Cognitive Structure ACRoss Consumer Ethnic Subcultures: a Comparative Analysis

Elizabeth C. Hirschman, New York University
ABSTRACT - This research examines variation in cognitive structure across six consumer ethnic subcultures: Chinese, English, Greek, Irish, Italian and Jewish. Patterns of cognition typifying each group are compared for four consumption-salient constructs: novelty seeking, information transfer, divergent processing ability, and consumption motives. Two factor comparison procedures are used to assess inter-group structural congruence. It is found that a wide range of inter-group congruence is present; in some cases consumers in different ethnic subcultures have very dissimilar cognitive structures; while in other instances strong similarity is present. Implications for construct generalizability across ethnic subcultures are discussed.
[ to cite ]:
Elizabeth C. Hirschman (1983) ,"Cognitive Structure ACRoss Consumer Ethnic Subcultures: a Comparative Analysis", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 197-202.

Advances in Consumer Research Volume 10, 1983      Pages 197-202

COGNITIVE STRUCTURE ACROSS CONSUMER ETHNIC SUBCULTURES: A COMPARATIVE ANALYSIS

Elizabeth C. Hirschman, New York University

ABSTRACT -

This research examines variation in cognitive structure across six consumer ethnic subcultures: Chinese, English, Greek, Irish, Italian and Jewish. Patterns of cognition typifying each group are compared for four consumption-salient constructs: novelty seeking, information transfer, divergent processing ability, and consumption motives. Two factor comparison procedures are used to assess inter-group structural congruence. It is found that a wide range of inter-group congruence is present; in some cases consumers in different ethnic subcultures have very dissimilar cognitive structures; while in other instances strong similarity is present. Implications for construct generalizability across ethnic subcultures are discussed.

INTRODUCTION

The concept of ethnicity has played an important role in sociology (Glazer and Moynihan 1975) and social psychology (Ember 1977), yet has received relatively less attention in consumer research (Hirschman 1981). Investigation of ethnicity within a consumption context has generally focused upon race as the primary dimension or behavioral differentiation (e.g., Solomon, Bush and Hair 1976). Despite the number of published articles concerned with the influence of racial ethnicity on consumer behavior (c. f ., Zaltman and Wallendorf 1979; Engel, Blackwell, Kollat 1978 for reviews), this research stream suffers from several inadequacies. Among these are the failure to investigate dimensions of ethnicity other than the black/white racial dichotomy, the failure to use emic measures of ethnic affiliation, and the failure to take into account variation in social norms when examining group differences (Hirschman 1981).

These shortcomings are likely attributable to at least two causes: First, inadequate comprehension of the social origins of ethnicity; and second, a lack of researcher knowledge about how ethnic groups may function as consumer subcultures. Each of these issues is now examined more closely.

Origins of Ethnicity

Ethnicity, as presently conceptualized within social science literature may originate from three sources: (1) race, (2) religion, and (3) nationality (Glazer and Moynihan 1975). Ethnicity therefore implies three possible sources of social commonality among people. First is that of common descent (Weber 1952); individuals may perceive an ethnic bond based upon their possession of shared racial characteristics. Hence, American Indians and American Blacks may each serve as ethnic groups due to the common racial heritage of their members. A second source of ethnic commonality is a set of shared metaphysical beliefs and values primarily promulgated via religion. Religious affiliation provides the individual not only with a certain form of spiritual ritual, but also with ethical criteria and a general world-view (Gurvitch 1973, Russell 1945). A third potential source of social commonality is that of shared history (Cohen 1977). This form of ethnicity is especially salient among nationality groups. Persons having the same nationality typically are the product of common set or historic events: these events produce an ethnic bond based upon a sense of a shared past and future.

ETHNIC GROUPS AS CONSUMER SUBCULTURES

Ethnicity may be brought into closer proximity with consumer research by conceptualizing ethnic groups as consumer subcultures. A subculture is a group of people who, while sharing some traits in common with the surrounding culture (e.g., language), may be differentiated from it by their beliefs, symbols, and/or material artifacts (Jorgensen 1979). Members of a subculture are identifiable as members of the general culture, but additionally possess certain characteristics by which they may be classified into a distinct-category. One major proposition is that ethnic groups, as subcultures, constitute differentiable consumer segments. That is, ethnic groups may represent substantially distinct patterns of consumption characteristics.

