Fl .' .I . , o-:": ' .| :;.. ”b. r; f'- p . ’ If" 0 r: . _ ‘0‘",‘ch... .131]. i: ‘ .J 1 .02“ I‘,‘ '3 ." ‘_',.'.-}- u . b. . ’ ,— sxloeconom|C?- 'l.”: .." ’J; . . . ;"‘\‘ :' 0‘ '. ..;.:. ”A - "‘. ' descriptions "‘l-‘I‘I "31»: 131-=7- - L1"; Consumer }| .- PFWUCI- -:l 35ch ' , . : _ ’ , u . ”- A12» I , ‘fih ("1% Ide.s[y|e' . Ir beneIlIS ‘ ‘;.rogand 1‘-_ 'r“" a.....‘,u-,‘ ,‘. ‘\ _ /- 3.:i ' .“ .’.c. A , for, "33:? » ... "PAL. ‘ psychograph'C. 45.23; deSIred .‘ {1.3-’7‘. beIIeIS “. I‘L. ‘ l, 'I'. I I r III i; I‘l‘ ' 'p.‘ and ., 2.4"%-~'.fi 2v 1";.:§‘2:13.I.i- and ”.53” a? :c'fls. , 4 ~ . WA": t' - . ,« 5;; ._ 7. -~ , +1. Situation virgngi 7 L- ,3. .;..' percep ions . . . . ._. (f 5., :0! 9. _ '. ._ .n. bags." 5 ._ . . i‘: ”3""“5 “-ii'--'.-I‘.-.; if T5 .:-.--...I , Personality " I”- ~ 1 - .3 _;’;I$r';.1‘.. :. 2. 1.3;.I-3' 3.1-:9“ . .‘ ...,, IfaIIS III 9|.‘d‘.’ -, : z 131‘? ruff. '- -. J; ~32 ,4 1.”.‘3 V ‘ A; ‘ ll. .3 . I :n >3 l‘ '93.. ;.-:.‘"{"'IL- ‘_“ A ;{'.E"l:“: 'tztn" “gJ 125.1" “7 II-“éf'iu‘s. «“fo '- 1 4! £{=-'.'- ‘. ""' l A..;.—I Ii? «awn-2.1-; arm. .. 39:3, " 9.45.3" If“ 49", .;:I:'.~*J;3%E‘fi “1‘?“ ‘4 i‘ <1 :51]: a." Brand .3}. -' ‘ 7w.» u-‘s‘fir: ' at! 1 .-x..-.-,.s . 1* -..~ .-~~o--.. .- ‘ .—A- Z" I” (5"‘5IX9‘II ~I {I.I‘:"§';':' It'5Résqfr'flj‘i'i‘I'." £5. 3i: Bits. (It. .3“ attitude ~- "9"va ; a... "'.::|"‘I"2}H;" a . pfigyggr‘rzll‘gfi 2‘ fi ‘1 ,3: V. . _’J preference—p nan - 55:}: il- ‘2.33.-‘o~£~:~is::>t-v';ax~‘;v£.-i:3 I? r; I: 4r 911:.- «a; , and '- “ms-"hr ' .2; $1.13» .- m- ~. -- .-.-* . 22' ‘ M - '- ?I“-.”i.‘;--E’.'71.31-3:3#’§‘;-"2‘{|9 z? a «‘5. U»? "1 Intentions ’3‘ ewa Irrvufirsfireéfiskae“143%5fl3~« k. ”905“"? '.~I..'1--|%éé?.§ 53025193559 Purchasing environment William L. Uilkie (1986) Consumer Behavior, LmLfi Person as a purchaser ’ I - . 0 o v . .‘, . ., 'II a f , k . o .4 h We. 5‘ "’ "1 . I I '. .Ia ‘ ',1"_. ‘ 31‘s 1 c '. “on -.r I" .’-_ p ,s‘ I ': _I'- ' 3 v F'.' , ‘. N." v.3” .3 ct - I p. ‘.‘..' 3...... ' ‘,'","°‘r¢ "'5' -.‘ :‘é‘ ’u'i o ‘. . “-‘W .‘p‘ ,P--"‘.‘.-’ ‘ I x v, .~ I m.)- « .3-.’.‘ 0! v. "s ‘ :oe..‘.‘A ' I ‘ .‘ .3 , _ n - n Actual purchase " behaviors .04.. .- r ‘1 H'Yg75." 9..., "7'1 a.‘ '51" 1‘?“ “#«l 'o' ":.J. new '-. "@314? ‘3: - Figure A-A 133 Motivation Goal ‘——"" hierarchy I ‘li {* l r—_.‘ Scanner Interrupt . Perceptual and interpretation Attention , —~ . v—IH _—_.. encoding interrupt and mechanisms response i» <> I '[I I InIormation Mfg? H Scanner Interrupt . . . . at . . Processing acquisition ‘ L . and I interpretation cancel“! and External interrupt and ' "' evaluation search mechanisms response I I f Scanner Interrupt Decision and interpretation 3 x _ i-—u i—o processes interrupt and mechanisms response Consumption Scanner Interrupt and A and interpretation I learning ' interrupt and processes mechanisms response Fig. 2.1 The basic structure of the theory. Source: James R. Bettman (1979) An Information Processing Theory of Consumer Choice, Reading, Mass.: Addison wesley Publishing Co. Figure A-S Bettman Information Processing Medel 134 ”CURE 2 AN EXPANDED MODEL OF VALUES, LIFE STYLES. AND CONSUHPTION I I uuuuuuuuuuuuuuuuuuuuu I . ’: er: srvr: ‘ Ixocsuous ' F-------1 r---4£-—-- . , . runes: : runnruau ‘ tusruuuzxrar '., none: as» l___> rtnc use ' , VALuus "" VALuzs “T‘ xxrzxesrs ACTIVITIES | I I ' A I r I I I . I ' r r I ' r I r . ' g ‘V t ' BMW!“ I , . . . are un : I _______ SOCIAL I coasuurrteu. saorrtuc. U‘ c“ "c’°‘5 , tusrrrurraus ' HZDIA :xrosunz PATTIIIS ' I I. I I ' V sacroacouonxc . 'L 0 or Q o . 1‘- 'Amu ' ' ............ . 1 I ' reasouattrv I. ................ gamma-um (- ,iixfi‘, ' r' - 1 3m ozszm pmoucr am: IUTES AND racemes: BENEFITS ‘/ I ' awn Art-mm urn nrrmrorrs r f j --g PURCIIASI I Source: James M. Carman (1978), "values and Consumption Patterns: A Closed Loop," American Marketing Association, p. 405. Figure A-6 Carmen's Closed Loop 135 W W Concrete 22mm ’ 'V Wis. Anflhmfls . Pleasant Llnes Homey Short Warm Long, Cold Curvmg Security Aanes Comfort Patterns Serenity Symmetry Support Assymmetry Soft Check Smooth Circles Dots Enduring P'a'd Lasting Forms Durable Shapes Neutral Proportions Consistency _ Delicate 5'19 _ Clear DECS“)! Mellow Weight Light Height Heavy Depth Dark Width Temperature Absorption Liquidity Solidity Porousness Color Spectmm Texture 33 E? m g E‘ ‘3 '0 .- .C h '— c ‘3'? e S s g s a a a E§._ E “E’ 3E0c‘r3r3'5 g _ .c: b g .93 g E o m o E g r: 32 9- 27$ "‘ 2‘ S C —c: 9520‘ 68033 .9 x6522 E r S ‘65 w = a ‘5 ‘8 - 28 g 8 .C: ..- q; 3 (1) CU —. e. (D — o. < a) ..1 a) > o. o co 0 O o E Figure A-7 More Detail on Consumption Set 136 APPENDIX B Questionnaire Appendix B Questionnaire USEJU£EIKHSE Using the list of components given on page 3. please make up (develop) your idea of the CONTENTS of a “living room.’ Begin by thinking in terms of the components that are needed to build or construct the major. or most prominent items in the living; room set. You will describe the items by checking off spaces provided nest in a list of components. Please TURN TO PAGE 3 and look over the list for a moment. Notice that you should think in terms of components which you can touch (like wood) and components that you cannOt touch or feel (like charm). Another group of people performed a similar task for a set of recreation items. By going through some of their answers. you will see how to do this task. They were given the following empty boxes and list of components to work from. Wood Canvas Rope Rubber Charm Prestige Expressivencss Speed EXAMPLE a! One person filled in the boxes and checked off the components as follows: OGI 0 II \Vood Canvas Rape Rubber X xx” 3 C! it Charm Prestige x Expressiycncss Speed x it ~ X 137 EXAMPLE #2 Another person answered Artist's Spinning Easel Wheel Wood at x Canvas 1: Rope x Rubber Charm 1: Prestige Expressiveness 1: Speed Notice that each person made up a different set using some of the same components. REMEMBER THAT YOUR TASK IS TO CONSTRUCT THE CONTENTS OF A uvmo ROOM usmc THE SPACES ON THE FOLLOWING PAGE. You may use the components in the list as often as you like. If a word in the list does not describe EXACTLY what you mean. DO NOT USE IT. Remember you 19. pp; have to construct the item down to the last string or nail. You are trying to list the item's major components. You must develop the living room based on your present income. NOW, PLEASE BEGIN: I. Fill in the BOX at the top of the column with an item name. Please make sure that you fill in at least five of the boxes. 2. Check off the components that apply. from the list. in the spaces beneath the item name. CHECK OFF NO MORE THAN SEVEN (7) COMPONENTS for each boxed item. PLEASE TURN TO THE NEXT PAGE AND FILL lN THE BOXES AND BLANKS. 138 Wood Stone Fabric Glass Brass Chrome Leather/Suede Wicker Simple Futuristic Cozy Natural Authentic Understated Harmony Classic Charming Dramatic Rustic Tradition Comfort Practical Gracious Distinctive Pretty impeccable 139 As you went through this exercise, did you have an overall picture in your mind of the living room that you created? yes no . Indicate whether the living room was I. your present living room 2. your ideal living room 3. other (please specify) What style did the living room you developed have? (CHECK ONE) I. Colonial 2. Traditional 3. Modern/Contemporarv 4. Victorian 5. French Provincial 6. Country 7. Eclectic __ 8. Other (please specify) THE NEXT TWO QUESTIONS CONCERN YOUR OWN ENTERTAINING HABITS. In which room do you do most of your entertaining? (CHECK ONE) I. living room 2. family room 3. den/library 4. dining room 5. kitchen 6. other For what purpose do you usually use your living room? In the first column, please check off all activities that apply. In the second column. please indicate how often you use your living room for the activities you checked in column one. dividing up TEN points between the activities that you checked in column one. (CHECK ALL THAT APPLY) Write Numbers that ADD UP to TEN entertaining reading watching television eating playing games listening/playing music other $9999.“? 140 \fi THE NEXT SET OF BACKGROUND QUESTIONS ARE FOR CLASSIFICATION PURPOSES ONLY. Please note that there is no way to identify who you are. So please answer the questions as honestly as you can. YOUR ANSWERS HERE AND THROUGHOUT THIS QUESTIONNAIRE WILL REMAIN STRICTLY CONFIDENTIAL. What is your sex? I. male 2. female What is your age? (Check one) I. IS-24 2. 25-34 3. 35-44 4. 45-54 5. . 55-64 6. 65-up What is your present marital status? (CHECK ONE) I. single 2. married 3. living together. not married 4. divorced/separated 5. widowed How many children do you have. LIVING AT HOME with you. in each of the following age categories? ‘ I. under 6 years old 2. 6 - l7 years old 3. IS years and up What is your current FAMILY (household) income before taxes? (CHECK ONE) Under 3 5.000 S 5.000 to less than Sl5.000 515.000 to less than 825.000 325.000 to less than 335.000 335.000 to less than 345.000 345.000 to less than $55,000 $55,000 to less than 365.000 565.000 to less than 575.000 375.000 to less than $85,000 $85,000 to less than 595.000 595.000 and above If greater than $95,000 please specify ' 5 "099°H9959Pr‘ What is the occupation of the head of your household? (If retired. please write “retired“ and the past occupation.) What is the occupation of the second wage earner? (Write N/A if there is only one wage earner; if retired. write ’rclircd" and the past occupation.) 141 What level of education has the head of your household completed? I. Some high school 2. High school graduate 3. Some college 4. 5. ll! College graduate Specify highest Graduate School / degree earned What level of education have you (your spouse) completed? (Write N/A if not applicable). I. Some high school 2. High school graduate 3. Some college __ 4. College graduate __ v Specify highest 5. Graduate school / degree earned THE FOLLOWING ARE SOME QUESTIONS ABOUT YOUR INTERESTS. Please answer the following questions by checking "I" for 'Dcfinitcly Disagrec' through '5' for "Definitely Agree.