The Effects of Adding Products to a Brand on Consumers' Evaluations of New Brand Extensions

Peter A. Dacin, University of Wisconsin-Madison
Daniel C. Smith, University of Pittsburgh
ABSTRACT - In this study, we examine whether increasing the number of products associated with a brand decreases or increases that brand's strength. We develop and test hypotheses related to both positions. The main conclusion of our study is that when products affiliated with a brand have relatively little variance in terms of their perceived attribute performance, consumers' confidence in using the brand to evaluate a new extension (i.e., brand strength) seems to increase as the number of products affiliated with the brand increases. This finding holds even when the products affiliated with the brand are not perceived to be similar to each other.
[ to cite ]:
Peter A. Dacin and Daniel C. Smith (1993) ,"The Effects of Adding Products to a Brand on Consumers' Evaluations of New Brand Extensions", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 594-598.

Advances in Consumer Research Volume 20, 1993    Pages 594-598

THE EFFECTS OF ADDING PRODUCTS TO A BRAND ON CONSUMERS' EVALUATIONS OF NEW BRAND EXTENSIONS

Peter A. Dacin, University of Wisconsin-Madison

Daniel C. Smith, University of Pittsburgh

ABSTRACT -

In this study, we examine whether increasing the number of products associated with a brand decreases or increases that brand's strength. We develop and test hypotheses related to both positions. The main conclusion of our study is that when products affiliated with a brand have relatively little variance in terms of their perceived attribute performance, consumers' confidence in using the brand to evaluate a new extension (i.e., brand strength) seems to increase as the number of products affiliated with the brand increases. This finding holds even when the products affiliated with the brand are not perceived to be similar to each other.

INTRODUCTION

One of the more valuable resources a company has is the reputation of its brands. An increasing number of companies are attempting to leverage this asset by extending their brand names into new product areas (e.g. Gerber day care centers, McDonald's children's' clothing). While this strategy facilitates new product acceptance with less marketing investment than the introduction of a new brand (Smith and Park 1992) these benefits also come with an increased potential for negative side effects.

A common view of the effects of adding products to a brand holds that as the number of products introduced under a brand name increases, the meaning of the brand becomes diluted-the brand is no longer associated with any single product. It is generally believed that this disassociation results in less favorable judgments of future brand extensions (see, for example Farquhar 1990, Ries and Trout 1983, Tauber 1981). Although this is a widely held position, there may also be conditions under which favorability of judgments of extensions does not automatically decrease as the number of products affiliated with the brand increases (Park, MacInnis and Jaworski 1986).

The purpose of this study is to empirically investigate these opposite effects. To accomplish this, for each perspective, we outline its theoretical justifications and formulate propositions about the effects of increasing the number of products affiliated with a brand on: a) the favorability of consumer judgments of subsequent extensions, and b) how confident consumers feel about these judgments. Then we design and report the results of an experiment that tests these propositions.

THEORY AND HYPOTHESES

The Negative Effects of Multiple Products

As noted, it is widely believed that adding products to a brand weakens it. This position follows from the reasoning that, as products are added to a brand, consumers no longer associate the brand with a particular product or highly related class of products. Consider the following statement that is reflective of this perspective:

"In an over-communicated society you are lucky if your brand can mean one thing. Almost never can it mean two or three things." (a statement offered by Al Ries in an interview with the Economist, February, 1990, p.78)

The theoretical justification for this perspective relies on categorization theory to describe how consumers form judgments about extensions based on their knowledge of other products affiliated with the brand (Aaker and Keller 1990, Bridges 1989, Farquhar, Herr and Fazio 1990). When categorizing a brand extension, consumers try to construct a relationship between the extension and the set of existing products affiliated with the brand based on perceived similarity. Successfully constructing this relationship, in turn, alters the meaning of the brand.

For example, until recently many consumers associated the brand name of Honda with motorcycles and automobiles. Based on these highly related products, consumers may have thought of Honda as being synonymous with "motorized transportation vehicles." However, when Honda introduced a line of lawn and garden equipment, some consumers may have related these new products to other Honda products based on their primary component parts (i.e., gasoline engines, etc.). In so doing, these consumers may have altered the meaning of Honda to "products that use gasoline engines."