PURPOSE

One of the most fertile areas for examining differences among consumer segments is the cognitive structure underlying consumption processes (Bettman 1979, Olson 1981). For example, consumers in various segments may utilize different attributes to characterize a product, may possess more/less complexity with regard to certain consumption domains, or may aggregate products into different categorical frameworks (Shocker, Day, Srivastava 1979). The present research compares the cognitive structures---of consumers from six nationality subcultures on four consumption constructs: product novelty seeking, product information transfer, divergent processing ability, and consumption motives. The first two or these constructs--novelty seeking and information transfer--were selected because prior research (Hirschman 1982) has shown them susceptible to religious affiliation differences. Hence it was reasoned that nationality differences might also be present. Divergent processing ability has been posited to be a unidimensional construct, insusceptible to subcultural variation (Horn 1976); hence the present research examines the validity of that proposition in a nationality subculture context. The final construct, consumption motives, has to the author's knowledge,never been examined in terms of its cross-subcultural similarity. Hence the present study represents an initial test of its generalizability.

The efficacy of subcultural nationality as a device for differentiating consumers is virtually unexplored, yet logically would seem to possess potential value. Several researchers have discerned important differences in consumption style on a cross-national basis (c.f. Douglas and Craig 1981); thus it is straightforward-to deduce that such differences may also be present among nationality-based subcultures. The ethnic subcultures examined in the present research are: Chinese, English, Greek, Irish, Italian, and Jewish.

Culture and Cognition

One way of assessing the similarity of two or more social groups is by measuring the similarity or cognitive structure they possess concerning the same stimulus. For example, to the extent that consumers in two subcultures organize cognitive content into dissimilar structures, they may possess dissimilar subjective realties (Triandis 1979; Szalay and Deese 1978). Deese, in a lengthy set of investigations (Deese 1965, 1969, 1979, 1974), has demonstrated that the cognitive structures characterizing individuals in various cultures and subcultures reflect differences in their organization of reality. Since consumption knowledge is a subjective phenomenon, then differences in cognitive structure represent differences in culture. Therefore, to the extent that members of one nationality group possess cognitive structures unlike those possessed by members of another group, they reside in separate consumption subcultures.

METHOD

Sample

A mailing list for New York City S.M.S.A. households having annual incomes greater than $25,000 and professional-managerial occupational status for the household head was purchased from a national marketing research firm. A total of 700 5-page questionnaires was mailed to listed households. The sponsor of the survey was identified as New York University. No monetary incentive was offered; respondents were told that survey results would be used for academic research. A stamped, return-addressed N.Y.U. envelope was included with the questionnaire. After three weeks 286 questionnaires were returned. A follow-up mailing yielded an additional 195 questionnaires. There appeared to be no systematic biases in the first and second wave of collected questionnaires, based on a comparison of socioeconomic data and marital status.

Within the sample were 33 persons identifying their subcultural nationality as Chinese; 41 persons identifying themselves as English [This option was listed on the questionnaire as WASP (English); in the present manuscript this group of respondents is termed English. as this is the more precise nationality label.]; 29 persons identifying themselves as Greek; 31 persons identifying themselves as Irish; 74 as Italian, and 172 as Jewish. The remaining 31 persons designated their nationality affiliation as Black, German, Hispanic, Japanese, Polish or stated that they had no nationality affiliation. Because the numbers in these latter nationality groupings were so few, these 31 persons were excluded from subsequent data analysis.

A simultaneous comparison of group means among the Chinese, English, Greek, Irish, Italian, and Jewish respondents revealed no significant differences across socioeconomic, life cycle, or age variables (Scheffe procedure, p<.15). By comparing the responses of consumers who differ only in the primary criterion of interest (nationality), but who are otherwise socially similar, we are able to draw stronger inferences from the data. Differences in the cognitive structures of Chinese, Greek, English, Irish, Italian, and Jewish consumers are more confidently attributed to their nationality, when factors such as socioeconomic status and life cycle stage are neutralized via use of an internally consistent sample. Thus, the generalizability of the findings is limited to similarly constituted populations, but internal validity is substantially enhanced (Calder, Phillips and Tybout 1980, Cook and Campbell 1975).

