Organizational Structure of Values - Crosby, Bitner

Embed Size (px)

DESCRIPTION

An article about organizational values.

Citation preview

  • J BUSN RES 1990:20:123-134

    123

    Organizational Structure of Values

    Lawrence A. Crosby

    Mary Jo Bitner Arizona State University

    James D. Gill Walker Research Inc., Tempe, AZ

    The study reported here attempts to overcome weaknesses in previous investiga- tions of the factor structure of the Rokeach Value Survey. The approach involves comparing a priori models of how values are organized by testing hierarchically nested models using confirmatory factor analysis. One model posits that values are independent, while the other models suggest that both terminal and instrumental values can be organized along a few underlying dimensions. The findings indicate that 1) instrumental values ratings reflect the importance of three underlying uni- polar dimensions: self-direction, conformity, and virtuousness; 2) terminal values ratings reflect the importance of one bipolar dimension, self-actualization versus hedonism, and two unipolar dimensions: idealism and security.

    Introduction

    The nature and structure of human values is a topic of continuing interest in marketing, other business fields (e.g., Hofstede, 1985), and the social sciences generally (see Clawson and Vinson, 1978). Within the marketing discipline, theories have been advanced regarding the impact of values on buyer behavior (e.g., Vinson et al., 1977). Also within marketing, empirical studies have established linkages to product/brand choice (e.g., Henry, 1976), store patronage (Becker and Conner, 1982), price-quality perceptions (Petit et al., 1985), leisure attitudes and activities (Jackson, 1973), shopping orientation (Darden et al., 1979), media usage (Becker and Conner, 1981), media preferences (Beatty et al., 1985), and the various in- tervening variables of consumer behavior (Homer and Kahle, 1988; Pitts and Wood- side, 1984). Values antecedents have also been investigated (e.g., Crosby et al., 1984; Kahle, 1986).

    Address correspondence to Lawrence A. Crosby, School of Business, Arizona State University, Tempe, AZ 85287.

    Journal of Business Research 20, 123-134 (1990) 0 1990 Elsevier Science Publishing Co., Inc. 1990 655 Avenue of the Americas, New York, NY 10010

    014%2963/90/$3.50

  • 124 J BUSN RES 1!990:20:123-334 L. A. Crosby et al.

    The interest in values research is accompanied by concerns regarding the con- ceptualization of values and the validity and reliability of values measures. For example, considerable effort in several disciplines has been devoted to investigating the factor structure of the Rokeach Value Survey (RVS) (Rokeach, 1973), probably the most commonly used measure of values. Using exploratory factor analysis, researchers have sought to explain the covariation among the 36 items in the RVS (Darden et al., 1979; Feather and Peay, 1975; Heath and Fogel, 1978; Maloney and Katz, 1976; McKernan, 1982; Vinson et al., 1977). Despite the considerable activity, there is disagreement among researchers regarding the dimensions un- derlying the 36 values of the RVS. This disagreement is not surprising, given the arbitrary assumptions that underlie exploratory factor analysis.

    The study described in this article attempts to overcome weaknesses in previous investigations of the factor structure of the RVS. The study is motivated by the- oretical concern with uncovering basic goal orientations underlying human activity and personality. The approach involves confirmatory factor analysis via LISREL (Joreskog and S&born, 1984). The models tested are based in theory and draw on previous research results.

    The Rokeach Value Survey

    With a few exceptions (e.g., Beatty et al., 1985; Homer and Kahle, 1988; Kahle et al., 1986), studies investigating the relationship between values and consumer behavior have typically employed the RVS (Munson, 1984). Rokeach (1973) has argued persuasively that values are separately organized into relatively enduring hierarchical structures of terminal and instrumental values. He defines values as . . . enduring beliefs that specific modes of conduct (instrumental values) or end- states of existence (terminal values) are personally or socially preferable to opposite or converse modes of conduct or end-states of existence (p. 5). The RVS was developed by the selection of 18 terminal and 18 instrumental values from a larger compilation of several hundred value descriptors based on . . . retaining only one from a group of synonyms or near synonyms. . . retaining only those judged to be maximally different from or minimally correlated with one another. . . . (Rokeach, 1973, p. 29).

