Is Brand Evaluation Independent of Other Brands?

Shuzo Abe, Yokohama National University
Masao Tanaka, Aoyamagakuin University
ABSTRACT - Normally, in consumer information processing research, overall evaluation of a brand is implicitly assumed to be independent of other brands. This paper questions this assumption of the inter-brand independence and empirically tests the validity of such assumptions. By extending the Multi-Attribute Attitude Model so as to includes the S Biai of other brands as well as the S Biai of the concerned brand, the influence form other brands' knowledge in brand evaluation was investigated. The results indicate that brand evaluation may be interdependent between or among brands.
[ to cite ]:
Shuzo Abe and Masao Tanaka (1989) ,"Is Brand Evaluation Independent of Other Brands?", in NA - Advances in Consumer Research Volume 16, eds. Thomas K. Srull, Provo, UT : Association for Consumer Research, Pages: 439-442.

Advances in Consumer Research Volume 16, 1989      Pages 439-442

IS BRAND EVALUATION INDEPENDENT OF OTHER BRANDS?

Shuzo Abe, Yokohama National University

Masao Tanaka, Aoyamagakuin University

ABSTRACT -

Normally, in consumer information processing research, overall evaluation of a brand is implicitly assumed to be independent of other brands. This paper questions this assumption of the inter-brand independence and empirically tests the validity of such assumptions. By extending the Multi-Attribute Attitude Model so as to includes the S Biai of other brands as well as the S Biai of the concerned brand, the influence form other brands' knowledge in brand evaluation was investigated. The results indicate that brand evaluation may be interdependent between or among brands.

INTRODUCTION

One important aspect of consumer information processing is that it is related to choice (Bettman 1979, Hansen 1976, Payne 1982). Consumers process information in order to achieve their goals, i.e., to choose one alternative from a number of alternatives. Marketers provide competitive information so as to make consumers form favorable Attitude toward their brand and finally purchase their brand. In consumer information processing research, this competitive aspect of information processing is treated at various stages of processing. For instance, when a consumer uses choice heuristics like lexicographic rule or additive difference rule, the comparison is done at the belief level and the formation of overall evaluations of brands are not necessary for brand comparison. However, when a consumer uses compensatory choice heuristics, the overall evaluations of brands must be formed before the comparison process begins. A good example is the linear compensatory choice heuristic known as Multi-Attribute Attitude Model. When the MultiAttribute Model is used, basically, two approaches are implicitly used in dealing with this choice aspect. One is the approach that assumes that a consumer chooses the brand which has the highest score of Attitude among alternative brands. The other approach is the introduction of an intervening variable Intention, which is thought to be very close to the concept of probability (Ajzen and Fishbein 1980; Fishbein and Ajzen 1975).

It is clear that both approaches share the basic assumption that competitive aspect of alternative brands become relevant only after the formation of overall evaluation, i.e., Attitude. However, this assumption seems to be rather strong assumption under the competitive environment of consumer information processing. It seems to be unrealistic to rule out the possibility that the overall evaluation of a brand is affected by the knowledge of other brands. It may even be possible that the knowledge of each brands is interrelated between or among other brands. For instance, if the size of a newly introduced brand is so compact, it may make the other brands look comparatively large and consequently may lower the overall evaluation of other brands. If this is the case, then a change of a belief on one brand toward the favorable direction affects the choice not only through the increase of overall evaluation score of that brand but also through the decrease of overall evaluation scores of other brands.

This study tries to shed light on this point, i.e., the interdependence of brand evaluation.

RESEARCH HYPOTHESES AND METHOD

First hypothesis in this study is S Biai of a brand is positively related to Ao (i.e., Attitude toward f the brand), while S Biais of other brands are i negatively related to Ao or Aact." Where Bi is the belief that the brand possesses the attribute i, and ai is the evaluative aspect of attribute i. Since brands in ) a market are in competitive relationships, the high 0 score of S Biai on one brand is thought to lower the overall evaluation of other brands. these 2 hypothesized inter-brand relationships are though to be the most prominent ones and therefore the most easily discernible ones, if any inter-brand relationships exist at all. For, even the inter-brand relationships are likely to exist in the form of causal flow rather than a simultaneous interdependence form [One of the reviewers pointed out that there might be negative correlations between S Biais of different brands and between Attitudes of different brands. We calculated the simple correlation coefficients to test this simultaneous inter-brand relationships. The results were either positive or no correlations, so they were not in the direction which inflates our hypothesized inter-brand relationships. The aforementioned simultaneous inter-brand relationships at the level of individual beliefs are difficult to discern by a nonexperimental study like this, so it is not tested here.], and the inter-brand relationships at the overall evaluation stage are the ones which should be tested first before delving into the more complicated interbrand relationships at individual belief level.