‘ Definitely Definitely Disagree Agree l 2 3 4 5 I usually have one or more outfits that are the very latest style I try to arrange my home for my children‘s convenience (If no children. please write N/C) I‘d like to spend a year in London or Paris I must admit I really don't like household chores I enjoy going through an art gallery I think I have more self- confidence than most people I would rather spend a quiet evening at home than go out to a party I am an active member of more than one service organization My idea of housekeeping is 'oncc over Iightly“ 142 When I must choose between the two I usually dress for fashion. not for comfort I like to feel that I am part of a group I enjoy going to concerts I am more independent than most people I like parties where there is lots of music and talk I think I have a lot of personal ability I usually read the sports page in the daily paper I take a lot of time and effort to teach my children good habits (if no children. please write N/C) I have personally worked in a political campaign or for a candidate or an issue I like to be considered a leader I like ballet I like to play golf on a regular basis I find cleaning my house an unpleasant task I would rather go to a sporting event than a dance I would like to take a trip around the world THANK YOU FOR YOUR TIME AND Definitely Disagree 143 ATTENTION ! Definitely Agree LIST OF REFERENCES LIST OF REFERENCES Ahtola, Olli T. (1975), ”The Vector Model of Preferences: An Alternative to the Fishbein Model," lgarnal at: Maglgatigg Raggazgh, Vol. XII (February), 52-59. Ajzen, Icek and Martin Fishbein (I977), 'Attitude--Behavior Relations: A Theoretical Analysis and Review of Empirical Research," W Ballatin. Vol. 84, 5_, 888-918. Alderson, Wroe (1957), M rk in B h vi r n Ex iv A i n Homewood, “1.: Richard D. Irwin, Inc. Alpert. Mark I. (1972), ”Personality and the Determinants of Product Choice,” .lgarnal at: Markatiag Researeh, Vol. IX (February), 89-92. Anderson, Paul F. (1986), ”On Method in Consumer Research: A Critical Relativist Perspective,” rn l f n m r h Vol. 13 (September). 155-173. Andreasen. Alan R. (1984), " Life Status Changes and Changes in Consumer Preferences and Satisfaction.” Jamaal 9f ansamag Raseargh. Vol. 11 (December), 784-794. Assael. Henry (1987), angamgr Bahavigr and Markating Agtign, 3rd ed., Boston: Kent Publishing Company. Bagozzi, Richard P. (1982), "A Field Investigation of Causal Relations Among Cognitions, Affect. Intentions, and Behavior." lgarnal 9f Markgtiag 3mm Vol. XIX (November), 562-584. Alice M. Tybout, C. Samuel Craig, and Brian Sternthal (1979), ”The Construct Validity of the Tripartite Classification of Attitudes," Jgaraal Qf Markating Raaaargh, Vol. XVI (February), 88-95. Bandura. A. (1978). ”The Self System in Reciprocal Determinism," Amarigan Win. 33 (April). pp. 344-58. Bass, Frank M. (1972), "Fishbein and Brand Preference: A Reply.” lgarnal Qf Markating Baggageh, Vol. IX (November), 461. Bass, Frank M. and W. Wayne Talarzyk (1972), ”An Attitude Model for the Study of Brand Preference," rn I f M rk in R r h Vol. IX (February). 93-96. 144 145 Belk, Russell W. (1985). ”Materialism: Trait Aspects of Living in the Material World,” WWW Vol. 12 (December), 265-280. (1982), ”Acquiring. Possessing. and Collecting: Fundamental Processes in Consumer Behavior," in W W eds. Ronald F. Bush and Shelby D. Hunt, Chicago: American Marketing Association. (1975), ”Situational Variables and Consumer Behavior,” Jgaznal 91f W (December). 157-164. (1976), ”It’s the Thought that Counts: A Signed Digraph Analysis of Gift-Giving." WW Vol. 3. 3.. 155-162. Kenneth H. Bahn and Robert N. Mayer (1986), ”Developmental Recognition of Consumption Symbolism,“ W 9. Berkman, Harold W. and Christopher Gilson (I986), Qansamar Behaviar. Boston: Kent Publishing Company. Bettman, James R. (1979), An Inf armatign Prgeessiag jljhaggy 9f ansamag thiga. Reading, Mass: Addison-Wesley Publishing Company. H. H. Kassarjian and Richard Lutz (1978), "Consumer Behavior," Ravigw at: Markating 1278. Chicago: American Marketing Association. Blin, Jean-Marie and Joe a. Dodson (I980), ”The Relationship Between Attitudes, Brand Preference. and Choice: A Stochastic View,“ W Vol. 26, 6, (June), 606-619. Boyd, Harper W., Jr. and Sidney J. Levy (1963), ”A New Dimension in Consumer Analysis," HarvarQB psi na§§ Raviaw, Vol. 4] (November-December), 129-140. Burten, Jess K. and Seymour S. Dorner (I941). ”The Wishmaker’s House -- A Coordinated Home-Furnishings Plan,” ,lgarnal 9f Ratailing, Vol. XVII. 2, 43- 48. Boyd, Harper W., Jr. and Sidney J. Levy (1963), "New Dimension in Consumer Analysis,” Harvard Busiaass Raviaw, (November-December), 129-140. Carman, J. M. (1978). ”Values and Consumption Patterns: A Closed Loop," in Adv vaaags ia ggnsamer Raggarah, H. Keith Hunt. ed. Provo, Utah: Association For Consumer Research, pp. 403- 407. Cassady, Ralph, Jr. (I942) ”Discussion,” lgarnal at Magkating. Vol. 4. 1 (April), 119-123. Coleman. Richard P. (1983), "The Continuing Significance of Social Class to Marketing,” Jamaal at: ansamgr Raggarah, 10 (December) pp. 265-280. (1960). "The Significance of Social Stratification in Selling,” in Magkatin ng; A Miami; ag Di sgipljna, Martin L. Bell ed., Chicago: American Marketing Association, pp. 171- 84. I46 Cohen. Joel B. Martin Fishbein and Olli T. Ahtola (I972), ”The Nature and Uses of Expectancy-Value Models in Consumer Attitude Research," Jgar nal at W Vol. IX (November), 456-460. Cooper, Lee G. (I987), ”Do We Need Critical Relativism?” rn l f n m r 85.35.31.913. Vol. 14, 1(June), 126-127. Csikszentmihalyi, M. and E. Rochberg-Halton (I981), WW3; W Cambridge, Eng; Cambridge University Press. Curtis, William W. (1972), ”Social Status or Income?” Jamaal 9f Magkating, Vol. 36(January), 67-68. Darden, William R. and William D. Perreault, Jr. (1975). ”A Multivariate Analysis of Media Exposure and Vacation Behavior with Life Style Covariates.