    One organizational issue debated in the consumer literature concerns the a priori classification of values into two groups: terminal versus instrumental. An argument for their separate treatment is that terminal and instrumental values are by definition different (ends versus means), although they may be causally related. Furthermore, past attempts to factor analyze the entire 36 x 36 correlation matrix of rankings (e.g., Feather and Peay, 1975; Rokeach, 1973) or ratings (Vinson et al., 1977) have produced factors that tend to be exclusively terminal or instrumental. Thus, drawing a means versus ends distinction seems appropriate (for a contrary opinion, see Heath and Fogel, 1978). Although the distinction is useful and intuitive, further reduction of the 36 values would aid theoretical understanding and perhaps lead to more parsimonious and generalizable models.

    Another challenge to understanding the dimensionality of the RVS is the con- troversy over value importance measurement via rankings or ratings (e.g., Munson and McIntyre, 1979; Reynolds and Jolly, 1980). While both methods have their advantages and disadvantages, rankings have been criticized as imposing an artificial

  • Organizational Structure of Values .I BUSN RES 1990:20:123-134

    contrast on the data (Alwin and Krosnick, 1985). In other words, rankings may force respondents to make trade-offs they would not otherwise make. Some people, e.g., may feel that self-direction and conformity are both important modes of conduct even though they are logically opposite. Since a goal of the data reduction is to uncover underlying value dimensions in the respondents psycho-logic (not the researchers), ratings would seem to be preferred, as they impose fewer con- straints on the data.

    Competing Models of Values Organization

    Independence Model of Terminal and Instrumental Values

    Noting generally low correlations among the values and citing factor analysis results that failed to account for the majority of the variance, Rokeach concluded that the 36 values could not be safely reduced (Rokeach, 1973). Other researchers have sought to uncover the organizational structure of the RVS and have found from four to ten dimensions. Differences may be explained partially by the use of rank- ings versus ratings and 36 x 36 versus 18 x 18 item correlation matrices. However, the possibility must also be acknowledged that Rokeachs conclusion is correct and that the values are basically independent, which leads to the following hypotheses:

    Hla: There is no systematic association among the instrumental values contained within the RVS, beyond that attributable to correlated measurement error.

    Hlb: There is no systematic association among the terminal values contained within the RVS, beyond that attributable to correlated measurement error.

    Implicit in Hla and Hlb is the recognition that correlated measurement error could potentially mask the independence of value items and, therefore, needs to be controlled. This can occur when ratings are used to measure values impor- tance. A drawback of ratings is their susceptibility to problems of response styles or sets (Alwin and Krosnick, 1985).

    Nonindependence Model of Instrumental Values

    While factor structures from previous RVS research are by no means identical, there are some striking similarities. A review of these studies (Feather and Peay, 1975; Rokeach, 1973; Vinson et al., 1977) suggests an external-internal organi- zation of instrumental values, as depicted in Table 1. Internal dimensions composed of varying subsets of self-direction values shown in Table 1 have carried labels such as inner-directed, competence, self-expansion, self-reliant, achievement-oriented. External dimensions have been of two types, one group reflecting conformity (e.g., other-directed, sociality, self-constriction, adherence to social norms), and the other focused on concern for others (e.g., religious morality, compassion, virtuous, al- truistic). Also, as previously noted, conformity and/versus self-direction have been hypothesized to underlie Kohns measures of parental values (Alwin and Jackson, 1982; Jackson and Alwin, 1980). Kohns values would be classified as instrumen- tal by Rokeachs definition (e.g., honest, clean, self-controlled, responsible). The characterization of instrumental values as internally or externally focused is con- sistent with the description of Riesman et al. (1950) of inner- versus other-directed