Measurement of Ethnicity

Because ethnic nationality was the central construct in this research, attempts were made to operationalize it using an emic measure. An emic measure of national ethnicity is one which permits the individual to ascribe nationality identification to him/herself. It is based on the individual's self-perceptions and not on the perceptions of the researcher, which may be biased by ethnocentrism. This method of measuring ethnicity is the approach deemed most appropriate by cross-cultural behavioral researchers, and especially those in cultural anthropology and social psychology (Cohen 1978; Ember 1977; Jorgensen 1979). In fact, Cohen (1978) insightfully argues that self-labeling is the only valid measure of ethnicity, since it represents the internal beliefs of the individual and hence reflects the salience and reality of the ethnic affiliation s/he experience. To assess their ethnic nationality, subjects were asked if there were any nationality group with which they identified. Eleven response categories were provided to this question: Hispanic, Irish, Greek, WASP (English), Italian, Jewish, Chinese, Japanese, Black, Other and None. A similar operationalization is used in Hirschman (1981).

Measurement of Novelty Seeking

Novelty seeking was defined as the consumer's innate willingness and ability to seek out novel information about products (Hirschman 1980). Novel information may be both sensory (i.e., a sensation, such as the feeling of stroking a cat) and cognitive (i.e., encountering a propositional statement, such as gasoline is $1.50 per gallon). The construct of novelty seeking is seen as representing the search for all types of product information by the consumer, hence a measure of it should include a diverse set of products about which information may be sought (Hirschman 1982).

In the present research this construct was operationalized by asking the respondent: "How willing are you to try something new in each area listed below?" Fifteen product areas were covered; responses were recorded on a 5-point scale anchored by: (5) "very great willingness" to (1) "very little willingness." Product areas included: new foods, new places to shop, new apparel, new home furnishings, new movies, new books, new magazines, new vacation sites, new restaurants, new political ideas, new religious ideas, new types of transportation, new hairstyles, new sports activities, and new dances.

Measurement of Information Transfer

This variable represents the frequency with which the consumer transmits information to others about various consumption domains. It was measured by asking the respondent: "How frequently do you give information to other people concerning the following areas of consumption? Please estimate the number of times per month you give information to someone on this topic."

Fifteen consumption areas identical to those used in assessing product novelty seeking were used as indicators of information transfer (Hirschman 1982).

Divergent Processing Ability

This construct represents the individual's ability to generate multiple responses to a given set of requirements (Horn 1976). Within a marketing setting, it may be represented by the consumer's ability to cite multiple criteria relevant to a given consumption problem and to name multiple solutions to the problem (Hirschman 1980). To measure divergent processing ability, respondents were given the following directions:

This is a question about your problem solving as a consumer. It is measuring two parts of the problem solving process: (1) the characteristics (i.e., criteria) you perceive as relevant for solving a problem, and (2) the solutions you might consider for solving the problem.

Listed below are four consumption problem areas. You are to first list all the difference criteria that a person (not necessarily you, just people generally) could consider in solving the problem.

Second, you are to list all the different solutions you can think of for solving the problem. These don't have to be solutions that you, yourself, have used or would use. The solutions simply have to be those that could reasonably be used to solve the problem.

The four consumption problems included: (1) selecting weekend night entertainment, (2) choosing a transportation mode, (3) deciding upon a residence, and (4) purchasing a pet.

Measurement of Consumption Motives

Respondents were next given a list of possible motives for engaging in their three favorite activities and asked to indicate which motives were relevant for each activity. The motives were derived from the extensive research on participatory involvement by Hilgard (1970) and Swanson (1978) and are believed to be a representative cross-section of motivational factors underlying human behavior. These motives included: fun/pleasure, physical stamina, escape reality, perfect performance, involvement, alertness, competition, adventure, and excitement. Operational measures of these motives are available in Swanson (1978). Each description was used as a single-item indicator of that motive. The score a given respondent received for a particular motive could range from 0 to 3, depending on how many of the three favorite activities she/he indicated were undertaken to fulfill the motive.

