Journal of Economic Psychology 26 (2005) 624–641 www.elsevier.com/locate/joep
A cross-cultural examination of the relationship between materialism and individual values William Kilbourne a
a,*
, Marko Gru¨nhagen b, Janice Foley
c
Department of Marketing, Clemson University, Clemson, SC 29634-1325, United States b Southern Illinois University Edwardsville, Edwardsville, IL 62026, United States c University of Regina, Regina, Sask., Canada S45 0A2
Received 1 September 2004; received in revised form 22 November 2004; accepted 22 December 2004 Available online 17 March 2005
Abstract This paper examines the relationship between openness, self-transcendence and self-enhancement values [Schwartz, S. H., & Sagiv, L. (1995). Identifying culture-specifics in the content and structure of values. Journal of Cross-Cultural Psychology, 26 (1), 92–116] and attitudes toward consumption as measured by the Richins and Dawson materialism scale [Richins M. L., & Dawson, S. (1992). A consumer values orientation for materialism and its measurement: Scale development and validation. Journal of Consumer Research, 19(December), 303–316]. The study encompassed three countries, Canada, Germany, and the US. The results indicated that both the materialism and value scales were invariant across countries and that the proposed causal model was invariant as well. The relationship between self-transcendence and materialism was negative, and the relationship between self-enhancement and materialism was positive. Ó 2005 Elsevier B.V. All rights reserved. JEL classification: A13; A14; D12; M30 PsycINFO classification: 2930; 3920 Keywords: Cross-cultural differences; Consumer research; Values; Attitudes
*
Corresponding author. Tel.: +1 864 656 5296; fax: +1 864 656 0138. E-mail addresses:
[email protected] (W. Kilbourne),
[email protected] (M. Gru¨nhagen),
[email protected] (J. Foley). 0167-4870/$ - see front matter Ó 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.joep.2004.12.009
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
625
1. Introduction Interest in materialism and its implications is apparent as far back as the early Greek philosophers. Pythagoras, for example, required that students relinquish their personal possessions before entering his school. Rudmin and Kilbourne (1996) provide a historical view of different attitudes toward materialism from the ancient world to the modern. Critiques of consumption do indeed transcend time and have remained forceful for thousands of years, though they have failed to deter the progress of materialism. This fact has been lamented by playwrights such as Shakespeare and Ibsen, romantic poets such as Emerson and Wordsworth, and economic historians such as Veblen and Tawney. Marketing scholars recently returned to an assessment of materialism in contemporary societies. While some studies have remained critical of consumption practices, referring to materialism as a ‘‘dark side’’ variable (Mick, 1996), others have suggested that materialism might have positive aspects as well (Holt, 1997). Despite the many examinations of materialism, it remains unclear how it relates to other aspects of life. Micken and Roberts (1999) suggest that materialism reflects a quest for certainty of identity. Schouten and McAlexander (1995) carry this idea further, arguing that consumers use products as a basis for social cohesion and interaction in subcultures of consumption. Contrary to this, it has been argued that materialism has a negative effect on perceptions of well-being (Richins & Dawson, 1992), quality of life (Muncy & Eastman, 1998), and sustainability (Norgaard, 1995). This suggests that the relationship of materialism to life in Western societies is still not well understood. Burroughs and Rindfleisch (2002) argue that the nature of materialism needs to be clarified, and that more researchers are beginning to examine it in the context of other life goals and values. Richins and Dawson (1992), Kasser and Ryan (1993), and Burroughs and Rindfleisch (2002) suggest that materialism can be considered the value consumers place on possessions, and this indicates that materialism is a value. Values researchers have argued, however, that the study of individual values should be undertaken in the context of the larger value systems that individuals hold (Schwartz, 1992). The purpose of this paper is twofold. The first purpose is to examine materialism as a variable in the context of values as suggested by values researchers. Thus, we will first determine if materialism is related to other values the individual might hold. The second purpose is to determine if this relationship holds across cultures. We propose to develop a causal model relating materialism to values and then to test the crosscultural invariance of that model.