Second hypothesis is "The effects from other brands are more evident in the case of Aact than in the case of Ao." In general, it is pointed out that Ao is less specific about the purchasing behavior and is somewhat less useful in explaining the purchasing behavior than Aact (Fishbein 1967). For example, a person who has very favorable Attitude toward Chadillac can have rather unfavorable Attitude toward purchasing a Chadillac, because Attitude toward buying a Chadillac reflects a number of specific factors like high price of the car, high maintenance costs and lack of parking space. This indicates that the concept of Aact is closer to Behavior or Behavioral Intention than the concept Ao is. Thus, to the extent Aact is closer to Behavior, it is thought to reflect more about the knowledge of other brands. For, as indicated above, Behavior or Behavioral Intention in choice setting is meaningless without incorporating the information on other brands. Here, one caveat is necessary. Normally, when Aact is used, Bi and ai are redefined so that they may be meaningful in the context of Aact. However, in this study we used the same Bi and ai for both Ao and Aact in order to make the comparison between Ao and Aact easier.

Margarine was used as the choice object. The major reason we chose margarine was that three leading brands altogether enjoy about seventy percent of margarine market in Japan. Therefore, most of the consumers were thought to be familiar with three brands. the estimated market share for each of three margarine (denoted here by brand N, brand R, and brand M) was 37%, 17%, and 16%, respectively.

Based upon an exploratory study on twenty consumers, we selected seven relevant attributes for choosing margarine. They were (1) ease of spreading, (2-) good taste, (3) economy, (4) good for health, (5) tasty looking color, (6) convenience of container, and (7) nutrition. All of these attributes have positive values, so we used seven point uni-polar impotance scales to measure the evaluative aspects of attribute. Beliefs were measured on a semantic differential type scales.

For measuring Ao following two scales were utilized:

As a whole, margarine S is

very good ----:----:--:--:----:----:----very bad

very desirable ----:--:--:----:----:--:--very undesirable

"act was measured on similar two scales with a slightly different preceding words as "As a whole purchasing margarine X is".

928 housewives of family size of two or more were randomly extracted form Tokyo, Kyoto, Osaka, and their vicinities. Data were collected by personal interview method during September 1985. This interview covered several products among which margarine was the one for this research purpose. Of 938 housewives, 138 were absent and 230 refused to corporate, therefore, the remaining 260 housewives answered the questionnaire. Two forms of questionnaire were used on a split run basis, i.e., one for collecting data on Ao and the other for collecting data on Aacts Of those who responded, we further limited the sample respondents to those whose evoked set plus hold set included all three brands of concern. If any one of these three brands belonged to foggy set, unawareness set, or reject set, then the respondent was excluded form the sample (Laroche et. al., 1983). In other words, the respondents who knew little about the three brands or who held very strong negative Attitude toward these brands were excluded from the analysis. The reason we excluded reject set was that it might have distorted the competitive relationships between or among brands due to the exceptional nature of reject set. Ideally, the analysis should have been restricted to the relationships between or among brands in evoked set, however we added hold set in view of the sample size. The effective sample sizes for each of the three brands with Ao and Aact separately are shown in the Table. Because, some of the answers for Attitude scales were incomplete, sample sizes were not identical over three brands.

Since we utilized two indicators for both Ao and Aact LISREL approach was used. The model we used for testing the inter-brand relationships is depicted in the Figure.

In the Figure the variables in circles are constructs or latent variables, and can be measured indirectly through indicators which are denoted by squares. xN, xR and xM are S Biai for brand N, brand R, and brand M, respectively. These exogenous variables have one to one correspondence with the indicators. Therefore, no measurement error is assumed for S Biai. h is the only endogenous variable, and it signifies either Ao or Aact. Two indicators (i.e. y1 and y2) are used to measure the endogenous variable h. e1 and e2 are measurement errors for y1 and y2, respectively. l1 and l2 are factor loadings which connect y1 and y2 to h, though the value of l1 can be set as 1.0 to make the unit of h to be identical with that of y1 . z is a specification error.

In the model depicted in the Figure, G N, G R, and G M signify three paths from x N, x R, and x M to h, respectively. If h is the Attitude toward (purchasing) brand N, then the value of G R and the value of G M are of interest, because these two paths show the effect of other brands' S Biais on the formation of Attitude toward (purchasing) brand N.

RESULTS AND DISCUSSION

The results of parameter estimation are shown in the Table. It is Clear that only three cases, i.e., Ao for brand M, Aact for brand R, and Aact for brand M pass the acceptable level of p that is 0.10. The p levels for the remaining three cases fall short of this level. Also, we need to be careful that some reliability coefficients are below satisfactory level of 0.6. So, we ought to be conservative in drawing conclusions from these results.