  • J BUSN RES 1990:20:123-134

    L. A. Crosby et al.

    Table 1. Hypothesized Instrumental Values Organization

    External

    Conformity dimension Cheerful Clean Obedient Polite Responsible

    Virtuous dimension Forgiving Helpful Honest Loving

    Internal

    Self-direction dimension Ambitious Broadminded Capable Courgeous Imaginative Independent Intellectual Logical Self-controlled

    personality. Similar distinctions are drawn between nonsocial and social motives (McGuire, 1974), self-perception differences reflecting an internal or external focus (Rogers, 1951; Wylie, 1961), and internal versus external locus of control (Lefcourt, 1966; Rotter, 1966).

    H2: Instrumental values are organized according to their internal or external focus. External values are further distinguished by whether they promote conformity or virtuousness. Thus, three dimensions are predicted, as shown in Table 1.

    An alternative configuration of instrumental values would suggest that these three dimensions are to some degree redundant. Fewer than three factors may exist with the values clustered at opposite ends of one or two bipolar dimensions (e.g., conformity versus self-directed). If this is the case, then H2 would not be able to withstand a test of discriminant validity.

    Nonindependence Model of Terminal Values

    A review of previous research (Feather and Peay, 1975; Rokeach, 1973; Vinson et al., 1977) suggests that terminal values may be organized as depicted in Table 2. The first distinction, societal or personal, pertains to whether the desired end- state applies mainly to the individual (e.g., sense of accomplishment, comfortable life) or would be a collective benefit (e.g., world of beauty, world at peace). This distinction is consistent with the notion that consumers can be segmented by the degree to which they consider the societal consequences of their consumption decisions (e.g., Webster, 1975). The distinction is also consistent with research involving the nine-item List of Values (LOV), where external/internal and aper- sonal/personal dimensions have been identified (Homer and Kahle, 1988).

    Terminal values can be positioned as lower or higher order. As noted by Rokeach, Maslows (1954) conception of higher-order and lower-order values can be fruitfully employed. However, it is recognized that a lower-order versus higher-order distinction is quite judgmental (Kahle, 1983). This nomenclature is retained, however, in light of Rokeachs explicit recognition of Maslows work. A more appropriate label might be material-psychological. In previous research involving the RVS, the self-actualization dimension has also been termed self- realization or delayed gratification. Hedonism corresponds to such dimensions as

  • Organizational Structure of Values

    Table 2. Hypothesized Terminal Values Organization

    J BUSN RES 1990:20:123-134

    127

    Higher Order

    Societal

    Idealism dimension A world of beauty (beauty of nature and

    arts) Equality (brotherhood, equal opportunity

    for all) Freedom (independence, free choice)

    Personal

    Self-actualization dimension Sense of accomplishment (lasting

    contribution) Inner harmony (freedom from inner conflict) Mature love (sexual and spiritual intimacy) Salvation (saved, eternal life) Self-respect (self-esteem) True friendship (close companionship) Wisdom (mature understanding of life)

    Lower Order

    Security dimension World at peace (free of war and conflict) Family security (taking care of loved ones) National security (protection from attack)

    Hedonism dimension A comfortable life (a prosperous life) An exciting life (a stimulating, active life) Happiness (contentedness) Pleasure (an enjoyable, leisurely life) Social recognition (respect, admiration)

    personal or immediate gratification. Values included under idealism have loaded on factors labeled noble and social harmony. Security and safety are terms that have been used to describe the fourth dimension.

    H3: Terminal values are organized according to whether they are personally or so- cietally focused and based on a higher-/lower-order distinction, as shown in Table 2. Thus, four dimensions are predicted.

    Again, a logical alternative to H3 is a reduced space with fewer dimensions, some of which might be bipolar. The potential for dimension reduction would be revealed in tests of discriminant validity.