2. Research on materialism 2.1. Definition and measurement A number of different, though similar, definitions of materialism have been advanced over time. The definition posed by Belk (1985) suggests that materialism is
626
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
‘‘the importance a consumer attaches to worldly possessions’’ (p. 291). Rassuli and Hollander (1986) refer to ‘‘. . . an interest in getting and spending’’ (p. 10). Burroughs and Rindfleisch (2002) add to this the belief that oneÕs ‘‘. . . well-being can be enhanced through oneÕs relationship with objects’’ (p. 349). In a more psychological approach, Micken and Roberts (1999) suggest that materialistic behaviors reflect ‘‘. . . a preference for certainty, with materialists relying on objects not just as identity markers, but identity fixers’’ (p. 513). Finally, Browne and Kaldenberg (1997) characterize materialism as a cluster of values and traits that focus on possessions. What all these definitions have in common is that they reflect the use of consumption to acquire more than instrumental or functional value in the things purchased. Collectively, the definitions suggest that the individual seeks a relationship with objects through which she or he is enhanced in some way. This recalls JamesÕ (1890) description of the self as ‘‘. . . in its widest possible sense. . . a manÕs Self is the sum total of all that he CAN call his. . .’’ (p. 291). He included in this collection clothes, houses, land, and bank accounts. From a conceptual perspective, this suggests that materialism is a multi-faceted construct relating to the importance of material objects to the individual. This construct has been derived empirically and tested by Richins and Dawson (1992), and it is the basis for the measurement that will be used here. This model characterizes materialism as a three-dimensional construct. The dimensions relate to beliefs that materialism leads to happiness, that success is best defined in material terms, and the significance individuals attach to materialism. The last dimension is referred to as centrality. It has been further demonstrated that materialism can be represented as a second order factor with the three original dimensions as the latent constructs, and that the original eighteen item scale can be reduced to a smaller number of items while maintaining its validity (Roberts, Manolis, & Tanner, 2003). 2.2. Approaches to materialism From the individual, or micro, perspective, recent research has related to measurement issues, psychological traits, value orientations, and ethics. Belk (1985) and Richins and Dawson (1992) have developed materialism measurement scales that can be used to determine the level of materialism in an individual. These scales have, in turn, been used to assess the consequences of materialism for the individual. The consensus in these studies has been that materialism is negatively related to such factors as happiness, well-being, and life satisfaction (Belk, 1985; Burroughs & Rindfleisch, 2002; Richins & Dawson, 1992). Materialism has been found to be related to family structure with higher levels of materialism associated with disrupted families (Rindfleisch, Burroughs, & Denton, 1997; Roberts et al., 2003). From a different perspective, recent research has suggested that consumption is frequently used as a means of self-identity. Holt (1997) demonstrates that consumption has been used to demarcate social class and that its meaning varies across groups. Using the same approach, Schouten and McAlexander (1995) argue that subcultures are formed and maintained by shared consumption practices. These results are consistent with McCrackenÕs (1988) contention that the cultural codes of
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
627
consumption objects are transferred through consumption. Holt (1998) also suggests that cultural capital is essential because consumption in the postmodern world is based on oneÕs understanding of the cultural codes as much as the ability to buy. Micken and Roberts (1999) argue that materialism reflects a quest for certainty of oneÕs identity by using consumption as identity fixers. Thus, this stream of research indicates that there are positive aspects of materialism as well as negative, confirming the characterization of materialism as both instrumental and terminal. From a societal, or macro, perspective, we find both positive and negative aspects of materialism as well. Kassiola (1990) and Tawney (1920) have characterized materialism as the negative consequence of industrialism, while others have argued that it is capitalist economic organization that results in the negative aspects of materialism (Bredemeier & Toby, 1960; Fromm, 1968, 1976; Galbraith, 1972; Heilbroner, 1956). Through the ‘‘ideology of consumption’’ (Hetrick, 1989) that equates material possessions with happiness, consumers are driven to accumulate possessions and organize their lives around the consumption of material goods (Daun, 1983). This produces cultures of consumption in which the normal functions of culture are supplanted by consumption (Featherstone, 1991). The social critiques of consumption by Scitovsky (1976), Slater (1980), Bredemeier and Toby (1960), Fromm (1976), and many others are critiques of these cultures of consumption and have led others, such as Mick (1996) and Burroughs and Rindfleisch (2002) to characterize materialism as the ‘‘dark side’’ of consumption. What the forgoing suggests is that materialism is a complex phenomenon that must be examined from multiple perspectives if it is to be understood. It is individual and social, beneficial and destructive, and domestic and international. Whichever form one considers here, it is clear that materialism is directly related to the value systems of individuals and, because these systems vary across cultures, materialism will vary as well. Because the consequences of materialism are cross-cultural, the research agenda should be expanded to the cross-cultural level as well. This motivates the approach taken here. We propose to examine materialism across cultures and from the perspective of value systems that have been indicated to be related to it.
3. Value systems 3.1. Values measurement Schwartz (1992) and Schwartz and Sagiv (1995) provide a conceptual framework for the study of values that is based on RokeachÕs (1973) value system. They consider values to consist of beliefs about desirable end-states and modes of conduct that guide evaluations and choices. Values exist in a system that is ordered by importance. More importantly, values transcend specific situations, and consequently, measures of values should not pertain to or reference any particular situation. This suggests, for example, that values would serve as guides for consumption behavior, but would not pertain to a specific consumption situation, as this would not be
628
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
consistent with the trans-situational aspect of values. Schwartz (1994) argues that these characteristics distinguish values from attitudes. SchwartzÕs (1994) measurement system, called SchwartzÕs value inventory (SVI), is based on a circumplex model containing ten types of values. The circumplex model characterizes values as being in a circular framework in which value domains adjacent to each other are more similar while those across from each other are different. The ten value types can be organized in a two dimensional system where each of the endpoints of the two dimensions are considered higher order values. One axis has openness to change and tradition as its endpoints while the second axis has selfenhancement and self-transcendence as its endpoints. While Schwartz suggests that his value inventory can be depicted as a two-dimensional framework, much research using it continues to use the four higher order domains as independent measures. This is the approach taken here as well. 3.2. Values and materialism Empirical research on materialism has had to deal with the nature of materialism and its relationship to other constructs. One of the primary issues has been characterizing materialism within a particular framework. The frameworks have included materialism as a set of traits as with Belk (1985) who related it to the personality traits of envy, possessiveness, and nongenerosity. Others such as Moschis and Churchill (1978) and Richins (1987) have measured materialism as an attitude structure focusing on the meaning of possessions to the individual. More recently, materialism has been characterized as a value and studied in that context. Richins