Following from this logic, once the meaning of the brand changes, so does its strength-the brand no longer provides a sound basis for making inferences about future extensions. Specifically, when categorizing an extension to other products affiliated with the brand, consumers are assumed to transfer their existing beliefs and feelings about the brand to the extension. The extent to which this occurs depends on the similarity or category fit between the extension and other products affiliated with the brand (Aaker and Keller 1990, Farquhar, Herr and Fazio 1990, Bridges 1989, MacInnis and Nakamoto 1990, Park, Lawson and Milberg 1991, The Minnesota Consumer Behavior Seminar 1987). As the number of dissimilar products affiliated with a brand increases, the meaning of the brand changes and becomes less distinct in the minds of consumers. As a result, it becomes difficult for consumers to use the brand in categorizing subsequent extensions, thus, reducing the likelihood of transferring brand affect to the extension.

However, when a brand is affiliated with a portfolio of highly related products, the meaning of the brand will change less than when it is affiliated with highly dissimilar products. That is, the meaning of the brand remains relatively distinct in the minds of consumers. Consequently, conditions of high similarity among products affiliated with the brand should facilitate the use of the brand as a basis for categorizing subsequent extensions, reducing the negative effect of having multiple products affiliated with the brand.

This discussion leads to the following propositions:

H1: As the number of products affiliated with the brand increases, the favorability of the judgments about an extension of the brand decreases.

H2: The negative effects of having multiple products affiliated with a brand on the judgments of an extension will decrease as the similarity of products affiliated with the brand increases.

The Favorable Effects Multiple Products

The theory outlined above assumes that in categorizing a new extension, consumers attempt to interpret the extension in terms of the entire set of other products affiliated with the brand. Indeed, what may be more critical is that consumers are able to identify at least one other product that is somehow related to the extension. For example, the Yamaha brand is associated with electronics, motorcycles, and musical instruments. Following the reasoning described earlier, one would conclude that the Yamaha name is highly diluted and should not have a favorable effect on consumer inferences of a new extension, say, watches. However, consumers may perceive the skills needed to manufacture watches as being similar to those needed to produce electronic devices and thus, successfully relate watches to the brand even though watches, on the surface, appear to be quite dissimilar to other products affiliated with the brand.

In addition, while the logic of the potential hazards of using a single brand to represent multiple products is intuitively appealing, there exists an opposing perspective. Many firms intentionally follow a strategy of brand extensions focusing on a particular abstract dimension rather than specific product characteristics (e.g., AMF's "We Make Weekends", Johnson and Johnson's "We Know Babies"). In essence, consumers can equate the abstract dimension with the brand in much the same way as they equate a specific product with a brand at its inception. So, while the meaning of brand associations may become abstracted as the number of products affiliated with the brand increases, it does not necessarily follow that this will negatively affect the favorability of judgments about subsequent extensions.

Given this, from a consumer decision making perspective, it is easy to argue that adding products to a brand can strengthen the brand. In this perspective, consumers rely on a brand name as a decision heuristic (Johnson and Russo 1984), and as a vehicle for managing perceived risk (Cox 1967). Therefore, whether a brand is harmed or helped by being affiliated with multiple products is found in the extent to which consumers rely on it as a basis for making inferences about a new extension. Reliance on a brand in evaluating a new extension, in turn, is expected to be a function of the extent to which consumers believe that their knowledge of other products affiliated with the brand is an accurate predictor of the experience they will have with the new product.

Given this conceptualization, we can see how adding products to a brand can strengthen it. The products affiliated with a brand essentially represent a "data base" from which consumers draw information in forming judgments about the experiences they can expect from an extension. As the number of products increases, the breadth of data increases-consumers have knowledge of the brand in multiple product contexts. Consequently, consumers' confidence in a brand is expected to depend not only on the breadth of data, but also on the degree of variability or consistency in the data.

For instance, consumers might find the quality of Izod jackets and slacks substantially above that of its suits and socks, but well below that of its polo shirts. In contrast, the quality of the product offerings associated with another brand, say Ralph Lauren, may be more consistent. The overall level of quality associated with Ralph Lauren products, then, may not be as high as the quality associated with some Izod products. However, because of the consistency in outcomes associated with the brand, we expect consumers to feel more confident in their judgments of a subsequent Ralph Lauren extension than an Izod extension.