‘ Jgarnal Qt: ansamar Rgsaarga, Vol. 2 (September), 93-103. Douglas, Mary and Baron Isherwood (1978). TM; Harmondsworth, Middlesex, England: Penguin Books Ltd. Easterlin, Richard A. (1973), ”Does Money Buy Happiness?” h P 1i r 30 (Winter) pp. 3-10. Engel, James F.. Roger D. Blackwell and Paul W. Miniard (1986), Qansamgr Bahavigr, 5th ed. Chicago: The Dryden Press. Evans, Franklin B. (1959), ”Psychological and Objective Factors in the Prediction of Brand Choice,” MW Vol. 32 (October), 340-369. "Fanatic Consumer Behavior: Collectors, Cognoscenti, Compulsives, and Other Captives of Deeply Involved Consumption,” (I987), AW m Vol XIV. Provo, UT: Association for Consumer Research. Feller. Robert L. (1987) editor, Artists’ Pigmants; A Hanghagk 9f Thair Histgry aag Charagtgrigtias. Valamg 1. Cambridge University Press and National Gallery of Art. Fenwick, Ian, D. A. Schellinck and K. W. Kendall (1983), ”Assessing the Relaibility of Psychographic Analysis," Markgting Sgignag, Vol. 2, _1_ (Winter), 57-73. Fincher, Jack (1988), ”Let’s Hear it for the Once-Pesky Weed for Russia," Smithsonian. Vol 19.2(May). 88- 94 Firat, Fuat (1986), Unpublished Lecture Material, Association For Consumer Research Meeting, Toronto, Ontario, Canada, October 1986. Fishbein, Martin (1967), "Attitude and the Prediction of Behavior," in mm MW M Fishbein. ed New York: Wiley and lock Ajzen (I974), "Attitudes Toward Objects as Predictors of Single and Multiple Behavioral Criteria," Psyghglggigal Raviaw, Vol. 81, L, 59-74. 147 Fisher, James B. (1987), ”Social Class and Consumer Behavior: The Relevance of Class and Status," in W Melanie Wallendorf and Paul Anderson, eds. Provo. UT: Association for Consumer Research, Vol. XIV, 492-496. Foxall, Gordon R. (1983), W London: The Macmillan Press Ltd. (1975), ”Social Factors in Consumer Choice,” rn I n mm 2 (June) pp. 60-64. Frank, R. E. and P. E. Green (1968), ”Numerical Taxonomies in Marketing Analysis: A Review Article,“ Jaatnal at: Matkating Rasaatgh, Vol. V (February), 83-98. Frankenberg, Ronald (1967), “Economic Anthropology,” in jljhemgs in Eggngmig Aathmmtggy, Raymond Firth, ed., London: Tavistock Publications. Fry, Joseph N. (1971), ”Personality Variables and Cigarette Brand Choice,“ Jamal Qt: Marketiag Raseatgh, Vol. VIII (August), 298-304. Glock, C. Y. and Nicosia, F. M. (1964), "Uses of Sociology in Studying ’Con sumption’ Behavior.” Jamaal Qf Matkgtiag, Vol. 28, (July), pp. 51-54. Green, Paul E., Yoram Wind and Arun K. Jain (1972), "Preference Measurement of Item Collections,” Jgarnal Qf Matkating Rasaargh, Vol. IX (November), 371- 377. R. E. Frank, and P. J. Robinson (1967), "Cluster Analysis in Test Market Selection," Managamgnt Sgiangg, Vol. 13, B-387-400. Hair, Joseph F.. Jr., Ralph E. Anderson. and Ronald L. Tatham (1987), Multivariata Data Analysis, New York: Macmillan Publishing Company. Harrell, Gilbert D. (1986), ansamgr Bahavigr, San Diego: Harcourt Brace Jovanovich, Publishers. and Peter D. Bennett (1974), "An Evaluation of the Expectancy Value Model of Attitude Measurement for Physician Prescribing Behavior," Jgarnal Qt Markating Baagargh, Vol. XI (August), 269-278. Hirschman, Elizabeth C. (1986), "Humanistic Inquiry in Marketing Research: Philosophy, Method, and Criteria,” Jaarnal 9f Markgting Raggargh, Vol. XXIII (August), 237-49. Hirschman, Elizabeth C. (1985), ”Primitive Aspects of Consumption in Modern American Society,“ Jgaraat at: ansnmer Rasaargh, Vol. 12 (September), 142- 154. Holbrook, Morris B. (1987), "What Is Consumer Research?” Jgarnal at: ansamat Em Vol. 14, 1 (June), 128-132. (1981), "Integrating Compositional and Dccompositional Analyses to Represent the Intervening Role of Perceptions in Evaluative Judgments," Jgatnal Qf Markatiag Rgsgarah, Vol. XVIII (February), 13-28. 148 and William L. Moore (1981), “Feature Interactions in Consumer Judgments of Verbal Versus Pictorial Presentations,'JQntna1 at: Qansnmat m Vol. 8 (June), 103-113. Hollander, Stanley C. (1984), Sumptuary Legislation: Demarkcting by Edict,“ WW (Spring). 4-16. (1971), "She ’Shops for You of With You’ Notes on the Theory of the Consumer Purchasing Surrogate," in George Fisk, ed., W Mar gating Thegr ry, Boston: Allyn & Bacon. 218- 240. Howard, John A. and Jagdish N. Sheth (1967), ”A Theory of Buyer Behavior,” in Changing Marketing Systams, Reed Moyer, ed. Chicago: American Marketing Association, 253-262. Inglis, J. and D. Johnson (1970), "Some Observations On, and Developments In, the Analysis of Multivariate Survey Data,” Jnnrnal Qf tha Markat Raggaran My, Vol. 12, 75-8. Jackson, Ralph W., Stephen W. McDaniel and C. P. Rao (1985), ”Food Shopping and Preparation: Psychographic Differences of Working Wives and Housewives," Jgnznal Qf ansnmgr Rasaargh, Vol. 12 (June), 110-113. Jacoby, Jacob (1976), ”Consumer Research: Telling It Like It Is," in Advangag in W Vol. 3, B. B. Anderson, ed. Atlanta: Georgia State University, 1-11. Donald E. Spellcr and Carol A. Kohn (1974), "Brand Choice Behavior as a Function of Information Load: Replication and Extention,” mm: ananmar Raaaargn, Vol. 3 (March), 33-42. Jones, Marshall R. (1955), Nebtaslta Symngsinm an Mgtivatinn, Lincoln, NE: University of Nebraska Press. Kassarjian, Harold H. (1987), ”How We Spent Our Summer Vacation: A Preliminary Report on the 1986 Consumer Behavior Odyssey,” in navanees in W Melanie Wallendorf and Paul Anderson, eds. Vol. XIV, Provo, Utah: Association for Consumer Research, 376-377. (1978), ”Presidential Address 1977: Anthropomorphism and Parsimony,‘ in W Vol.5, Chicago: Association for Consumer Research, xiii-xiv. and Mary Jane Sheffet (1981), "Personality and Consumer Behavior: An Update," in Eatsngatjvgs in ansnma; Bahavigr, H. H. Kassarjian and T. S. Robertson, eds., Glenview, 111.: Scott Foresman, 160-180. Kernan, Jerome B. (1987), ”Chasing the Holy Grail: Reflections on ’What is Consumer Research?" Jgnrnal 9f Qansnmar Rasaargh, Vol. 14, 1 (June), 133- 135. Kinnear, Thomas C. and James R. Taylor (1976), ”Psychographics: Some Additional Findings," Jgntnal Qt: Markgting Rasearah, Vol. XIII (November), 422-425. 