ANALYTICAL PROCEDURE

The primary analytical technique employed for comparing the cognitive structures of the nationality subcultures was factor analysis. The theoretical and mathematical assumptions underlying factor analysis render is quite appropriate for cross-subcultural cognitive comparisons, such as those undertaken here (c.f., Acito, Anderson, Engledow 1980; Anderson, Engledow 1977; Kim and Mueller 1978). Factor analysis was used in the present study to test the proposition that the cognitive structures underlying novelty seeking, information transfer, divergent processing ability, and consumption motives are dissimilar among the nationality subcultures examined. If discrepancies are found in cognitive structures across nationality subcultures, then our ability to generalize findings from one subculture to another may be much more circumscribed than perhaps was previously assumed. Each nationality group may, in effect, constitute a distinct segment in a cognitive sense.

Technical Approach

Items representing the separate measures for each of the four constructs investigated were input to a factor analysis procedure (PA2: SPSS-V9), which extracts principal factors by utilizing communality estimates in the diagonal of the correlation matrix. To orient the obtained factors toward simple structure (Thurstone 1942; Cattell 1952), a Varimax rotational procedure was employed. After principal factors had been extracted from the data set for each nationality group and rotated to achieve simple structure, two factor comparison processes were undertaken to determine cross-group congruence. The first technique is the computation of a coefficient of congruence (Wrigley and Neuhaus 1955), the second is Cattell's salient variable similarity index (Cattell et al, 1969). The coefficient of congruence, CC, is a measure of interfactor similarity that is sensitive to both pattern and magnitude differences of the loadings. The measure is not equivalent to a correlation coefficient, because the two sets of loadings have not been standardized. The sampling distribution of CC is unknown, hence tests of statistical significance are precluded (Anderson and Engledow 1977; Levine 1977).

The second factor comparison procedure used was the salient variable similarity index. Levine (1977) argues following from Cattell (1944), that pattern similarity between paired factors is the "crucial issue" in tests of their comparability, rather than similarity of loading magnitudes. To test the similarity of paired factor loading patterns, the salient variable similarity index, S, is germane. The index is derived from classifications of loadings into salient and hyperplane categories. A hyperplane loading is operationally defined as a near-zero loading, usually interpreted as the range from -.10 to +.10. Loadings outside this range are classified as salient in defining the factor (Cattell and Baggaley 1960). The salient loadings are next classified as to their algebraic sign (+ or -). The index produces a comparison of the difference between the number of hits and misses as a proportion of a weighted sum of the cell frequencies. It reaches a maximum of one when correspondence is perfect, a minimum of minus one when the factors are perfect reflections, and zero when there is total incongruence. An approximate test of significance is available for S in Cattell et al (1969) and is used in the present study.

FINDINGS

For each of the four constructs examined--novelty seeking, information transfer, divergent processing ability, and consumption motives--three tables were developed: (1) the set of unrotated eigenvalues for the extracted principal factors, (2) a comparison of rotated primary factor loadings, and (3) the set of computed congruence coefficients and computed salient variable similarity indices. (Due to space limitations, only the latter set of tables are presented. The others are available from the author).

Because an average of four factors was extracted for each of the six nationality groups on each of the four constructs investigated, the computation of comparative statistics across groups on all factors for each construct quickly becomes overwhelming. Hence, comparisons are based on the rotated loadings of only the primary (first) factor.

Novelty Seeking

Using a cut-off criterion for factor extraction of eigenvalue > 1.0, the number of factor dimensions underlying the 15-item novelty seeking scale ranged from four (Irish, Italian) to six (Greek); the Chinese, English and Jewish samples were intermediate with five dimensions. Explained variance accounted for by the first factor ranged from a high of 40 percent for the Irish sample to a low of 25 percent for the Jewish sample. Despite this range, all first factors were at least 1.5 times as powerful in explaining variance as the second factors extracted; hence, they are appropriately viewed as primary factors (Harman 1976, Jensen 1980).