Therefore:

H3a: Under conditions of high consistency in past outcomes, consumer confidence in judgments about brand extensions will increase as the number of products affiliated with the brand increases.

H3b: Under conditions of low consistency in past outcomes, consumer confidence in judgments about brand extensions will decrease as the number of products affiliated with the brand increases.

The consistency of past outcomes associated with the brand should also account for some of the variation in consumer evaluations of brand extensions. Low consistency in past outcomes will tend to create conditions of uncertainty in consumers' expectations about the brand that should lead to reluctance on the part of consumers to use any specific past outcomes to generalize to brand extensions. Therefore, conditions of low consistency in past outcomes are expected to give rise to less favorable evaluations of subsequent brand extensions than conditions of high consistency in past outcomes.

H4: In general, judgments of brand extensions will be more favorable under conditions of high consistency in past outcomes associated with the brand than conditions of low consistency of past outcomes.

Finally, we expect the effect of the number of products affiliated with a brand on favorability of extension evaluations to be moderated by the degree of consistency in past outcomes. Following from the same logic used to develop H3a and H3b we hypothesize that:

H5a: Under conditions of high consistency in past outcomes, favorability of judgments about brand extensions will increase as the number of products affiliated with the brand increases.

H5b: Under conditions of low consistency in past outcomes, favorability of judgments about brand extensions will decrease as the number of products affiliated with the brand increases.

METHOD

The above hypotheses were tested using a 2 X 2 X 2 between subjects factorial design with two levels of similarity of extension to the products associated with the brand (similar and dissimilar), two levels of consistency of outcomes of other products associated with the brand (consistent and inconsistent), and two levels of the number of products affiliated with the brand (three and seven).

Stimuli

We incorporated all three manipulations in a brand ratings table typical of those found in magazines such as Consumer Reports or Consumer Guide. We manipulated the consistency of outcomes of other products associated with the brand by altering the pattern of ratings found in the table. Furthermore, to ensure that subjects' perceptions of consistency did not affect their overall feelings towards the brand we also included a summary evaluation of the brand in the table.

We manipulated similarity and the number of products affiliated with the brand by altering the products listed in the ratings table. Products listed in the "similar" conditions were chosen to be either similar or dissimilar to each other and to the brand extension (electric clothes irons). The three-product similar manipulation consisted of: small kitchen appliances, electric razors, and hair dryers/curling irons. The three-product, dissimilar manipulation consisted of: kitchen appliances, 10 speed bicycles and, sports watches. For the seven-product manipulation, we added four products to each table. In the similar case, we included small electric power tools, electric garage door openers, hand held vacuum cleaners and electric pencil sharpeners. In the dissimilar case, lawn and garden tools, electric garage door openers, camping equipment (tents, backpacks), and wooden patio furniture were added. Manipulation checks revealed that these manipulations were successful.

We selected the hypothetical brand name "Jasil" for two reasons. First, pretests revealed that subjects were unaware of this name, therefore, previous experience could not interfere with the manipulations. Second, pretests also demonstrated that associating Jasil with a variety of unbranded products did not alter the evaluation or cognitive associations of these products. This was important since the experiment required that we manipulate both the beliefs subjects had for the brand (i.e., consistency) and the number of products they affiliated with the brand. This would have been difficult to achieve with well established brands.

Subjects and Procedures

One hundred twenty undergraduate students were recruited from undergraduate business classes. For their participation, subjects were offered extra credit. The subjects were randomly assigned to the eight cells so that each cell contained 15 subjects.

An experimenter handed subjects a booklet as they entered the laboratory. The first page of the booklet contained a paragraph orienting subjects to the task. It told them that they would see a table consisting of a number of products being offered under the Jasil brand name and that following this table, they would be responding to several questions about the products. Subjects were given two minutes to read these instructions and then were told to turn the page.

The second page of the booklet presented the brand ratings table containing the manipulations. Instructions informed subjects that this table was from Consumer Reporter, a publication that, as a service to its readers, reviews consumer products in terms of their performance on a number of important attributes. Each product was rated on six attributes obtained from pretests. These were: durability, customer satisfaction, value for the money, goodness of warranty, style/design and, reliability/engineering. Subjects studied the table for four minutes to study the table and then turned the page and continued through the booklet, responding to all the questions on the remaining pages.