149 Kirkpatrick, Jerry (1983), ”Theory and History in Marketing,” W We. Vol 4. 1.. 44- 49 Kluckhohn, Clyde (1951), ”Values and Value Orientations in the Theory of Action: An Explanation in Definition and Classification,” in Inward a Qanatal r A i T. Parsons and E. Shills, eds.. Cambridge: Harvard University Press, 388-431. Kotler, Philip (1988), Want, Englewood Cliffs, NJ: Prentice- Hall. (1986), Pringiplgs 9f Matltgtjng, 3rd ed., Englewood Cliffs, NJ: Prentice-Hall, Inc. (1965), "Behavioral Models for Analyzing Buyers.” 1 rk Vol. 29 (October), 37-45. Kunkle, John H. and Leonard L. Berry (1968), ”A Behavioral Conception of Retail Image,” Jdnznat at: Marketing. Vol. 32 (October), 21-27. Lancaster. Kelvin J. (1971). W New York: Columbia University Press. (1966a), "A New Approach to Consumer Theory," 11 li' Lenngnay, Vol. 74, a, 132-157. (1966b), "Change and Innovation in the Technology of Consumption," Amagigan Egdndm mi; Raviaw, Vol. LVI, 2 (May), 14- 23. Laroche, Michel and Jacques E. Brisoux (1981), ”A Test of Competitive Effects in the Relationship among Attitudes and Intentions," in Ih_e__(;1_la_ngjng WW 1981 Educators? Conference Proceedings, Kenneth Bernhardt, ct. al., editors. Chicago: American Marketing Association. Lastovicka, John L. (1982), "On the Validation of Lifestyle Traits: A Review and Illustration, WWW Vol. XIX (February). 126- 138. Lauer, Jeanette C. and Robert H. Lauer (1979), "The Language of Dress: A Sociohistorical Study of the Meaning of Clothing in America,” W Bavigw 9f Amatigan Stadigs, Vol. 10, 3 (Winter), 305-323. Laumann, Edward O. and James S. House (1970), ”Living Room Styles and Social Attributes: The Patterning of Material Artifacts in a Modern Urban Community." Wm 54 (April). 9. 326. Lee, Dorothy (1959), ”Culture and the Experience of Values," in A. H. Maslow, ed., Naw W New York: Harper and Row. Levy, Sidney J. (1966), "Social Class and Consumer Behavior,” in Joseph W. Newman, ed. Qn [gnawing tha ansnmgt, New York: John Wiley and Sons, pp. 146-60. 150 __ (1963). ”Symbolism and Life Style." W Stephen A. Greyser, ed., Chicago: American Marketing Association. Lilien, Gary L. and Philip Kotler (1983), W New York: Harper and Row, Publishers. Lutz, Richard J. (1977), "An Experimental Investigation of Causal Relations Among Cognitions, Affect, and Behavioral Intention,“ W Rm Vol. 3 (March), 197-208. Machlup. Fritz (1963). W Enslcwood CliffS. NJ: Prentice-Hall Inc. Mazis, Michael B., Olli T. Ahtola, and R. Eugene Kippel (1975), "A Comparison of Four Multi-Attributc Models in the Prediction of Consumer Attitudes,” Vol. 2 (June), 38-52. McAlister, Leigh (1979), "Choosing Multiple Items from a Product Class," M1131 at: Qensnmer Researeh. Vol. 6 (December), 213-224. McAlister, Leigh (1982), "A Dynamic Attribute Satiation Model of Variety-Seeking Behavior, Jengnal at: Censnmer Researen, Vol. 9 (September), 141-150. McAlister, Leigh and Edgar Pessemier (1982), ”Variety Seeking Behavior: An Interdisciplinary Review," W Vol. 9 (December), 311-322. Mittelstaedt, Robert A. (1971), "Criteria for a Theory of Consumer Behavior," in We; Robert J. Holloway, Robert A. Mittelstaedt and M. Vcnkatesan, Boston: Houghton Mif f lin Company, 8-13. Mochis, George P. (1981), ”Socialization Perspectives and Consumer Behavior," in mm eds. Ben M. Enis and Kenneth J. Roering, Chicago: American Marketing Association. (1976), "Social Comparison and Informal Group Inf lucnce," Jentnal Qt: Marketing Researeh, 13 (August), pp. 237-244. Molloy. John T. (1977). W New York: Warner Books. Morrison, D. G. (1967), ”Measurement Problems in Cluster Analysis,'_Manage_rnent Sejenee, Vol. 13, B-775-80. Mukerji, Chandra (1978), "Artwork: Collection & Contemporary Culture,” WW 84. pp. 348-365. Myers, J. H. and Jonathan Guttman (1974), ”Life Style: The Essence of Social Class.” in Lille Style and Beyenegranhies, William D. Wells, ed., Chicago: American Marketing Association, 235-256. Myers, J. H., R. Stanton and A. Haug (1971), ”Correlates of Buyer Behavior -- Social Class Vs. Income,” Jenrnal elj Marketing. 35, pp. 8-15. 151 Nakanishi, Masao and James R. Bettman (1974), ”Attitude Models Revisited: An Individual Level Analysis,” W Vol. 1 (December), 16-21. Nicosia, Francesco M. and Robert N. Mayer (1976), ”Toward a Sociology of Consumption,” Jenrnal at: Cenenme; Reeearell, Vol. 3 (September), 65-75. Nord, Walter R. and J. Paul Peter (1980), "A Behavior Modification Perspective on Marketing." We. 44 (Spring). pp. 36-47. Olashavsky, Richard W. and Donald H. Granbois (1979), ”Consumer Decision Making -- Fact or Fiction?" Jentnal Qt: Cenennle; Researen, Vol. 6 (September), 93-100. Peak, Helen (1955), “Attitude and Motivation." in W Metljtjdn, Marshall R. Jones. ed., Linclon, NE: University of Nebraska Press. Peterson, Robert A., George Albaum and Richard F. Beltramini (1985), ”A Meta- Analysis of Effect Sizes in Consumer Behavior Experiments.” LaumaLQE Wren. Vol. 12 (June). 