In order to have the variable loadings on each factor more closely approximate a simple structure pattern (Cattell 1960), all axes were rotated according to the Varimax criterion (Kim and Mueller 1978). Despite the fact that rotation can cause a redistribution of variance among the factors such that the potency of the primary factor is lessened, erosion was not substantial in the present case. The average primary factor eigenvalue decline was from 4.6 to 4.3, which represented an insubstantial decrease in explained variance. Further, the factor loadings approached a pattern of simple structure, with most values exceeding .30 or being below .10.

Visual examination of the high loadings for each nationality group revealed a pattern of ethnic inconsistency in the primary dimension underlying novelty seeking, as hypothesized. These cross-group discrepancies in primary factor loadings are reflected in the computed coefficients of congruence and indices of similarity (Table 1). Both measures of factorial resemblance indicate that a low level of ethnic similarity is present. The average coefficient of congruence is 57.06; the average index of similarity is 63.71, which is not significant at the p = .01 probability level (Cattell et al 1966). The structure of cognitive similarities and dissimilarities among the various nationality groups for novelty seeking was as follows: The English and Irish exhibited a moderately high level of similarity (CC=.84); The Jewish group was most similar to the Chinese (CC=.83), Greeks (CC=.88), and Italians (CC=.80). There were also several major inconsistencies: the cognitive structure underlying Irish novelty seeking was found very dissimilar to that characterizing Italians or Jews (CC = .19, .21), respectively. Likewise, the cognitive consistency of the Greeks with the Irish (CC= .36), and the English with the Italians (CC = .34) and Jews (CC = .30) was quite low. This suggests that rather stringent limits may be in effect for our ability to generalize results obtained within one of these subcultural groups to another, as regards the construct of novelty seeking.

Information Transfer

The first factors for information transfer were at least 1.; times as powerful in explaining variance as the second factors obtained; thus they may be designated as primary factors. The average eigenvalue after rotation dropped only from 6.6 to 6.4, an insubstantial decrease. Each of the rotated factors exhibited a loading pattern approximating simple structure. As before, substantial nationality group uniqueness was apparent in the coefficients of congruence and indices of similarity shown in Table 9. The average coefficient of congruence was 65.8; the mean index of similarity was 78.8, which, although it was significant at the p = .01 probability level, is not "high" in a substantive sense. The pattern of congruence coefficients was more homogeneous for information transfer, than was true for novelty seeking. There was no wide range of values; rather most are between .55 and .75 -- representing a low to moderate level of structural similarity.

The pattern of cross-group similarity is not same for information transfer as for novelty seeking, indicating some important differences in the cognitive manifestations or these two constructs, even though they were measured with similar scales. The highest levels of congruence were between the English and Greeks (CC = .79), and the English and Italians (CC = .80). The Italians and Jews displayed moderate similarity in structure, as well (CC = .77). These findings suggest that social norms governing information transfer are somewhat consistent among the English, Italian, and Greek nationality subcultures and between the Italian and Jewish subcultures. However, the similarity of their underlying cognitive structures for information transfer is not extremely high, even though moderate consistency is present. Thus behavioral generalizability from one group to another in activities regarding the transfer of consumption information may be limited.

This caveat is even more relevant for nationality groupings displaying lower levels of structural consistency. For example, to generalize findings from studies conducted on the information transfer patterns of consumers in the Chinese subculture to those who are members of the English, Greek, Irish, or Italian subcultures is likely inappropriate. In fact, the generalization of information transfer findings among any of the groups examined here would be of questionable validity. Thus nationality-based subcultural affiliation may pose a threat to the external validity of information transfer studies, especially those constructed from cognitive structural measures.

Divergent Processing Ability

The proportion of variance accounted for by the first factors ranged from a high of 71 percent (Irish) to a low of 43 percent (Greek). This proportion is generally considered to pe "high/very High" for an initial factor (Harman 1976), and is indicative of a largely unidimensional construct (Jensen 1980). This pattern had been anticipated, given the posited nature of divergent processing ability (Horn 1976).