After questions about the products in the table, we gave subjects a scenario that stated that the same manufacturer decided to introduce a line of electric clothes irons under the Jasil brand name. Following this, subjects responded to questions about this brand extension.

In order to determine whether subjects identified the purpose of the study, randomly selected subjects were asked extensive debriefing questions after the experiment. The subjects' responses to these questions indicated that none had guessed the purpose.

Measures

Subjects' Perceived Consistency of Outcomes. This measure was composed of two seven-point items. The first item asked subjects to indicate the extent to which they felt that the attribute ratings varied between products in the table (varies a lot - does not vary at all). The second item asked how consistent subjects felt the products were in terms of their attribute ratings (very consistent - very inconsistent). The correlation between these two items was .87. An average of these two items was used as a perceived consistency score.

Subjects' Perceived Similarity of Products Affiliated With the Brand and Similarity of the Extension to Other Products Affiliated with the Brand. Following from previous research, (see for example, Aaker and Keller 1990, MacInnis and Nakamoto 1990), subjects were asked to indicate how similar they felt the products affiliated with the brand were in terms of: (1) the types of needs the products satisfy, (2) situations in which the products are used, and (3) physical features. Each item used a 7-point scale anchored by very similar - very dissimilar. Coefficient alpha for the three-item scale was 0.74. The three items were averaged to represent consumer perceptions of similarity.

The above dimensions were also used to measure the perceived similarity of the brand extension (electric clothes irons) to the other products affiliated with the brand. Coefficient alpha for this scale was 0.77.

Overall Evaluation of the Jasil Brand Name and Overall Evaluation of the Jasil Clothes Iron. We measured subjects' evaluations of the Jasil brand name and the Jasil electric clothes iron using 7-point scales. One scale asked subjects how they felt about the Jasil brand name and the other scale asked subjects how they felt about the Jasil electric clothes iron. Each scale was anchored by very favorable - very unfavorable.

Expectations for Various Attributes for Any Future Products Affiliated with Jasil and the Confidence in These Expectations. For this measure, subjects were presented with a list of the six attributes that appeared in the ratings table. Beside each attribute was a 7-point scale anchored by far above average - far below average on which the subjects were to indicate their expectation for that attribute for any future product associated with the Jasil brand name. Coefficient alpha for this scale was 0.97. An average of these items was used to indicate subjects' expectations.

In addition, a second 7-point scale was associated with each attribute. This scale, anchored by very confident - not confident at all was used to obtain the subjects' rating of their confidence in their expectation for that attribute. Coefficient alpha for this scale was 0.76. Again, an average of these items was used.

Expectations for Various Attributes for the Jasil Clothes Iron and Confidence in These Expectations. Identical measures to those just described were used to obtain subjects' expectations and confidence in their expectations for the Jasil electric clothes iron brand extension. Coefficient alpha for the extension expectation scale was 0.96 and for the confidence scale 0.74.

Knowledge of clothes irons. A measure of knowledge was taken because it was felt that knowledge might affect subjects' judgments of the product. However, in all analyses, knowledge had no effect as a covariate.

RESULTS

As a preliminary test of the hypotheses concerning the role of similarity on the effects of having multiple products affiliated with a brand (H2), and the effects of consistency and number of products affiliated with the brand on the favorability of judgments (H4, H5a, H5b), we ran a 3-way analysis of variance. In this analysis, we used consumers' evaluations of the brand extension as the dependent variable and the manipulated similarity, consistency of past outcomes and number of products affiliated with the brand as independent variables.