and Dawson (1992), for example, argue that materialism is a value that guides conduct in a variety of contexts in addition to consumption. Richins (1994) also examined the relationship between possessions and individual material values, concluding that individuals find different meaning in their possessions depending on their personal values. Thus, the concepts of materialism and values have been examined from different perspectives. Materialism can be seen as part of the individualÕs attitude structure, as a set of traits that reflect personal values, or as a value itself. While it is not clear which of these is the correct characterization of materialism, the Richins and Dawson scale appears to be both situation specific and is based on a Likert scale normally used to measure attitudes. Consequently, we consider it here to be a measure of attitudes rather than a measure of values. Because it is argued by both Rokeach (1973) and Schwartz (1994) that values are of a higher order and more enduring than attitudes and guide their development, we consider the materialism scale to be a measure of attitudes that is influenced by values. Because the SVI was developed in a cross-cultural context and has been demonstrated to be valid across cultures, it is the basis for the value measurement instrument we will use here. It has been used in many contexts including environmentalism (Stern, Dietz, & Guagnano, 1998; Stern, Dietz, Kaloff, & Guagnano, 1995), consumption (Grunert & Juhl, 1995), and materialism (Burroughs & Rindfleisch, 2002). When considering the four value domains as independent constructs, it was found that the values with the greatest predictive ability generally were
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
629
self-transcendence and self-enhancement. In the environmental domain, openness to change was moderately predictive of intentions to behave but much less so than the other variables (Beckmann, Kilbourne, & van Dam, 1998), and Stern et al. (1995, 1998) concluded that self-transcendence and self-enhancement were predictive of intentions but openness was not. They also concluded that, in the environmental domain, tradition was only moderately and inconsistently predictive. Burroughs and Rindfleisch (2002), in the context of values conflict, argue that because materialists have been found to be self-focused (Belk, 1985; Fournier & Richins, 1991), the variable with the greatest predictive validity should be selfenhancement in studies of materialism. Their research supports this argument, but in addition, indicates that self-transcendence and tradition are most distant from materialism. Openness falls between transcendence and tradition on one side and self-enhancement on the other. This suggests that both self-enhancement and selftranscendence should be predictive of materialism, albeit in opposite directions, and openness and tradition should be less predictive with openness slightly more so than tradition. Because of the typical negative correlation between self-enhancement and self-transcendence and its predictive ability in other contexts cited, we chose to use both higher order variables in this study. Because openness and tradition have been shown to be weakly predictive in several contexts, our beginning model of values, shown in Fig. 1, included both domains. The model presented in Fig. 1 suggests two fundamental aspects in the relationship between values and materialism. First, it is argued that the causal flow is from values to materialism consistent with our conclusion that materialism is measured as an attitude constellation rather than a value. Second, it is suggested that, following Roberts et al. (2003), materialism is represented as a second order factor model with success, centrality, and happiness as first order latent constructs. Finally, the model
equality
self-tr admire
peace justice
suc
achieve
wealth success
self-en need
power
cen
simple
mater exciting
important
open
have better happy
authority
variation curious
hap
discipline
tradit
fam_sec parents
Fig. 1. Proposed model of values and materialism.
630
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
suggests that self-transcendence values will be inversely related to materialism, and self-enhancement values will be positively related to materialism. The relationship between tradition and openness values and materialism is less clear. The limited evidence suggests that openness will be positively related to materialism and tradition will be negatively related. In both cases, the relationship will be less significant than the other two value domains. The model will be tested across three countries, Germany, Canada, and the US, using the invariance procedures suggested by Steenkamp and Baumgartner (1998) and Byrne and Campbell (1999). 3.3. Materialism and values across cultures Ger and Belk (1996) examined cross-cultural differences in materialism, comparing economically advanced cultures to less advanced cultures. Their conclusion tended to refute InglehartÕs (1981) thesis that as cultures become more affluent, their focus will shift to less materialist aspects of life. Rather, there was no consistent pattern of materialism between more and less affluent countries. Germany and the US, for example were among the highest of the 13 countries on materialism measures despite the fact that they were among the most affluent in the set. Further, the US was higher than Germany. Feather (1998) drew the same conclusion about Canadians arguing that there was little evidence for post-materialist values despite high levels of affluence. He further suggested that Americans had a stronger emphasis on success, mastery, and achievement and less on equality and harmony than Australians. Canadians fell between the other two. Thus, while all three countries are individualistic, some like Australia are more horizontal than others and favor more equality between individuals, while others are more vertical, like the US, and tolerate more inequality. Canada is difficult to position because it has some characteristic of both of the other countries. From another perspective, Schmuck, Kasser, and Ryan (2000) compared US and German college students examining the degree to which they were intrinsically or extrinsically motivated. Their use of the terms extrinsic and intrinsic parallels SchwartzÕs self-enhancement and self-transcendence domains respectively. They characterize Germans as being less capitalistic and less individualistic than Americans despite the similarities in development between the two countries and conclude that American students are more extrinsically motivated than German students. They further suggest that this leads to more ego-involved activities in Americans. Thus we would hypothesize that American college students would be higher on materialism than would Germans. Canadians, being more like Americans than Germans would likely fall between the two, but closer to the US students. The most frequently used measure of cross-cultural differences is HofstedeÕs (1980). He developed five measures of cultural differences that have been used frequently in cross-cultural research. While subject to much criticism, they are still frequently used and will be used here to further develop hypotheses about the three countries studied. The five measures are power-distance, individualism, uncertainty avoidance, masculinity, and long-termism. The three that relate to the present study indirectly are power-distance, individualism, and uncertainty avoidance. We suggest
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
631
Table 1 HofstedeÕs value measures Country
Power-distance
Individualism
Uncertainty
US Canada Germany
40 39 35
91 80 67
46 48 65
that these three measures roughly parallel self-transcendence, self-enhancement, and tradition/openness respectively. The basis for the argument is that power-distance relates to social equality and justice, which are the basis for the measure of self-transcendence in the SVI. Individualism relates to wealth and power in affluent societies and these are the basis for the self-enhancement domain. Finally, uncertainty avoidance in a culture suggests that tradition is important because it eschews change. Traditional cultures are more predictable and more stable while cultures open to change are less predictable. While the correlation between the constructs is certainly not perfect, they do parallel one another sufficiently to contribute to the development of hypotheses about the three countries under study. HofstedeÕs measures for the US, Canada, and Germany are presented in Table 1.
