Editor’s note: Melvin Prince is president of Prince Associates, a Darien, Conn., research firm, and professor of marketing at Southern Connecticut State University, New Haven, Conn.

Research method

Descriptions of prestige brands were developed by a panel of marketing faculty judges who agreed on a subset of descriptive items that specifically indicated concepts of brand excellence and status conferral.

Survey respondents were exposed to four brands within a single product category. They were invited to state which of a series of brand descriptions especially applied to specified brands in the category under investigation. Each of the indicators was evaluated for its contribution to the latent class scales for brand excellence and status conferral, based on its significance for the model fit. Significance was tested by means of the maximum likelihood ratio chi-squared statistic (L2) with and without the indicator. The items were analyzed for the fit to latent dimensions, in order to assess the discriminant validity of the constructs. Questionnaires also contained a seven-point rating scale to obtain overall prestige ratings for the same brands presented. These prestige ratings were used as the criterion for external validation of the multidimensional scale.

Product categories in the present study were diverse, consisting of what judges viewed as prestige brands of golf equipment and wine. Brands of golf equipment analyzed in the study were TaylorMade and Titleist. Brands of wines analyzed were Kendall-Jackson and Robert Mondavi.

Data analysis used exploratory latent class factor analysis a) to explain correlations between nominal indicators of brand excellence and status conferral, and b) to classify consumers by factor levels for the constructs assigned to prestige brands. The analysis was replicated for two brands in each of the two categories, to test for scale test-retest reliability. Each product category was presented to an independent sample. The latent class factor analysis used Version 4 of the Latent GOLD computer program (Vermunt and Magidson, 2005).

The study was conducted within a 25-mile radius of a public university located in the Northeast. The sample frame consisted of adults, 25-49, who were aware of the prestige brands to which they were exposed. Purposive sampling was employed from September 2005 to April 2006. Two surveys were conducted: n=120 for prestige golf equipment brands and n=130 for prestige wine brands. Personally conducted interviews were obtained from respondents who were qualified and agreed to be interviewed.

Results

Overall model fit

In a first stage of confirmatory latent class factor analysis, a two-factor measurement model was used to test for inherent scalability. This confirmatory model constrained indicators unrelated to a concept to zero values. Probabilities from this analysis indeed demonstrated the existence of a two-factor model fit. The latent class confirmatory factor analysis was then replicated with a latent class factor solution involving no constraints on indicators. This unconstrained model provided a slightly better fit. Unconstrained results are shown in Table 1.

Looking at the table, it’s clear that the two-factor solution also consistently fits the measured brands. The fit for the brands is may be seen by the relative correspondence of degrees of freedom and L2 values. A more rigorous statistical test completes the argument. Bootstrap L2 values are calculated to minimize the effects of sparse data. For all brands, bootstrap p-values >.05 were found. This robust result provides strong and reliable evidence of a scalable model involving two latent factors.

Factor structure

Factor loadings for the two-factor solution are shown in Table 2. For all brands, highest (“products meet the highest standards”), strong reputation (“has a strong reputation for excellence”) and quality (“known for superior quality products”) load primarily on factor 1, while envied (“brand users are envied by others”) and feel good (“brand users feel good that others cannot afford it”) load primarily on factor 2. Interpretatively, factor 1 is labeled brand excellence, and factor 2 status conferral.

To illustrate the factor structures, TaylorMade prestige golf equipment loadings for highest, strong reputation and quality are .66, .64 and .61 on factor 1 vs. loadings of less than .30 for factor 2. Kendall-Jackson prestige wine loadings for envied and feel good are .82, and .68 on factor 1 vs. loadings of less than .20 on factor 2.

Communalities (R2) of indicators are consistently significant for the four brands. As cases in point, these values for the contributions of factors to the variance of each of the indicators range from .39 to .62 for Titleist prestige golf equipment, and from .29 to .79 for Robert Mondavi prestige wines. Finally, the explained total variance of each of the two factors is substantial for each brand, although it differs somewhat between product categories. Explained variances for prestige golf equipment for factors 1 and 2 are 55.6 percent and 44.4 percent (TaylorMade), 56.5 percent vs. 43.5 percent (Titleist). For prestige wines, explained variances for the respective factors are 47.7 percent vs. 52.3 percent (Kendall-Jackson) and 36.3 percent vs. 63.6 percent (Robert Mondavi).

Factor comparisons were made between pairs of prestige brands to assess reliabilities of comparative indicator loadings. Correlations of factor loadings were calculated for all five pairs of brands, regardless of product category. In every case the correlation was very strong, with p< .05. From Table 3 we see correlations between Taylor vs. Titleist, Taylor vs. Kendall-Jackson, Taylor vs. Robert Mondavi, Titleist vs. Kendall-Jackson and Titleist vs. Robert Mondavi scoring .81, .73, .72, .91 and .73, respectively.