    Sample and Procedure

    The data were obtained from a multiwave survey of adults conducted via mail in California and Colorado. Five hundred households participating in a national panel were selected from each state, with the use of a quota sampling procedure. House- holds were selected for inclusion by the matching of household characteristics with census data for each state. A random procedure was used to select an eligible adult within each household. A total of 461 respondents participated in all phases of the survey (46% overall response rate), 418 of whom had complete data on all the value items.

    A rank, then rate approach was used to measure the importance of the 18 terminal and 18 instrumental values included in the RVS (Rokeach, 1973). For reasons previously stated, value ratings were believed to be the appropriate mea- sures of importance, given the purpose of the study. Yet, it was recognized that value rankings may give rise to more precise data in that respondents are encour- aged to make fine distinctions about the relative importance of the values. In an effort to encourage data quality, but still allow respondents to judge some values

  • 128 J BUSN RES 1!990:20:123-134 L. A. Crosby et al.

    as equally important, a two-step procedure was employed. Respondents began by ranking the list of alphabetically arranged terminal values. Next, each of the ter- minal values was rated on a 7-point Extremely Important to Not at All Im- portant scale. values.

    Analysis and Results

    The two-step piocedure was thkn repeated for the instrumental

    The efforts to test Hl-H3 followed the recommended procedures for confirmatory factor analysis (CFA) (Anderson and Gerbing, 1988; Gerbing and Anderson, 1988). Confirmatory factor analysis is the preferred method for testing the unidimension- ality of scales, i.e., that the alternate indicators (values) of the same underlying construct (values dimension) share just one trait in common. The CFA was confined to the ratings data. While the measurement scales of both the ratings and rankings are ordinal, ratings are more amenable to causal analysis than ranks that have the added problem of being ipsative in nature (Alwin and Krosnick, 1985).

    The analysis is best understood with reference to the models in Figures 1 and 2 (ignoring for now the numerical estimates). Consistent with H2 and H3, each value is depicted as a function of one value dimension, a methods bias, and a unique (random) error component (Q). The methods bias factor attempts to account for the inherent positive correlation of values ratings (Alwin and Krosnick, 1985). The overall goodness of fit of these models, evaluated according to the similarity of predicted with actual correlations, provides a direct test of unidimensionality (Gerbing and Anderson, 1988, p. 187).

    The proposed models were tested using LISREL (Jbreskog and Siirbom, 1984) with the analysis performed on the correlation matrices of instrumental and terminal values ratings. The maximum likelihood (ML) estimation method was employed, providing significance tests of individual parameter estimates and overall model fit. The following results were obtained regarding the overall fit of the two models:

    Instrumental (Fig. 1) Terminal (Fig. 2)

    X2 df 416.71 114 341.64 111

    Prob. GFI RMS

    ,000 .91 .05 ,000 .92 .OS

    The significant chi-square results (normally indicating a lack-of-fit) must be in- terpreted in light of several considerations. First, the magnitude of the chi-square is dependent on sample size (Jiireskog, 1978, p. 447). The sample of 418 is relatively large and in . . . large samples almost any model . . . is likely to be rejected (Long, 1983, p. 75). Second, ML assumes a multivariate normal distribution of the observed variables, which is not the case for the values ratings. Violations of multivariate normality tend to inflate the chi-square. Third, the other indicants of model fit are within an acceptable range. For both instrumental and terminal values, the goodness-of-fit index (GFI) exceeded Bentler and Bonetts (1980) heuristic of .90. Also, the Root Mean Square Residual (RMS) values of .05 indicated only modest deviation between the predicted and actual correlations.