4. Derived hypotheses HofstedeÕs measures indicate that on the power distance measure, the US and Canada are highest and Germany the lowest, though the difference between them is relatively small. This suggests that on the self-transcendence measure the differences between countries should be small with Germany being the highest. For the individualism measure, Germany is lowest while the US and Canada are similar to each other. Thus, for the self-enhancement measure, Germany should be the lowest with the US and Canada being higher and more similar. Finally, for uncertainty avoidance, Germany is highest and the US and Canada are lower but similar. This suggests that for the openness measure, Germany should be the lowest on openness and highest on tradition both of which lead to greater certainty. The US and Canada should be higher on openness and lower on tradition than Germany. These hypotheses are also suggested by the literature cited above. For the materialism scale, it is hypothesized that because of differences in achievement motivations and power-distance, German students will have lower scores on materialism as a measure of success. This hypothesis is derived from Feather (1998) and Schmuck et al. (2000). On materialism, Canadian students should fall between the US and Germany. Unfortunately, there is insufficient evidence, other than anecdotal, to develop hypotheses about centrality and happiness differences between countries. For the structural equation model being developed, we propose the following hypotheses. Because of its emphasis on equality and justice, the self-transcendence measure will be negatively related to materialism, while self-enhancement, because of its focus on wealth and power, will be positively related. Openness will be
632
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
positively related to materialism because of its focus on variety and excitement which many affluent cultures associate with consumption. Tradition will be negatively related to materialism because of its focus on traditional values that are less related to material consumption.
5. Methods 5.1. Sample and procedures The study consisted of convenience samples of university students from three countries, the USA, Canada, and Germany. The final sample sizes were 139, 168, and 97 respectively, the sample was 46% female, and the median age was 22. While all respondents were students, this was not considered inappropriate because the purpose of the study was to assess the relationship between values and materialism rather than to determine the level of materialism within the group. The latter type of measure might be biased because of the nature of the student sample, but the former should be less dependent on the sample characteristics. This approach had the added benefit of controlling for age differences because the variation in age was small (91% of the sample was between 18 and 30), and this homogeneity is helpful in making comparisons between countries (Ger & Belk, 1996). For the US sample, the questionnaire was distributed in class and the respondents were asked to complete and return it the following week. All participation was voluntary and no class credit was given for participation. For the German sample, the questionnaire was translated into German, as few students were sufficiently fluent in English to respond to all the questions. Standard back translation protocol was used in the translation process. The German sample was also administered as a voluntary assignment without class credit, and students were asked to return the questionnaire over the course of a two-week period. In Canada, the questionnaire was distributed and completed in class under the supervision of a research assistant. Participation was voluntary, and completion of the questionnaire received no class credit or other benefit. 5.2. Measurement instrument The measurement instrument consisted of 57 items from SchwartzÕs value inventory (SVI), nine items from Richins and DawsonÕs (1992) materialism scale, and four demographic variables. There were also 45 items measuring environmental attitudes that were not a part of this study. Because of the large number of items in the questionnaire, only a subset of the materialism scale was used. We chose nine items from the original 18 using the three from each domain that had the highest correlation with the domain. Some scale items were reversed to correspond to the original instrument. The procedure for model development followed that suggested by Steenkamp and Baumgartner (1998) and Byrne and Campbell (1999). We first assessed the
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
633
cross-cultural invariance of the two measurement models to determine if they provided sufficient invariance to accomplish the two objectives of the study. These were comparing the values and materialism means across the countries surveyed and developing a causal model of the relationship between values and materialism. In developing the models, the fit criteria used were the goodness of fit index (GFI), adjusted goodness of fit index (AGFI), Tucker–Lewis index (TLI), comparative fit index (CFI), and the root mean square error of approximation (RMSEA). Standard cutoff criteria were used for each index. In addition, the constrained models were tested with sequential Chi-square difference tests as suggested by Anderson and Gerbing (1988). 5.3. Measurement model for values For the SVI, we used a Structural Equation Model approach, beginning with all the items shown to be valid across cultures for the four higher order domains. We then began reducing them successively to achieve an acceptable fit in all three countries. This was done using modification indices and correlations to remove the items contributing least to the model fit. The final measurement model for values consisted of the three observed variables for each of the four domains. However, the four-factor measurement model specified did not fit the data well when metric invariance was assessed. Each of the fit indices was reduced substantially and sufficient modifications were not available to improve them. Examination of the factor loadings revealed the tradition factor had consistently low correlations so it was removed from the model leaving a three-factor value model. This result was consistent with the cited literature indicating that this factor typically did not predict well. We then reassessed the three-factor value measurement model. Coefficient alphas for the observed variables making up the three latent constructs were 0.64 for self-transcendence, 0.71 for self-enhancement, and 0.62 for openness. In assessing the configural invariance for the three factor model, each of the items loaded significantly onto their respective factors (p < 0.01) indicating convergent validity. The correlation between factors was less than unity indicating discriminant validity. The final fit statistics for each stage of the analysis are presented in Table 2. The model was shown to possess configural, metric, and partial (co)variance invariance (Byrne & Campbell, 1999; Steenkamp & Baumgartner, 1998). The covariance from self-transcendence to openness was not invariant and was freed to vary in the analysis. This suggests that the correlations between those factors cannot be compared from the model. Because this was not the primary purpose of the research, the analysis continued. In order to compare latent factor means across countries it was necessary to demonstrate scalar invariance. This was accomplished by constraining the intercepts to be equal for the observed variables and comparing the resulting model to the original configurally invariant model. The results indicated that the model was partially scalar invariant. This is sufficient for comparing means so long as two items in each construct are invariant (Steenkamp & Baumgartner, 1998). This condition was achieved enabling the comparison of the latent factor means.