Given the fact that so much variance is accounted for in each nationality sample by the primary factor, one would anticipate that the factor loading patterns would also display consistency across samples. This expectation was confirmed by an examination of the data in Table 3. The coefficients of congruence and indices of similarity computed for divergent processing ability show that the mean coefficient of congruence was 85.9 and the average value of the similarity index was 83.3, which was significant at--the p <.001 level of probability. Although all nationality groups were relatively consistent in their cognitive structures for this construct, the strongest pattern of similarity occurred among the Greek, Irish, Italian, and Jewish consumer subcultures. This set of groups had an average congruence of 92.3, which represents a very high level of cognitive similarity.

From this it is possible to conclude that findings regarding divergent processing ability are likely transferrable across the Greek, Irish, Italian and Jewish subcultures. That is, findings derived from research on divergent processing using one of these groups will probably be generalizable to the other groups. Further, the generally moderate to high levels of congruence among the other nationality groups -- Chinese, English -- suggest that an adequate degree of external validity will be present for divergent processing research conducted on any of the six nationality groups investigated. The only exception to this would be generalization between the Chinese and English subcultures, whose congruence was the lowest (CC = .69).

Consumption Motives

In all cases, the first factors exceeded the second factors in proportionate explained variance by two or more times. Hence, all first factors qualified as primary factors. All factors exhibited a strong pattern of simple structure. Eigenvalue erosion as a result of the rotation procedure was minor; the average decline was from 4.38 to 4.03.

Examination of the loadings across nationality groups revealed great structural similarity; almost all motives displayed high, positive correlations with the primary factor. This was readily apparent in the computed coefficients of correlation and indices of similarity given in Table 4. The average coefficient of congruence was 90.06, and the mean similarity index value was 89.20; both were quite high, and the latter was significant at beyond the p = .0001 level of probability. The high congruence of structure found for all the nationality groups examined suggests that research on consumer motives conducted within one of these subcultures should be readily transferable to the others

Discussion

The influence of national ethnicity on consumers' cognitive structure has never been addressed in a systematic fashion within consumer research. Results from the present study indicate that the cognitions of consumers drawn from similar SES and life cycle strata, but differing in nationality, were substantially incongruent in their structure for two constructs: novelty seeking and information transfer. Thus for these two aspects of consumption, nationality subcultures should be viewed as constituting distinct segments, much as was found previously for religious subcultures (Hirschman 1982).

Conversely, for two other consumption-relevant constructs--divergent processing ability and consumption motives--a high level of cross-group congruence was found. Thus, studies utilizing similar measures should produce results generalizable across subject populations similar in composition to those examined here.

Although the present research was developed primarily to demonstrate a methodological issue of importance to consumer research, its findings possess some intrinsic theoretical worth, as well. Subcultural antecedents to the social norms governing consumers' cognitive structures are little explored, or even acknowledged, in the literature. Yet studies which link cognitions to their social origins may play a valuable role in increasing our knowledge of the development and longitudinal transmission of consumption processes. Research which further develops this strand of potential causation could prove valuable to consumer behavior theory.

TABLE 1

NOVELTY SEEKING: COEFFICIENT OF CONGRUENCE AND INDEX OF SIMILARITY

TABLE 2

INFORMATION TRANSFER: COEFFICIENTS OF CONGRUENCE AND INDICES OF SIMILARITY

TABLE 3

DIVERGENT PROCESSING: COEFFICIENTS OF CONGRUENCE AND SIMILARITY INDICES

TABLE 4

CONSUMPTION MOTIVES: COEFFICIENTS OF CONGRUENCE AND SIMILARITY INDICES

REFERENCES

Acito, Franklin, Ronald D. Anderson, Jack L. Engledow (1980), "A Simulation Stud of Methods for Hypothesis Testing in Factor Analysis", Journal of Consumer Research, Vol. 7, 2, September. 141-150.