    Supporting the interpretation that the models in Figures 1 and 2 adequately fit the data, tests using the unweighted least squares (ULS) estimation method yielded

  • Organizational Structure of Values J BUSN RES 1990:20:123-134

    129

    c=.69,

    28 _Xl Ambltlous

    F& 21 _x2 Broadminded

    F& 48 _ X6 Courageous

    F=.46, 28 ) X10 lmaglnative

    Xl 2 Intellectual

    F=.64, X18 Self-Controlled

    Xl 7 Responsible

    Figure 1. Instrumental values model (maximum likelihood estimates). Key: * = not sig- nificant (t-value < 2.00).

    extremely high GFIs: instrumental = .98; terminal = .98. Unlike ML, ULS makes no assumptions about the distribution of the observed variables. Because statistical tests of individual parameter estimates are not possible with ULS, the ML solutions were relied on for the remainder of the analysis (the ML and ULS parameter estimates were all very similar in magnitude).

    A hierarchical testing procedure was used to assess the need for a methods bias factor in the two models (Bentler and Bonett, 1980). Essentially, this involved eliminating the methods latent variables (& for instrumental, & for terminal), re- estimating the models, then testing the change in x2 against the change in degrees of freedom. The results of this analysis appear in Table 3. The number 1 models for instrumental and terminal values correspond to Figures 1 and 2, respectively.

  • 130 J BUSN RES 1990:20:123-134 L. A. Crosby et al.

    a Methi Bias -

    .4

    .E

    .E

    .l

    .L

    .L

    .L

    .

    1

    .I

    Sense of Accomplishment

    X10 Inner Harmony

    Xl 4 Salvation

    Xl 5 Self-Respect

    Xl 7 True Friendship

    Xl8 Wisdom

    F =.53 13 _ - Xl A Comfortable Life

    I =.70 50 -+ X2 An Excltmg Life

    E =.46 - Xl3 Pleasure

    =.6_7, 57 ~ Xl6 Social Recognltlon

    -a Hedkm

    Figure 2. Terminal values model (maximum likelihood estimates). Key: * = not significant (t-value < 2.00); ** = fixed parameter; also: +,, = .28, +,4 = Sl, & = -.44, & = .09*, q134 = .77.

    The number 2 models contain no methods effect. Throughout Table 3, the Ax2 and Adf are evaluated relative to model number 1. As the results show, the change in x2 was statistically significant for both instrumental and terminal values, indicating a need to retain the methods factors.

    The general approach of comparing hierarchically nested models was extended to assess further the construct validity of the values dimensions. The competing independence versus nonindependence hypotheses (Hla versus H2 and Hlb versus H3) were examined by comparing the fit of the methods-only models (nos. 3) with the full, hypothesized models (nos. 1). As Table 3 reveals, rejecting the inde- pendence models and associated hypotheses was possible.

  • Organizational Structure of Values J BUSN RES 1!990:20:123-134

    Table 3. Tests of Nested Models

    x2 df Ax2 Adf P Instrumental values models

    Model number 1. Hypothesized model with method control 2. Hypothesized model without method (no e4) 3. Hypothesized model with only method (no &, &, k3) 4. Self-Direction = conformity (+,, = 1) 5. Self-Direction = virtuous (+,3 = 1) 6. Conformity = virtuous (I& = 1)

    Terminal values models Model number 1. Hypothesized model with method control 2. Hypothesized model without method (no &) 3. Hypothesized model with only method (no t,, .$, .&, 54) 4. Self-Actualization = hedonism (I& = 1) 5. Self-Actualization = idealism ($13 = 1) 6. Self-Actualization = security (+,4 = 1) 7. Hedonism = idealism (+Z3 = 1) 8. Hedonism = security (& = 1) 9. Idealism = security (9% = 1)

    417 615

    1226 474

    446

    342 549

    1006 342

    354 347 353 352

    114 - - - 132 198 18 < 0.001 135 809 21 < 0.001 115 57 1 < 0.001 Failed to converge 115 29 1 < 0.001

    111 - - - 129 207 18 < 0.001 135 664 24 < 0.001 112 0 1 N.S. Failed to converge 112 12 1 < 0.001 112 5 1 < 0.05 112 11 1 < 0.01 112 10 1 < 0.01