634
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
Table 2 Fit statistics for values measurement model Invariances
v2
df
GFI
AGFI
TLI
CFI
RMSEA
SCDT
Configural Metric Full metric Scalar Partial scalar Covariance Partial covariance Variance Partial variance
87 114 106 146 121 124 111 125 117
70 82 81 92 89 87 84 90 89
0.96 0.94 0.95 na na 0.94 0.94 0.94 0.94
0.91 0.90 0.91 na na 0.90 0.91 0.91 0.91
0.96 0.94 0.95 0.99 0.99 0.93 0.95 0.94 0.95
0.98 0.95 0.96 0.99 0.99 0.95 0.96 0.95 0.96
0.025 0.031 0.028 0.030 0.030 0.044 0.028 0.031 0.028
0.01a 0.06 <0.01b 0.02 <0.01c 0.05 0.01d 0.05
a b c d
Free Free Free Free
error covariance from peace to exciting for Canada. curious and authority for Germany and peace for the US. covariance from self-transcendence to openness. variance of self-enhancement for Germany.
5.4. Measurement model for materialism A similar approach was used for the invariance analysis of the materialism scale. We began with the measurement instrument as a three-factor model (success, centrality, and happiness) with three observed variables in each factor. Coefficient alphas for the three sets of observed variables were 0.67, 0.67, and 0.70 respectively. The measurement model for materialism was shown to possess configural, metric, variance, and covariance invariance across the countries. Full scalar invariance was again assessed by constraining the intercepts to be equal across countries. The result of this analysis indicated that the constrained model did not fit the data well as the SCDT was <0.01. Modification indices suggested freeing both ‘‘admire’’ and ‘‘need’’ to get a better fit. This was done and the resulting model fit the data well. The SCDT for the model yielded a p-value of 0.07 which was acceptable. This indicated that the means of the latent constructs could be compared across the countries in the sample. The complete set of fit statistics for this assessment is presented in Table 3. After the invariance assessments were completed and the model was demonstrated to fit the data sufficiently, we tested the final model to determine if the second order factor model of materialism suggested by Roberts et al. (2003) was appropriate in this cross-cultural context. The second order model was tested and found to meet the criteria specified with the exception of the SCDT = 0.04 which was slightly below the 0.05 level desired. Steenkamp and Baumgartner (1998) argue that a lower SCDT is acceptable if all the other fit statistics are acceptable. Because the SCDT was only slightly below the 0.05 criterion and all other fit indices were well within acceptable ranges, the assessment continued. We next tested path invariance for the second order materialism model to see if it was invariant across the countries in the sample. We constrained the paths to be equal and compared this fit to the original model. This resulted in acceptable fit indices and a SCDT of 0.05. This indicated that the materialism model was best repre-
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
635
Table 3 Fit statistics for materialism measurement model Invariances
v2
df
GFI
AGFI
TLI
CFI
RMSEA
SCDT
Configural Metric Scalar Variance Covariance Partial covariance Second order Invariance
135 153 160 164 174 168 158 164
73 85 89 91 97 96 86 91
0.93 0.92 na 0.92 0.91 0.92 0.92 0.92
0.87 0.88 na 0.88 0.88 0.88 0.88 0.88
0.91 0.91 0.099 0.91 0.91 0.92 0.091 0.91
0.94 0.93 0.099 0.92 0.92 0.93 0.93 0.92
0.046 0.045 0.045 0.045 0.044 0.043 0.046 0.045
0.12 0.07a 0.05 0.03 0.08b 0.04 0.05
Note: Haywood cases were assigned a value of 0.01. a Free intercepts for admire and need. b Free covariance centrality ! happiness for Germany.