The results of this analysis appear in Table 1. In this table, a higher number represents more favorable evaluations. None of the interactions were significant (three way interaction [F(1,112)=0.03, p=n.s.]; similarity x number of products [F(1,112)=0.05, p=n.s.]; similarity x consistency [F(1,112)=0.24, p=n.s.]; consistency x number of products [F(1,112)=0.08, p=n.s.]). There were significant main effects for both the similarity and consistency manipulations (similarity [F(1,112)=4.21, p<0.05]; consistency [F(1,112)=8.39, p<0.05]). These results suggested initial support for the role of the similarity of a brand extension to current products affiliated with the brand (H2) and of the consistency of outcomes associated with the brand (H4). Furthermore, the mean evaluation in these conditions indicated that; a) under conditions of low consistency, evaluation decreased as the number of products affiliated with the brand increased and, b) under conditions of high consistency, evaluation increased as the number of products affiliated with the brand increased. Although these contrasts were not statistically significant, the directions were consistent with the predictions of H5a and H5b. Finally, the evaluation of the extension was not affected by the number of products affiliated with the brand (H1) [F(1,112)=0.85, p=n.s.].

TABLE 1

MEANS AND (STANDARD DEVIATIONS) FOR EVALUATIONS OF THE JASIL ELECTRIC IRON BRAND EXTENSION

TABLE 2

MEANS AND (STANDARD DEVIATIONS) FOR CONFIDENCE FOR JASIL BRAND ASSOCIATIONS

In order to further test the hypothesis concerned with the role of similarity on the effects of having multiple products affiliated with a brand (H2) and to test the effect of consistency of past outcomes on consumer confidence in their judgments (H3), the 3-way analysis of variance was run on the dependent variable of confidence in outcomes associated with the brand. The results appear in Table 2. In this table, a higher number represents greater confidence. Of all the interactions, only the consistency x number of products interaction was significant [F(1,112)=5.13, p_0.05]. Two main effects, similarity [F(1,112)=3.01, p_0.05] and consistency [F(1,112)=7.11, p_0.05], were also significant. The main effect for consistency was interpretable since the significant interaction in which it appears was ordinal. Furthermore, although none of the contrasts were statistically significant, under conditions of low consistency, confidence decreased as the number of products affiliated with the brand increased while, under conditions of high consistency, confidence increased as the number of products affiliated with the brand increased. The direction of these findings were consistent with the predictions of H3a and H3b.

Regression analysis was used as an additional test of the effects of consistency and similarity on subjects' judgments of the brand extension. In this analysis, we used the continuous measures of similarity and consistency to test for the relative contributions of each variable on the judgments of favorability for the electric iron brand extension. Tests for multicollinearity revealed that it was not a problem.

The incremental F test in the hierarchical regression for the consistency x similarity interaction was not significant [DR2=0.01 [F=1.53, p=n.s.]. In the main effects model, the standardized betas for the variables were 0.174 for similarity [t=1.66, p_0.10] and 0.355 for consistency [t=5.104, p_0.05]. The model explained approximately 16% of the variance in the dependent variable. These results suggested that the perceived consistency of past outcomes had a much more important role on subjects' judgments about the brand extension than did subjects' perceived similarity of the brand extension to other products affiliated with the brand. This supported the role of consistency and partially supported the role of similarity in judgments of favorability.

In summary, the results of these analyses suggested that both the favorability of consumers' inferences about an extension and the confidence in these inferences increased with the consistency of past outcomes. Similarity, on the other hand, had no effect on confidence and, compared to consistency, had a relatively small effect on the favorability of judgments about the extension.

DISCUSSION

This study investigates the effects of the number of products affiliated with a brand. While there is a commonly held belief that detrimental effects increase as the number of products affiliated with the brand increases (Farquhar 1990, The Economist 1990, Tauber 1981), no strong evidence for this emerges from this study. In general, the number of products affiliated with a brand does not affect subjects' judgments of a brand extension.

Another belief found in extant literature is that extensions should be as similar as possible to existing products affiliated with the brand (Aaker and Keller 1990, Farquhar, Herr and Fazio 1990, Bridges 1989, MacInnis and Nakamoto 1990, Park, Lawson and Milberg 1991, The Minnesota Consumer Behavior Seminar 1987). While our study finds some support for this relationship, similarity provides only a partial explanation for consumers' judgments of brand extensions.

A major contribution of this study is that we extend the thinking about the effects of adding products to a brand by focusing attention on the degree of consumers' reliance on a brand in decision making. This differs considerably from current thought which focuses primarily on the extent to which a brand's meaning changes. Adding a product to a brand need not result in unfavorable judgments of that extension. Indeed, our findings support this. Furthermore, we find that the number of products affiliated with the brand moderates the relationship between consistency and confidence. Under high consistency, there tends to be a positive relationship between the number of products affiliated with the brand and confidence. Under low consistency the opposite is true.