    Although there was systematic variation among the values over and above the methods effect, it did not necessarily follow that the values dimensions were prop- erly specified. Both discriminant and convergent validity needed to be established. Again, by following procedures recommended by Anderson and Gerbing (1988), discriminant validity was assessed for each pair of instrumental and terminal values constructs by constraint of the estimated correlation parameter (4,) between them to 1.0, then performance of another chi-square difference test (i.e., Ax2 and Adf relative to the no. 1 models). As reported in Table 3, discriminant validity was supported by a significantly higher x2 in all cases but one (except where the model failed to converge altogether). Among the terminal values, there appeared to be a lack of discriminant validity between self-actualization and hedonism. With &, = 1, the model fit the data as well as when &, was estimated; all of the self-actual- ization values loaded negatively on 5, and all the hedonism values loaded positively on c2. In other words, self-actualization and hedonism were perfectly (albeit neg- atively) correlated, and, rather than two dimensions, there was only one bipolar dimension underlying the data, anchored at one end by self-actualization and at the other end by hedonism. The other component of construct validity, convergent validity, was assessed by determination of whether each values pattern coefficient on its posited underlying construct factor was significant (Anderson and Gerbing, 1988). The magnitude of the coefficients is reported in Figures 1 and 2. For the instrumental values, the ML estimates were based on the hypothesized model (no. 1). For the terminal values, the ML estimates were based on the alternate model (no. 4), with the correlation between self-actualization and hedonism fixed at &, = - 1.0 so as to generate all positive pattern coefficients. With one exception among the instrumental values (responsible on conformity) and two exceptions among the terminal values (mature love on self-actualization and social recognition on hedonism), the coefficients were statistically significant.

    The final step in the measurement process involved assessing the reliability of

  • 132 J BUSN RES 1990:20:123-134

    L. A. Crosby et al.

    the values constructs. In addition to unidimensionality, the usefulness of a scale also depends on its reliability or freedom from random measurement error. Ac- cording to Gerbing and Anderson (1988), reliability should be assessed after uni- dimensionality is established (p. 190). With the use of the formula fi= (%,)2/ ((CA,)+ CG,,), these estimates of reliability were computed: self-direction = .81, conformity = .57, virtuousness = .65, self-actualization/hedonism = .62, idealism = .58, security = .67. These results suggested fair-to-moderate reliability in the mea- surement of the constructs. While improvements in reliability could no doubt be obtained by deletion of values from the analysis (i.e., increase of the specificity of the latent variables), this was considered incompatible with the research purpose of the testing of models representing the organization of the entire RVS.

    Conclusion

    The CFA results failed to support the independence model of values (Hla, Hlb) while supporting the nonindependence models of both instrumental and terminal values, as specified in Figures 1 and 2. With the exception of the self-actualization and hedonism dimensions (as discussed above), the organizational structures were as hypothesized in H2 (instrumental) and H3 (terminal). These findings support the development of more parsimonious and, hopefully, more generalizable models. Based on the results reported here, future research may attempt to respecify the models by either a) deleting responsible, mature love, and social recognition; b) allowing them to load on different factors; or c) representing them as unique values measured by single indicators. The choice will depend on further development of the theory of value organization and on empirical tests performed on new data.

    Additional work is also needed to begin the development of nomological net- works that link the values dimensions to consumer behavior. For example, it might be speculated that consumers whose instrumental values are external/conformity focused might be more prone to normative compliance and national-brand con- sumption, while those who are internally focused might be more likely to experi- ment with new products and to complain when they are dissatisfied. With respect to terminal values, it might be speculated that those who are motivated by societal/ higher-order (idealism) concerns would be more likely to support regulation of business, ecological consumption, and consumer boycotts. On the other hand, those consumers whose terminal value structure is dominated by self-actualization values might exhibit greater than normal demand for educational products and be less materialistic in their lifestyles than would those whose terminal value structure reflects a hedonistic orientation. These latter consumers may be heavy purchasers of leisure products and services and more likely to seek out highly stimulating consumption environments.