sented as a second order factor model with success, centrality, and happiness as first order factors and materialism as the second order factor. The final fit statistics for the values measurement model were acceptable with GFI = 0.94, AGFI = 0.91, TLI = 0.95, CFI = 0.96, and RMSEA = 0.028. The model was shown to be a three factor model that was invariant across countries. The three latent factors were self-transcendence, self-enhancement, and openness. The final fit statistics for the second order materialism model were acceptable with GFI = 0.92, AGFI = 0.88, TLI = 0.91, CFI = 0.92, and RMSEA = 0.045. At each stage of the analysis for both measurement models, SCDTs were performed. The SCDT statistics ranged from 0.04 to 0.12 indicating that the constrained models were not significantly different from the baseline model as required (Anderson & Gerbing, 1988). This indicates that latent construct means can be compared for the three countries, and that we could proceed with the development of the structural model proposed in Fig. 1 with the exception that the values measurement model has three latent constructs. Because tradition has been shown to be only weakly predictive, this was not considered to be a significant deficiency. In the final materialism model, the importance of the three latent constructs varied considerably. The factor with the highest path coefficient for all three countries was success. The second most important was happiness while centrality was the weakest relationship though still highly significant (p < 0.01). This pattern of significance was consistent across countries.
6. Results 6.1. Latent factor means While structural equation modeling does not allow for the direct comparison of all means to each other, it does allow for the comparison of all means to a specified base mean. Like regression models with dummy variables, one sample must be
636
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
specified as the base against which the others are compared. Because the hypotheses more consistently represent Germany as the sample that might be different, we chose to use it as the base for comparison. For the self-transcendence domain, the means for the three countries were the same when considering traditional p-values. This was not unexpected because the three countries were similar to each other on HofstedeÕs power distance measure. For self-enhancement, the US and Canadian students were significantly higher (p < 0.01) than German students in the sample. This is consistent with the hypothesis derived. For the openness measure, Canadian students were significantly lower than German students (p < 0.055), but the US students fell in between the others and were not different from either German or Canadian students. For the materialism measures, we again specified Germany as the base against which to compare the others. The results of this analysis indicate that for the success measure, US and Canadian students were significantly higher (p < 0.01) than German students as hypothesized. US and Canadian students were not different from each other. For both the centrality and happiness dimensions of materialism, there were no differences between the three samples. 6.2. Full model The final stage of the analysis required that the structural model proposed be tested. The final structural model in Fig. 2 was tested first for overall fit across countries. The results of this analysis indicated that the model fit the data with GFI = 0.88, AGFI = 0.83, TLI = 0.90, CFI = 0.91, and RMSEA = 0.033. To test structural invariance, the links from self-transcendence, self-enhancement, and openness to materialism were constrained to be equal for the three countries. This model also achieved an acceptable fit with GFI = 0.87, AGFI = 0.83, TLI = 0.90, CFI = 0.91, and RMSEA = 0.033. Thus, the fit statistics for the constrained model
equality
admire achieve
self-tr
suc
justice
success
(-) wealth
need simple
cen
mater
(+)
self-en
authority power
important have better
peace
(0)
hap
exciting
open
happy
variation curious
+ indicates positive relationship - indicates negative relationship 0 indicates no relationship
Fig. 2. Final model of values and materialism.
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
637
Table 4 Standardized path coefficients for structural model US Self-transcend Self-enhance Openness a b
Canada a
0.21 0.63b 0.05
b
0.39 0.62b 0.11
Germany 0.09 0.90b 0.07
p < 0.10. p < 0.01.
were not reduced by imposing the constraints across countries. In addition, the SCDT test yielded a p-value of 0.13 indicating that the constrained model predicted as well as the original model. The only fit index that was lower than desired was the GFI which was slightly lower than 0.90. However, because the other indices, particularly the RMSEA and the AGFI, were in acceptable ranges, we concluded that the structural model was invariant across countries as predicted. The results of the analysis of the structural equation model partially confirmed the hypotheses. The self-enhancement domain had the strongest impact on materialism for all three countries. As hypothesized, the path coefficients, presented in Table 4, were positive and highly significant (p < 0.01) in all three countries. While the constrained model indicated that the path coefficient for self-transcendence was significant (p < 0.01), examination of the individual standardized coefficients indicated that most of the parameter is accounted for by the US (p < 0.08) and Canada (p < 0.01). The path coefficient for Germany was not significant. Thus, while the hypothesized relation did materialize, it was accounted for by two of the three countries only. For the openness path coefficients, no significant relationships were found. This was not unexpected because previous research indicated that the results of this variable were inconsistent in different contexts.