97-103. Pollay, Richard (1986), Unpublished Lecture Material, Association For Consumer Research Meetings, Toronto, Ontario, Canada, October 1986. Punj, Girish and David W. Stewart (1983), "Cluster Analysis in Marketing Research: Review and Suggestions for Application,” W m Vol. 20 (May), 134-148. Rainwater, Lee (1974), W New York: Basic Books. Rassuli, Kathleen M. and Stanley C. Hollander (1986), ”Desire -- Induced, Innate. Insatiable?” WW2. Vol. 6. 2 (Fall). 4-24. Reibstein, David J. (1978), ”The Prediction of Individual Probabilities of Brand Choice," Jenrnal ef genenmer Researel], Vol. 5 (December), 163-168. Reynolds, Fred and William Darden (1974), "Construing Life Style and Psychographics,‘ in Life Style and Psyehegraphiee. William D. Wells, ed., Chicago: American Marketing Association. Riesman, David (1950), jljhe Lenely Crewd, Yale University Press. Rook, Dennis W. (1985), ”The Ritual Dimension of Consumer Behavior,” Jenrnal ef W Vol. 12 (December), 251-264. Rosenberg, Milton J. (1956), "Cognitive Structure and Attitudinal Affect Jentnal LAW (November). 367- 372 Ryan, Michael, J. and E. H. Bonfield (1980), ”Fishbein’s Intentions Model: A Test of External and Pragmatic Validity.” Wag, Vol. 44 (Spring), 82-95. 152 Saranson. Irwin 6., Ronald E. Smith, and Edward Diener (1975), "Personality Research: Components of Variance Attributable to the Person and the Situation." MW Vol 32 (February). 199- 204. Schaninger, Charles M. (1981), "Social Class Versus Income Revistied: An Empirical Investigation." W 18 (May). pp. 192-208. Sheth, Jagdish N. (1979), "The Surpluses and Shortages in Consumer Behavior Theory and Research." WWW Vol. 7, a, (Fall), 414-426. (1972), "Reply to Comments on the Nature and Uses of Expectancy- Value Models in Consumer Attitude Research," 11 k i Regain, Vol. IX (November), 462- 465. and Dennis E. Garrett (1986), Marketing jljheety, Cincinnati: South- Wcstern Publishing Co. and W. Wayne Talarzyk (1972), "Perceived Instrumentality and Value Importance as Determinants of Attitudes," MW Vol. IX (February), 6- 9. Sirgy, Joseph (1985), "A Conceptualization of the Consumer Behavior Discipline," Jenrnal ef the Aeademy ef Marketing Seienee, Vol. 13, 1 (Winter), 104-121. Speece, Deborah L., James D. McKinney and Mark I. Appelbaum (1985), "Classification and Validation of Behavioral Subtypes of Learning-Disabled Children," Jenrnal ef Edneatienal Peyenelegy, Vol. 77, _l_, 67-77. Solomon, Michael R. (1983), "The Role of Products as Social Stimuli: A Symbolic Interaction Perspective," Jentnal elf gensnmer Reeearen, Vol. 10 (December), 319-329. Stewart-Abernathy, Leslie C. (1986), "Household Consumption in an Ozark Mountain Farmstead Before the Great War: An Archeological Perspective on Consumption and Cognition in an Agricultural Hinterland," Lecture given at Association for Consumer Research Meetings. Toronto, Ontario, Canada, October 1986. Talarzyk, W. Wayne (1972), "A Reply to the Response to Bass, Talarzyk, and Sheth," Jenrnal ef Marketing Researeh, Vol. IX (November), 465-467. Urban, Glen L. and Gerald M. Katz (1983). "Pre-Test-Market Models: Validation and Managerial Implications," Jenrnal ef Marketing Researeh, Vol. XX (August), 221-234. Uusitalo, Liisa and Jyrki Uusitalo (1981), "Scientific Progress andR Research Tradi- tions in Consumer Research," in Vol. 8. Kent B. Monroe ed., Ann Arbor, MI:v Association for Consumer Research, pp. 559-563. 153 Wells, William D. and Douglas J. Tigert (1971), "Activities, Interests and Opinions" WV01 11 (August). 27 34 Wells, William D. (1975). "Psychographics: A Critical Review." Jenznal elj Maw V01 XII (Malt). 196- 213 Westbrook, Robert A. and Claes Forncll (1979), "Patterns of Information Source Usage Among Durable Goods Buyers," MW 16 (August), pp. 303- 312. Wilkie, William L. and Edgar A. Pessemier (1973), "Issues in Marketing’s Use of Multiattribute Attitude Models, WWW Vol. (November), 428-441. Wilkie, William L. (1986), Wallet; New York: John Wiley and Sons. Wind, Jerry (1977), "Toward a Change in the Focus of Marketing Analysis: From a Single Brand to an Assortment." Jenrnal elf Marketing. (October), pps. 12, 143. and Paul Green (1974), "Some Conceptual, Measurement, and Analytical Problems in Life Style Research," in WEEKS. William D. Wells, ed., Chicago: American Marketing Association. Woodson, Larry G., Terry L. Childers and Paul R. Winn (1976), "Intergenerational Influences in the Purchase of Auto Insurance," in W. Locandcr (cd.), Marketing Leelg Qntwatd; 127e Easiness Preeeedings (Chicago: American Marketing Association, 1976), 43-49. Wynn, George W. (1982), "Consumer Behavior Models: An Introduction," in Qensnmer Behavier, James U. McNeal and Stephen W. McDaniel, cds., Boston: Little, Brown, 29-37. Zajonc, Robert B. (1980), "Feeling and Thinking: Preferences Need No Inferences," W Vol. 35 (February), 151-175. and Hazel Markus (1982), "Affective and Cognitive Factors in Pre- f ercnces," MW Vol. 9 (September), 123-131. Zielinski, John and Thomas S. Robertson (1981), "Consumer Behavior Theory: Excesses and Limitations," in Advanees in Qensntnet Researen IX, Andrew Mitchell, ed. (St. Louis, MO: Association for Consumer Research, 1981), 8- 12. General References Aaker. David, A. Richard P. Bagozzi, James M. CarmAn, and James M. MacLachlan (1980), "On Using Response Latency to Measure Preference," MW Vol XVII (May). 