    References

    Alwin, Duane F., and Jackson, D. J., The Statistical Analysis of Kohns Measures of Parental Values. Systems Under Indirect Observation: Causality, Structure and Prediction. K. G. Joreskog and H. Weld, eds., North-Holland, Amsterdam, 1982, pp. 197-223.

    Alwin, Duane F., and Krosnick, Jon A., The Measurement of Values in Surveys: A Comparison of Ratings and Rankings, Public Opinion Quarterly 49 (1985): 535-552.

    Anderson, James C., and Gerbing, David W., Structural Equation Modeling in Practice: A Review and Recommended Two-Step Approach, Psychological Bulletin, in press.

  • Organizational Structure of Values J BUSN RES 1990:20:123-134 133

    Beatty, Sharon E., Kahle, Lynn R., Homer, Pamela and Misra, Shekhar, Alternative Measurement Approaches to Consumer Values: The List of Values and the Rokeach Value Survey, Psychology and Marketing 2 (3, 1985): 181-200.

    Becker, B. W., and Connor, P. E., The Influence of Personal Values on Attitude and Store Choice Behavior, in Educators Proceedings. B. J. Walker et al., eds., American Marketing Association, Chicago, 1982: pp. 21-24.

    Becker, B. W., and Connor, P. E., Personal Values of the Heavy User of Mass Media, Journal of Advertising Research 21 (1981): 37-43.

    Bentler, P., and Bonett, D. G., Significance Tests and Goodness of Fit in the Analysis of Covariance Structures, Psychological Bulletin 88 (1980): 588-606.

    Clawson, C. Joseph, and Vinson, Donald E., Human Values: A Historical and Interdisciplinary Analysis, in Advances in Consumer Research. H. Keith Hunt, ed., Association for Consumer Research, Provo, UT, 1978, pp. 396-402.

    Crosby, L. A., Gill, J. D., and Lee, R. E., Life Status and Age as Predictors of Value Orientation, in Personal Values and Consumer Psychology. R. E. Pitts and A. G. Woodside, eds., Lexington Books, Lexington, MA, 1984, pp. 201-218.

    Darden, W. R., Erdem, O., Darden, D. K., and Powell, T., Consumer Values and Shopping Orientations, in Proceedings of the Southwestern Marketing Association, San Antonio, TX, 1979.

    Feather, N. T. and Peay, E. R., The Structure of Terminal and Instrumental Values: Dimensions and Clusters, Australian Journal of Psychology 27 (1975): 151-164.

    Gerbing, David W., and Anderson, James C., An Updated Paradigm for Scale Development Incorporating Unidimensionality and Its Assessment, Journal of Marketing Research 25 (1988): 186-192.

    Heath, R. L., and Fogel, 0. S., Terminal and Instrumental? An Inquiry into Rokeachs Value Survey, Psychological Reports 42 (1978): 1147-1154.

    Henry, W. A., Cultural Values do Correlate with Consumer Behavior, Journal of Marketing Research 13 (1976): 121-127.

    Hofstede, Geert, The Interaction Between National and Organizational Value Systems, Journal of Management Studies 22 (1985): 347-357.

    Homer, Pamela M., and Kahle, Lynn R., A Structural Equation Test of the Value-Attitude- Behavior Hierarchy, Journal of Personality and Social Psychology 54 (1988): 638-646.

    Jackson, D. J. and Alwin, D. F., The Factor Analysis of Ipsative Measures, Sociological Methoa!s and Research 9 (1980): 218-238.

    Jackson, Royal G., A Preliminary Bicultural Study of Value Orientations and Leisure Attitudes, Journal of Leisure Research 5 (1973): 10-22.

    Joreskog, K. G., Structural Analysis of Covariance and Correlation Matrices, Psychometrica 43 (1978): 443-477.