We recognize that there are limitations with our study. One of the study's major limitations emerges from our use of a single exposure to a Consumer Reporter table to manipulate the consistency of the outcomes of products associated with the brand, which we hypothesize to affect confidence. In a natural setting, a consumers' confidence in using information about a brand to make judgments about an extension will, no doubt, develop over multiple exposures and experiences. However, the purpose of this study is to examine a number of theory based predictions. Therefore, we require tight controls over a variety of cognitive phenomena. To accomplish this we consciously sacrifice some external validity. While we have no doubt that the confidence we create in the minds of the subjects may be relatively short lived and unstable compared to confidence built over multiple exposures and experiences with the brand, the results we obtain in the context of our study are very encouraging.

Our results demonstrate the importance of accounting for consistency in future research concerned with the effects of brand extensions. Furthermore, a number of managerial implications emerge from the role of consistency and confidence on brand dilution. The first group of implications are concerned with the strategies necessary to manage favorable judgments of brand extensions. Since the key determinant of brand dilution is the consistency in previous outcomes with products affiliated with the brand, the task of marketers is to monitor the outcomes consumers attach to the products affiliated with the brand. Consumers will only use the brand to make judgments about future extensions when they have built some confidence in the brand as a result of associating it with consistent outcomes.

The second group of implications deals with the way in which managers should evaluate the importance of a brand in brand extensions. Contrary to current thinking, a brand does not necessarily suffer as more products are added to it. While the meanings associated with a brand change as products are added, consumers are still able to rely on the new meanings to make inferences about subsequent extensions. Indeed, the results of this study suggest that it is more important to consider the consistency of outcomes that consumers associate with the brand than simply the sheer number of products affiliated with the brand or the similarity of an extension to the current products associated with the brand.

REFERENCES

Aaker, David A. and Kevin L. Keller (1990), "Consumer Responses to Brand Extensions," Journal of Marketing, 54 (January), 27-41.

Bettman, James A. and C. Whan Park (1980), "Effects of Prior Knowledge and Experience and Phase of Choice Process on Consumer Decision Making Processes," Journal of Consumer Research, 7 (December), 234-248.

Bridges, Sheri (1990), "A Schema Unification Model of Brand Extensions," Working Paper, Stanford University.

Cox, Donald F. (1967), Risk Taking and Information Handling in Consumer Behavior. Cambridge, MA: Harvard Business School.

Farquhar, Peter H. (1990), "Managing Brand Equity," Marketing Research, (September), 24-33.

Farquhar, Peter H., Paul Herr, Russell H. Fazio (1989), "Extending Brand Equity to New Product Categories," Working Paper, Center for Product Research, Carnegie Mellon University.

Johnson, Eric and J. Edward Russo (1981), "Product Familiarity and Learning New Information," Journal of Consumer Research, 13 (June), 38-47.

MacInnis, Deborah J., and Kent Nakamoto (1990), "Cognitive Associations and Product Category Comparisons: The Role of Knowledge and Context," Working Paper, School of Business, University of Arizona.

Minnesota Consumer Behavior Seminar (1987), "Affect Generalization to Similar and Dissimilar Brand Extensions," Psychology and Marketing, 4 (3), 225-237.

Park, C. Whan, Robert Lawson, Sandra Milberg (1991), "Bases of Categorization Judgments for Brand Extensions," Journal of Consumer Research, 18 (September) 185-193.

Park, C. Whan, Deborah J. MacInnis, Bernard Jaworski (1986), "Strategic Brand Concept Management," Journal of Marketing, 50 (October), 135-145.

Smith, Daniel C. and C. Whan Park (1992), "The Effect of Brand Extensions on Market Share and Advertising Efficiency," Journal of Marketing Research, 29 (August), 296-313.

Sujan, Mita (1985), "Consumer Knowledge: Effects on Evaluation Strategies Mediating Consumer Judgments," Journal of Consumer Research, 12 (June), 31-46.

Tauber, Edward (1981), "Brand Franchise Extensions: New Products Benefit From Existing Brand Names," Business Horizons, (March/April), 36-41.

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