237- 44 154 Ajzen. lock and Martin Fishbein (1980). WWW 851131811; Englewood Cliffs, NJ: Prentice-Hall, Inc. Ajzen, Icek and Martin Fishbein (1973), "Attitudinal and Normative Variables as Predictors of Specific BehaviorS" W W Vol 84 5.. 41 -57 Arensberg, Conrad M. and Arthur H. N1ehof f (1980), "American Cultural Values," in James P. Spradley and Michale A. Rykiewich, eds., jIjhe Neeireme; Readings eg Ameri r111ee Celter re, Goston: Little, Brown and Company, pp. 363- 379. Bagozzi, Richard P. (1984), "A Prospectus for Theory Construction in Marketing," Wing. 48 (Winter). pp. 11-29. Bandura, A. (1977), W Englewood Cliffs, NJ: Prentice-Hall. Becker, Gary S. (1974), "A Theory of Social Interaction," rn P Ii ' l Esteem (Nov./Dec.), 82, pp. 1063-93. and Robert T. Michael (1976), "On the New Theory of Consumer Behavior," in Eeenemie Apereeeh te Human Behaviet, ed. Gary S. Becker, Chicago: University of Chicago Press, pp. 131-49. Becherer, Richard C., Fred W. Morgan, and Lawrence M. Richard (1979), "Person- Situation Interaction within a Consumer Behavior Context," Jeernel efi 12mm 102 (July). pp. 235-242. Bem, D. J. (1965), "An Experimental Analysis of Self- Persuasion" W MW Vol 1 p 199 Bengston, V. L. and M. C. Lovejoy, (1973), "Values, Personality, and Social Structure" WM 16 (July/August) pp 880- 912 Blumberg, Paul (1974), "The Decline & Fall of the Status Symbolism thoughts on Status in a Post-Industrial Society, W 21, pp. 490-498. Burns, Alvin C. and Mary Carolyn Harrison (1979), "A Test of the Reliability of Psychographicsf Wham Vol. XVI (February). 32-38. Campbell, D. T. (1963), "Social Attitudes and Other Acquired Behavioral Disposi- tions," in Sigmund Koch ed., muggy, Vol. 6, New York: McGraw-Hill. Canter, D. and C. Kerry (1982), "Approaches to Environmental Evaluation," Inter getienal Reviewef Applied Pleyehe egy, Vol. 31, No. 2, p. 147. Cantril, Hadley (1965), jljhe Pattern ef fleman Ceneerge, New Burnswick, NJ: Rutgers University Press. Coleman, Richard P. and Bernice L. Neugarten (1971), Seeial States in the City, San Francisco: Jossey-Bass, Inc. and Lee P. Rainwater, with Kent A. McClelland (1978), mm W New York Basic Books 155 Davis, J. (1972), "Gifts and the UK. Economy," Mag, 7(3), pp. 408-429. Day, George S. (1979), MW New York: Free Press. DeFleur, M. L. and F. R. Westie (1963), "Attitude as a Scientific Concept," Seeiai EQLees, Vol. 42, No. 1, pp. 17-31. Dichter, Ernest (1985), "What’s in an Image," WW 2 (Winter), pp. 75-81. (1960), jljiie Sttategy ef Desire, New York: Doubleday. Duescnberry, J. (1949), v' n v' r Cambridge, Mass: Harvard Business Press. Durant, Will and Ariel Durant (1968), W New York: Simon and Schuster. Emshoff, J. E. (1971), Analysis ef Behavierei Syetems, New York: MacMillan Publishing Co., Inc. Fedor, Donald B. and Gerald B. Ferris (1981), "Integrating OB Mod with Cog- nitive Approaches to Motivation." W 6 (January), pp. 115-125. Felson, Marcus (1976), "The Differentiation of Material Lifestyles: 1925-1966," Wan. 3. pp. 397-421. Gardner, Meryl Paula (1985), "Mood States and Consumer Behavior: A Critical Review," Ween, Vol. 12 (December), 281-300. Gensch, Dennis H. and Wilfred W. Recker (1979), "The Multinomial, Multiattribute Logit Choice Model," W33, Vol. XVI (February), 124-132. Gerth, H. and Mills, C. W. (1948), r r' E i i1 , London: Routledge and Kegan Paul. Gilbert, Dennis and Joseph A. Kahl (1982), ' ° A N w We, Homewood, IL: The Dorsey Press. Gordon, Jean and Jan McArthur (1986), "Interior Decorating Advice as Popular Culture: Women’s Views Concerning Wall and Window Treatments, 1870- 1920." 121W V01. 9. 3. 15-23. Gordon, Milton M. (1947), "The Concept of the Subculture and Its Applications," Seeiflfleeee, 26 (October) pp. 40-42. Gutman, J. (1982) "A Means-End Chain Model Based on Consumer Categorization Processes, Marketing, 46 (1). pp. 60-72 Hamm, B. Curtis and Edward N. Cundiff (1969), "Self Actual and Production Perception, r 1 rk in R r h 6 (November), pp. 470-472. 156 Harris, Marvin (1968), WW1, New York: Harper & Row. Hauser, John R. (1986), "Agendas and Consumer Choice," MW Beeeaeeii, Vol. XXIII, (August), 199-212. Hauser, John R. and Steven M. Shugan (1983), "Defensive Marketing Strategies," Marketinsjsience. V01. 2. .4. (Fan). 319-360. Henry, Walter A. (1976), "Cultural Values Do Correlate with Consumer Behavior," Jeatgal 91' Marketing Reseateh, 13, 121-7. Hirschman, Elizabeth C. (1980), "Innovativeness, Novelty Seeking, and Consumer Creativity," W Vol. 7 (December), 283-295. Holbrook, Morris B. and Neville C. Hughes (1978), "Product Images: How Structured Rating Scales Facilitate Using a Projective Technique in Hypothesis Testing," jijhe Jeernal eff Esyehelegy, 100, pp. 323-328. Holman, Rebecca H. (1981), "Product Use as Communication: A Fresh Appraisal of a Venerable Topic," Review eij Marketiag 1281, Ben M. Enis and Kenneth J. Roering, eds., Chicago: American Marketing Association. Homans, George (1961), Seeial Behavier; It§ Eiementaty Fetme, New York: Harcourt. Horton, D. and T. Dixon (eds) (1968), V 1 vi - B h vi r T Englewood Cliffs: Prentice-Hall Inc. Howard, John and Arch G. Woodside (1974), "Personal Values Affecting Consumer Psychology," in Robert E. Pitts and Arch G. Woodside, Eeteenal Values and W Lexington, Mass.: Lexington Books, pp. 3-12. Kelly. G. A. (1955). W New York: Norton. Levy, Sidney J. (1981), "Interpreting Consumer Mythology: A Structural Approach to Consumer Behavior," Mam Vol. 45 (Summer), 49-61. McLeod, Jack M. and Garrett J. O’Keefe (1972), "The Socialism Perspective and Communication Behavior in F. G. Kline and P. J. Tichenor (eds) Curr rrent WWW pp 121-168. Merton, Robert K. (1968), Seeial Theery anti Seeial Streetare, 3rd ed., New York: The Free Press. Miller, Kenneth E. and James L. Ginter (1979), "An Integration of Situational Variation in Brand Choice Behavior and Attitude," JennaLeLMaLketing Reaeaeeh, Vol. XVI (February), 111-123. Miniard, Paul W. and Joel B. Cohen (1983), "Modeling Personal and Normative Influences on Behavior," Jeeraal e1: Ceneemer Reeeareh, 10 (September), pp. 169-180. 157 Munson, J. Michael and Shelby H. McIntyre (1979), "Developing Practical Procedures of the Measurement of Personal Values in Cross-Cultural Marketing." WWW 16. no pages? Murdoch, George P. (1945), "The Common Denominator of Cultures," in Ralph Linton (cd.), jijiie Seienee pf Man ip the Werlti Crisis, New York: Columbia University Press, 123-142. Myers, J. H. and J. Mount (1973), "More on Social Class vs. Income as Correlates of Buyer Behavior," Wag, 37, pp. 71-73. Noerager, Jon P. (1979), "An Assessment of CAD: A Personality Instrument Developed Specifically for Marketing Research," W M (February). pp 53- 59 Ostrom, Thomas M. (1969), "The Relationship Between the Affective, Behavioral, and Cognitive Components of Attitude," ,ieprnai pf Experimental Seeial Psyehelegy, Vol. 5, 12-30. Peter, J. Paul (1980), "Some Observations on Self-Concept in Consumer Behavior," in Jerry C. Olson, ed., A v n in n m R r h Vol. 7, A n n Arbor, MI: Association for Consumer Research, pp. 625-626. Petrinovich, L. (1979), "Probabilistic Functionalism: A Conception of Research Method," Ameriean Peyehelpgiet, 34 (May), pp. 373-390. Pirie, M. C. (1960), "Marketing and Social Class: An Anthropologist’s View," Management Review, 49 (September) pp. 45-48. Pitts, Robert E. and Arch G. Woodside (1984), "Personal Values and Market Segmentation: Applying the Value Construct," in Robert E. Pitts and Arch G. Woodside. WW Lexington. Mass.: Lexington Books, pp. 55-67. Poe, Donald (1977), "The Giving of Gifts: Anthropological Data and Social Psychological Theory," Cerpeli Jeernai pf Seeial Relatiene, 12(1), pp. 47-63. Prelinger, Ernst (1959), "Extention & Structure of the Self, Jeprnal pf Peyehelegy, 47, pp. 13-23. Prown, Jules David (1982), "Mind in Matter," Wintetthpr Pettiplie, Vol. 17, i, 1- 19. (1980), "Style as Evidence," Winterthp; Eertfeiip, Vol. 15, 3, 197-210. Reynolds, Thomas J. and Jonathan Gutman (1984), "Advertising as Image Manage- ment," leprnal pi Advertising Reeeareh, 24 (February/March), pp. 27-36. Rokeach, Milton (1973), The Nature pf Heman Valpes, New York: The Free Press. 158 Rothschild, Michael L. and William C. Gaidis (1981), "Behavioral Learning Theory: Its Relevance to Marketing and Promotions," W 45 (Spring), pp. 70-78. Sampson, R. (1972) "Using the Repertory Grid Test," Ilettinal pf Marketipg 3mm IX (February). Seibold, D. R. (1980), "Attitude - Verbal Report - Behavior Relationships as Causal Processes: Formalization, Test, and Communication Implications in Cushman and McPhee edS. WW2. New York: Academic Press. Sherif, Muzafer and Carolyn Sherif (1963), "Varieties of Social Stimulus Situa- tions," in im 1 D rmin n f h vi ed. S. B. Sells, New York: Ronald Press, pp. 82-106. Sherry, John F. Jr. (1983), "Gift Giving in Anthropological Perspective," 1911:1131 pi: Censpmer Beeeareh, Vol. 10, 157-168. Simon, H. (1957), Mpdela pf Man, New York: John Wiley and Sons, Inc. Sirgy, M. Joseph (1982), "Self-Concept in Consumer Behavior: A Critical Review," leptnal ef Censpmer Researeh, 9 (December), pp. 287-300. Skelly, Florence R. (1979), "Emerging Attitudes and Life Styles," paper presented to American Association of Advertising Agencies Annual Meeting, White Sulphur Springs, West Virginia, May 16-19, 1979. Snow, R. (1974), "Representative and Quasi-Representative Designs for Research on Teaching," Review pf Edpeatipnal Reeeareh, 44 (Summer), pp. 265-291. Vinson, Donald and J. M. Munson (1976), "Personal Values: An Approach to Market Segmentation," in Kenneth Bernhardt, ed., M rk in ' 177 -1 7 Beyengi, Chicago: American Marketing Association, pp. 313-317. Jerome E. Scott and Lawrence M. Lamont (1977), "The Role of Personal Values in Marketing and Consumer Behavior," ,lpptiial pi: Matketing, 41 (April), pp. 44-50. Ward, Scott (1978), "Contribution of Socialization Theory to Consumer Behavior Research, Ameriean Behaviet Seientiet, 21 (March-April). (1974), "Consumer Socialization," nl f n m r r h 1 (September), pp. 1-16. Warshaw, Paul R. (1980), "A New Model for Predicting Behavioral Intentions," ,lpprnal pf Matketing Researep, Vol. XVII (May), 153-73. Wind, Yoram (1969), "Incongruency of Socioeconomic Variables and Buying Behavior," in Philip R. McDonald, cd. i M rk in A ' W Series No. 30. pp. 362-67. (1978), "Issues and Advances in Segmentation Research," Jeernai pt: MW 15 (August). pp 317- 37