    Joreskog, K. G. and Sorbom, D., Lisrel VI: Analysis of Linear Structural Relationships by the Method of Maximum Likelihood, Scientific Software, Inc., Mooresville, IN, 1984.

    Kahle, Lynn R., The Nine Nations of North America and the Value Basis of Geographic Segmentation, Journal of Marketing 50 (1986): 37-47.

    Kahle, Lynn R., ed., Social Values and Social Change: Adaptation to Life in America, Praeger, New York, 1983.

    Kahle, Lynn R., Beatty, Sharon E., and Homer, Pamela, Alternative Measurement Approaches to Consumer Values: The List of Values (LOV) and Values and Life Style (VALS), Journal of Consumer Research 13 (December 1986): 405-409.

    Lefcourt, H. M., Internal Versus External Control of Reinforcement: A Review, Psychological Bulletin 65 (1966): 206-220.

    Long, J. S., Covariance Structure Models: An Introduction to LISREL, Sage Publications, Beverly Hills, CA, 1983.

  • 134 J BUSN RES 1990:20:123-134 L. A. Crosby et al.

    Maloney, J. and Katz, G. M., Value Structures and Orientations to Social Institutions, Journal of Psychology 93 (1976): 203-211.

    Maslow. A. H., Motivation and Personality, Harper and Row, New York, 1954.

    McGuire, W. J., Psychological Motives and Communication Gratification, in The Uses of Mass Communications. J. G. Blumler and C. Katz, eds., Sage Publications, Beverly Hills, CA, 1974, pp. 167-196.

    McKernan, J., Value Systems and Race Relations in N. Ireland and America, Ethnic and Racial Studies 5 (1982): 1566174.

    Munson, J. Michael, Personal Values: Considerations on Their Measurement and Application to Five Areas of Research Inquiry, in Personal Values and Consumer Psychology. R. E. Pitts and A. G. Woodside, eds., Lexington Books, Lexington, MA, 1984, pp. 13-33.

    Munson, J. Michael, and McIntyre, S. H., Developing Practical Procedures for the Measurement of Personal Values in Cross-Cultural Marketing, Journal of Marketing Research 16 (1979): 48-52.

    Petit, Kathy L., Sawa, Sonja L., and Sawa, Ghazi H., Frugality: A Cross National Moderator of the Price-Quality Relationship, Psychology and Marketing 2 (1985): 253-266.

    Pitts, R. E., and Woodside, A. G., Personal Values and Consumer Psychology, Lexington Books, Lexington, MA, 1984.

    Reynolds, T. J., and Jolly, J. P., Measuring Personal Values: An Evaluation of Alternative Methods, Journal of Marketing Research 17 (1980): 531-536.

    Riesman, D., Denny, R., and Glazer, N., The Lonely Crowd, Yale University Press, New Haven, CT, 1950.

    Rogers, C., Client-Centered Therapy, Houghton Mifflin, Boston, 1951.

    Rokeach, M., The Nature of Human Values, The Free Press, New York, 1973.

    Rotter, J. B., Generalized Expectancies for Internal Versus External Control of Reinforcement, Psychological Monographs, 80 (1966).

    Vinson, D. E., Munson, J. M., and Nakanishi, M., An Investigation of the Rokeach Value Survey for Consumer Research Applications, in Advances in Consumer Research. W. D. Perreault, ed., Association For Consumer Research, Provo, UT, 1977, pp. 247-252.

    Vinson, D. E., Scott, J. E., and Lamont, L. M., The Role of Personal Values in Marketing and Consumer Behavior, Journal of Marketing 41 (1977): 44-50.

    Webster, F. E. Jr., Determining the Characteristics of the Socially Conscious Consumer, Journal of Consumer Research 2 (1975): 188-196.

    Wylie, R. C., The Self Concept, University of Nebraska Press, Lincoln, 1961.