Predictive validity

Factor scores for each brand rated are assigned to respondents. An overall brand prestige rating given on a seven-point Likert scale is used as the criterion of predictive validity for the two subscales of brand excellence and status conferral. These brand prestige ratings are regressed on the subscales, by multiple stepwise regressions (Table 4). The regression fit is significant for each of the prestige golf and prestige wine brands, with p < .01.

Classification

Each respondent is also classified for each brand by levels of brand excellence (factor 1) and status conferral (factor 2). These levels are ascribed based on the respondent’s response pattern and the associated probabilities for each level of each factor. There are four types: 1) utterly banal: low brand excellence, low status conferral; 2) baselessly ostentatious: low brand excellence, high status conferral; 3) unaffectedly fulfilling: high brand excellence, low status conferral; and 4) irrefutably prestigious: high brand excellence, high status conferral. Segments of respondents for these types for each brand are shown in Table 5.

For TaylorMade and Titleist golf equipment brands, the largest segments are types of consumers who see the brands as utterly banal (low on both dimensions, 52-54 percent). Differences in segments between TaylorMade and Titleist are seen for the baselessly ostentatious type (low brand excellence and high status conferral classes, 22 percent vs. 12 percent). For the unaffectedly fulfilling type (high brand excellence and low status conferral) it is the other way around, 18 percent vs. 27 percent. However, for both brands, the smallest type found is that of respondents who score the brands irrefutably prestigious (high on both dimensions, 6-7 percent).

For Kendall-Jackson and Robert Mondavi prestige wine brands, the most frequent types also see the brands as utterly banal (scoring the brands low on both dimensions, 42-33 percent). Differences in prevalence between Kendall-Jackson and Robert Mondavi are seen for the baselessly ostentatious type (low brand excellence and high status conferral classes, 12 percent vs. 19 percent). For unaffectedly fulfilling type (high brand excellence and low status conferral) segments the reverse is true, 36 percent vs. 30 percent. For both brands, the smallest type found is the irrefutably ostentatious type (respondents who score the brands high on both dimensions). The difference is noteworthy (10 percent vs. 17 percent). In sum, classification and cluster assignment appear to be quite sensitive to measured brands and product categories.

Building blocks

Brand prestige constructs can be building blocks embedded in more comprehensive models of prestige brand motivation and consumption. For example, causal relations between such variables as consumer sensitivity to status symbols, prestige brand values and consumer terminal values, e.g., hedonism, belonging, etc., may be explored in future research.

The present research on scale development was based on two quite different product categories: wine and golf equipment. While seemingly robust, the scale may be further tested and validated on prestige brands in a number of other product categories, such as fashionable lingerie, upscale automobiles, super-premium spirits, luxury food and beverage suppliers and higher-end appliances. In the course of such testing and continual development, the instrument may be refined and upgraded with new items, measurement levels and techniques. The generalizability of the scale should be established by conducting research that includes demographic and behavioral segments, as well as other geographic locations where cultural differences may be at work.

Finally, the scale should be utilized in real-world settings where marketing problems of prestige brands need to be solved. Strategic and tactical applications will permit the ultimate validation test. They also will generate new, important streams of research on this critical and neglected area. 

References

Dubois, B., Laurent, G. “Luxury Possessions and Practices: an Empirical Scale.” European Advances in Consumer Research 1995;2;69-77.

Eastman, J. K., Goldsmith, R. E., Flynn, L. R. “Status Consumption in Consumer Behavior: Scale Development and Validation.” Journal of Marketing Theory and Practice 1999; 41-52 [summer].

Holbrook M. “Introduction to Consumer Value.” In Holbrook M. editor. Consumer Value: A Framework For Analysis and Research. New York: Routledge; 1999. p. 1-27.

Magidson, J. and Vermunt, J. “Latent Class modeling as a probabilistic extension of k-means clustering.” Quirk’s Marketing Research Review, March 2002.

Oliver, R. L. “Value as Excellence in the Consumption Experience.” In Holbrook M. editor. Consumer Value: A Framework For Analysis and Research. New York: Routledge; 1999. 43-61.

Solomon, M. R. “The Value of Status and the Status of Value.” In Holbrook M. editor. Consumer Value: A Framework For Analysis and Research. New York: Routledge; 1999. 63-83.

Vermunt J.K., Magidson J. Technical Guide for Latent Gold 4.0: Basic and Advanced. Belmont (Mass.): Statistical Innovations Inc.; 2005.

Vigneron, F., Johnson, L. W. “A Review and a Conceptual Framework of Prestige-Seeking Consumer Behavior.” Academy of Marketing Science Review 1999;1; 15 pp.

Vigneron, F., Johnson, L. W. “Measuring Perceptions of Brand Luxury. Brand Management 2004; 11(6); 484-506.