7. Conclusions There are two main conclusions to be drawn from the study. While the SVI has been shown to be valid across countries, the Richins and Dawson (1992) materialism scale has not. The results of the analysis indicate that the reduced form of the scale, with three indicator variables for each latent construct, produces a valid second order factor model of materialism. The latent constructs are happiness, centrality, and success and the model was shown to possess configural, metric, scalar, and (co)variance invariance for German, American, and Canadian college students. This confirms that the measure has useful application across cultures and that the reduced form containing only nine of the original items provides a useful measure of materialism as an attitude structure. The second objective of the study was to determine the causal sequence for values and materialism assuming that the materialism scale is a measure of attitudes toward consumption rather than a true value labeled as materialism. The hypothesized causal sequence indicates that values affect materialism as in Fig. 1. In addition to this,
638
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
the causal model was demonstrated to be invariant across all three countries. The relationship between self-transcendence values and materialism was negative, and the relationship between self-enhancement values and materialism was positive as hypothesized. As with previous research, tradition and openness are tenuous in the relationship. While the materialism scale was shown to be invariant across the cultures examined, success was the latent construct with the highest path coefficient. Centrality and happiness, while significant, were less so. It was also shown that students in the different countries had the largest differences on the success measure. Thus, success is the most important construct in the materialism model and it is the construct in which there were the greatest group differences. Clearly the strongest indicator of materialism was self-enhancement. Examining the indicator variables in this domain suggests this result because they focus on wealth and power and, within affluent societies, consumption is most often used as an indicator of success. Schmuck et al. (2000) argue that the German focus on intrinsic rather than extrinsic goals suggests that they might be less materialistic, and this is also seen in the research showing that US and Canadian students associate materialism with success more than Germans. However, all three groups had relatively high measures of materialism supporting FeatherÕs (1998) contention that post-materialist values are not becoming more evident in affluent cultures, and that all individualist cultures are not equally horizontal. Self-enhancement values are strong in all three cultures, but more so in the US and Canada. From the perspective of self-transcendence, the results indicate that both American and Canadian students do not associate equality and justice with materialism. They are also less concerned about equality than are German students. The statistical evidence suggests that German students do not associate equality and justice as matters of materialism. An explanation for this result is that German students, because their culture is more horizontal, do not see material possessions as a distinguishing factor in social institutions of society as much as the American and Canadian students. While all three groups were equally self-transcendent, markers of success for Germans would be intrinsic rewards relating more to personal growth than material status as suggested by Schmuck et al. (2000). Americans, for example, frequently equate equality with equality of possessions (Bredemeier & Toby, 1960). Thus, it appears that value systems are related to attitudes toward materialism. While materialism has been studied in very different contexts in the past, and its effects have been shown to vary by context, its relationship to values has only limited support. We have demonstrated that values are related to materialism, and that the effect varies by country. This suggests that future research in materialism consider the context of values, and that it take a multi-national approach. We would have a better understanding of this phenomenon if, for example, cultural values were shown to be converging with globalization as suggested by some researchers. We suggest that both the values context and multi-national samples should provide us with a better understanding of increasing materialism as a factor in globalization. The results of this study provide support for the study of values and materialism, and indicate that future research on materialism should consider individual values
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
639
and take a multi-national approach. The SVI and the Richins and Dawson materialism scale provide an adequate means through which to carry out such a research agenda. The samples from the three countries were limited to students which increased internal validity. While reasonably high levels of materialism were found in each sample, these measures might be even larger with an older and more diverse population with higher incomes. Future studies should focus on such diverse samples to confirm this conjecture. A notable limitation of the study, as with all such studies using correlational methods, is that the nature of the relationship tested is problematic. While we conclude that individual values affect oneÕs attitudes toward materialism, it cannot be stated unequivocally. It is also possible that, as a culture becomes more materialistic through trade liberalization, for example, individual values will change to become more self-enhancing and less self-transcendent. Such causation can only be determined through experimentation that is not feasible as a research method. References Anderson, J. C., & Gerbing, D. W. (1988). Structural equation modeling in practice: A review and recommended two-step approach. Psychological Bulletin, 103, 411–423. Beckmann, S. C., Kilbourne, W. E., & van Dam, Y. (1998). Anthropocentrism, value systems, and environmental attitudes: A multi-national comparison. In P. Andersson (Ed.), Proceedings of the 27th European marketing academy conference (pp. 1807–1813). Stockholm, Sweden. Belk, R. W. (1985). Materialism: Trait aspects of living in the material world. Journal of Consumer Research, 12(December), 265–280. Bredemeier, H. C., & Toby, J. (1960). Social problems in America: Costs and casualties in an acquisitive society. New York: Wiley. Browne, B. A., & Kaldenberg, D. O. (1997). Conceptualizing self-monitoring: Links to materialism and product involvement. Journal of Consumer Marketing, 14(1), 31–44. Burroughs, J. E., & Rindfleisch, A. (2002). Materialism and well-being: A conflicting values perspective. Journal of Consumer Research, 29(December), 348–370. Byrne, B. M., & Campbell, T. L. (1999). Cross cultural comparisons and the assumption of equivalent measurement and theoretical structure: A look beneath the surface. Journal of Cross-Cultural Psychology, 30(5), 555–574. Daun, A. (1983). The materialistic life-style: Some socio-psychological aspects. In L. Uusitalo (Ed.), Consumer behavior and environmental quality (pp. 6–16). New York: St MartinÕs Press. Feather, N. T. (1998). Attitudes toward high achievers, self-esteem, and value priorities for Australian, American, and Canadian students. Journal of Cross-Cultural Psychology, 29(6), 749–760. Featherstone, M. (1991). Consumer culture & postmodernism. London: Sage Publications. Fournier, S., & Richins, M. L. (1991). Some theoretical and popular notions concerning materialism. Journal of Social Behavior and Personality, 6(6), 403–414. Fromm, E. (1968). The revolution of hope: Toward a humanized technology. New York: Perennial Library. Fromm, E. (1976). To have: Or to be? New York. Harper and Row. Galbraith, J. K. (1972). The new industrial state. London, UK: Andre Deutsch. Ger, G., & Belk, R. W. (1996). Cross-cultural differences in materialism. Journal of Economic Psychology, 17(1), 55–77. Grunert, S. C., & Juhl, H. J. (1995). Values, environmental attitudes, and buying of organic foods. Journal of Economic Psychology, 16(1), 39–62. Heilbroner, R. (1956). The quest for wealth; a study of acquisitive man. New York: Simon and Schuster. Hetrick, W. (1989). The ideology of consumerism: A critique. In R. Bagozzi, & J. P. Peter (Eds.), Proceedings of the American Marketing Association winter conference (pp. 287–296). Orlando, FL.