Anderson, Ronald and Jack Engledow, (1977) "A Factor Analytical Comparison or U.S. and German Information Seekers", Journal or Consumer Research, March, 185-196.

Baker, John R. (1974) Race, Oxford: Oxford University Press.

Bettman, James R., (1979), An Information Processing Theory of Consumer Choice, Reading, Mass.: Addison-Wesley Publishing Company.

Calder, Bobby J., Lynn W. Philips and Alice M. Tybout, (1980), "The Design, Conduct and Application or Consumer Research: Theory vs. Effects Oriented Studies", Proceedings of the Educators' Conference, American Marketing Association, 307-311.

Cattell, R. B., K. B. Balcar, J. C. Horn and J. R. Nesselroade, (1969) "Factor Matching Procedures: An Improvement or the S Index, with Tables," Educational and Psychological Measurement, 29: 781-792.

Cattell, R. B. and A. R. Baggaley (1960), "The Salient Variable Similarity Index for Factor Matching", British Journal of Statistical Psychology, 13, (May), 33-46.

Cattell, R. B. (1952) Factor Analysis. New York: Harper and Brothers.

Cohen, Ronald, (1978) "Ethnicity: Problem and Focus in Anthropology", Annual Review of Anthropology, Annual Reviews, Inc., Palo Alto, CA: 7: 379-403.

Cook, T. and D. Campbell (1975), "The Design and Conduct of Experiments and Quasi-experiments in Field Settings", in Handbook of Industrial and Organizational Research, M. Dunnette. ed.. Chicago: Rand McNally.

Day, George S., Allan D. Shocker and Rajendra K. Srivastava (1980), "Customer Oriented Approaches to Identifying Product Markets", Journal of Marketing, Fall, 8-19.

Deese, James, (1975) "Mind and Metaphor: A Commentary", New Literary History, 6, 211-217.

Deese, James (1965), The Structure of Associations in Language and Thought. Baltimore: Md. Johns Hopkins Press.

Deese, James, (1963), "On the Structure of Associative Meaning", Psychological Review, 63, 161-175.

Douglas, Susan P. and C. Samuel Craig, (1981), "Marketing Research in the International Environment", Handbook or International Business, Ingo Walter (ed.), New York: John Wiley and Sons.

Ember, Carol R., (1977) "Cross Cultural Cognitive Studies", Annual Review of Anthropology, Palo Alto, Ca. Annual Reviews, Inc., 6: 33-56.

Engel, James, Roger Blackwell and David Kollat (1978) Consumer Behavior, 3rd Edition, Holt, Rinehart, Winston.

Glazer, Nathan and Moynihan, Daniel P. (eds.) (1975) Ethnicity: Theory and Experience, Cambridge, Mass.: Harvard University Press.

Gorsuch, R. L. (1974), Factor Analysis, Philadelphia, W. B. Saunders, Company.

Harman, Harry H., (1976), Modern Factor Analysis, Chicago: University of Chicago Press.

Hilgard, Josephine (1970), Personality and Hypnosis: A Study of Imaginative Involvement, Chicago: University of Chicago Press.

Hirschman, Elizabeth C. (1989), "American Jewish Ethnicity: Its Relationship to some Selected Aspects of Consumption", Journal of Marketing, June.

Hirschman, Elizabeth C. (1982), "Religious Differences in Cognitions Regarding Novelty Seeking and Information Transfer", Advances in Consumer Research, Volume 9, Association for Consumer Research, forthcoming.

Hirschman, Elizabeth C., (1980), "Innovativeness, Novelty Seeking, and Consumer Creativity", Journal or Consumer Research, 7 (December), 283-295.

Hirschman, Elizabeth C., and Susan P. Douglas, (1981) "Hierarchical Cognitive Content: Towards a Measurement Methodology," in Advances in Consumer Research, Association for Consumer Research: Ann Arbor, HI: Vol. 8., Kent B. Monroe (ed.), 100-5.

Horn, John L. (1976), "Human Abilities: A Review of Research and Theory in the Early 1970's", Annual Review of Psychology, Annual Reviews, Inc., Palo Alto, CA. 437-85.

----------------------------------------