The estimated values for Gs, which are main focus in this study, show that all the signs of Gs match the hypothesized ones, except for G M in Aact for brand R. Some of these are statistically significant at the 0.05 level. In the formation of Attitude toward brand N, which is the leading brand, the S Biai Of brand R is found to have significant negative effect on the Attitude score. In the case of Aact, G R in brand N, G N in brand R, and G N in brand M are found to be significant. As the Multi-Attribute Attitude Model indicates, both in Ao and in Aact the S Biai of the concerned brand is found to have positive significant effect on Attitude.

FIGURE

MODEL OF INTER-BRAND DEPENDENCE

TABLE

STANDARDIZED PARAMETER ESTIMATES

The fact that there are some cases in which Gs are not significant means that our first Hypothesis is not fully supported. However, the findings of some cases which are consistent with our first Hypothesis is much more meaningful in this study. For, it shows that the assumption of independent evaluation does not hold att the time.

In this study, we have no direct way of testing our second Hypothesis, which indicates interdependence is more evident in Aact than in Ao. We have only three cases for each type of Attitude. The only available way of indirectly testing our second Hypothesis is to compare the results between Ao and Aact. The fact that there are more significant Gs in the case of Aact than in the case of Ao is in favor of our second Hypothesis.

In addition to the limitations which stem from the fact that only small number of cases were tested, at least three limitations can be pointed out with regard to this study.

First, in testing the interrelationships among completing brands, we relied on the Multi-Attribute Attitude Model and tried to find out the relationships between Attitude and S Biai. However, this may be criticized of being too narrowly assuming such relationships. Inter-brand relationships may exist among Attitudes or they may exist at the level of beliefs. Moreover, we must admit that the Fishbein's Model is not the sole form of forming overall evaluation of brand. Consumers may use other forms of multi-attribute combination rules as well (Hansen, 1976).

Secondly, the data we used were nonexperimental data. Therefore, only weak inferences could be made in terms of finding relationships between or among brands. To investigate fully the interrelationships in brand evaluation, consumers' memory structure is needed to be taken into consideration.

Thirdly, the analysis of interdependence among three brands in our study was rather aggregated one in that each individual respondent's ranks of preference among three brands did not necessarily correspond to the ranks of market share. If the sample size were large enough, it would be desirable to subdivide the sample according to each respondent's preference order and to carry out the analysis separately for each subgroup.

However, if the interrelationships among brands in brand evaluation were found to be the case, then one important implication could be drawn form this study. The implication is that a marketer should bear it in mind that a consumer's evaluation of his brand is always linked to the evaluations of other brands. So, if a marketer succeeded in changing consumers' beliefs about his brand, it not only would hoist the overall evaluation of his brand but also would lower the overall evaluations of competitors' brands. This means more than a relative change of preference caused by the image hike of one brand, for, the decrease of competing brands' evaluations may cause the change in preference order more likely. In other words, it means even an ordinary noncomparative ad. has comparative nature, and is often better than a comparative ad. which arouses refutative cognitive response (Gorn and Weinberg 1984). It also means that, in marketing research, when the effect of a product change is to be measured, the research ought to cover the competitors' brands as well as the changed brand.

In conclusion, although this study does not offer any definite remarks regarding the interdependence in brand evaluation, it casts doubts on the existing models and encourages further studies in this field.

REFERENCES

Ajzen, Icek., and Martin Fishbein (1980), Understanding Attitudes and Predicting Social Behavior, (Englewood Cliffs N.J.: Prentice-Hall).

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

Fishbein, Martin (1967), "Attitude and the Prediction of Behavior,: in Readings in Attitude Theory and Measurement, ed., Martin Fishbein, (New York: John Wiley & Sons), 477-492.

Fishbein, Martin, and Icek Ajzen (1975), Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research, (Reading Mass.: Addison:Wesley).

Gorn, Gerald J., and Charles B. Weinberg (1984), 'The Impact of Comparative Advertising on Perception and Attitude: Some Positive Findings," Journal of Consumer Research, 11(2), 719-727.

Hansen, Fleming (1976), "Psychological Theories of Consumer Choice," Journal of Consumer Research, vol. 3(Dec.), 117-142.

Laroche, Michel, Jerry Rosenblatt, Jacques E. Brisoux, and Robert Shimotakahara (1983), "Brand Categorization Strategies in RRB Situations: some Empirical Results," in Advances in Consumer Research, 10, eds. Richard P. Bagozzi, and Alice M. Tybout, (Ann Arbor: Association for Consumer Research), 543-554.

Payne, J. W. (1982), " Contingent Decision Behavior, "Psychological Bulletin, vol. 92, No.2, 382-402.

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