640
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
Hofstede, G. (1980). Culture’s consequences: International differences in work-related values. Beverly Hills, CA: Sage Publications. Holt, D. B. (1997). Poststructuralist lifestyle analysis: Conceptualizing the social patterning of consumption in postmodernity. Journal of Consumer Research, 23(March), 326–350. Holt, D. B. (1998). Does cultural capital structure American consumption? Journal of Consumer Research, 25(June), 1–25. Inglehart, R. (1981). Post-materialism in an environment of insecurity. American Political Science Review, 75, 880–900. James, W. (1890). The principles of psychology. New York: H. Holt and Company. Kasser, T., & Ryan, R. M. (1993). A dark side of the American dream: Correlates of financial success as a central life aspiration. Journal of Personality and Social Psychology, 65(August), 410–422. Kassiola, J. (1990). The death of industrial civilization. Albany, NY: State University of New York Press. McCracken, G. (1988). Culture & consumption. Bloomington, IN: Indiana University Press. Mick, D. G. (1996). Are studies of dark side variables confounded by socially desirable responding? The case of materialism. Journal of Consumer Research, 23(September), 106–119. Micken, K. S., & Roberts, S. D. (1999). Desperately seeking certainty: Narrowing the materialism construct. In E. Arnould, & L. Scott (Eds.), Proceedings of the association for consumer research (pp. 513–518). Moschis, G. P., & Churchill, G. A. (1978). Consumer socialization: A theoretical and empirical analysis. Journal of Consumer Research, 15(November), 599–609. Muncy, J. A., & Eastman, J. K. (1998). Materialism and consumer ethics: An exploratory study. Journal of Business Ethics, 17(2), 137–145. Norgaard, R. B. (1995). Beyond materialism: A coevolutionary reinterpretation of the environmental crisis. Review of Social Economy, 54(4), 475–493. Rassuli, K. M., & Hollander, S. C (1986). Desire – induced, innate, insatiable? Journal of Macromarketing, 6(Fall), 4–24. Richins, M. L. (1987). Media, materialism, and human happiness. In M. Wallendorf, & P. Anderson (Eds.), Proceedings of the advances in consumer research (pp. 352–356). Provo, UT. Richins, M. L. (1994). Special possessions and the expression of material values. Journal of Consumer Research, 21, 522–533. Richins, M. L., & Dawson, S. (1992). A consumer values orientation for materialism and its measurement: Scale development and validation. Journal of Consumer Research, 19(December), 303–316. Rindfleisch, A., Burroughs, J. E., & Denton, F. (1997). Family structure, materialism, and compulsive consumption. Journal of Consumer Research, 23(March), 312–325. Roberts, J. A., Manolis, C., & Tanner, J. F. (2003). Family structure, materialism, and compulsive buying: A reinquiry and extension. Journal of the Academy of Marketing Science, 31(3), 300–311. Rokeach, M. (1973). The nature of human values. New York: Free Press. Rudmin, F. W., & Kilbourne, W. E. (1996). The meaning and morality of voluntary simplicity: History and hypotheses on deliberately denied materialism. In R. W. Belk, N. Dholakia, & A. Venkatesh (Eds.), Consumption & marketing: Macro dimensions (pp. 166–215). Cincinnati, OH: South-Western College Publishing. Schmuck, P., Kasser, T., & Ryan, R. M. (2000). Intrinsic and extrinsic goals: Their structure and relationship to well-being in German and US College students. Social Indicators Research, 50, 225–241. Schouten, J. W., & McAlexander, J. H. (1995). Subcultures of consumption: An ethnography of the new bikers. Journal of Consumer Research, 22(June), 43–61. Schwartz, S. H. (1992). Universals in the content and structure of values: Theoretical advances and empirical tests in 20 countries. In M. Zanna (Ed.), Advances in experimental social psychology (pp. 1–65). Orlando, FL: Academic. Schwartz, S. H. (1994). Are there universal aspects of the structure and contents of human values? Journal of Social Issues, 50(4), 19–45. Schwartz, S. H., & Sagiv, L. (1995). Identifying culture-specifics in the content and structure of values. Journal of Cross-Cultural Psychology, 26(1), 92–116. Scitovsky, T. (1976). The joyless economy. New York: Oxford University Press.
W. Kilbourne et al. / Journal of Economic Psychology 26 (2005) 624–641
641
Slater, P. E. (1980). Wealth addiction. New York: Dutton/Plume. Steenkamp, J.-B. E. M., & Baumgartner, H. (1998). Assessing measurement invariance in cross-national research. Journal of Consumer Research, 25(June), 78–90. Stern, P. C., Dietz, T., & Guagnano, G. A. (1998). A brief inventory of values. Educational & Psychological Measurement, 58(6), 984–1001. Stern, P. C., Dietz, T., Kaloff, L., & Guagnano, G. A. (1995). Values, beliefs, and proenvironmental action: Attitude formation toward emergent attitude objects. Journal of Applied Social Psychology, 25(18), 1611–1636. Tawney, R. H. (1920). The acquisitive society. New York: Harcourt, Brace and Howe.