The Usefulness of Product Warranties For Reputable and New Brands

Citation:

Daniel E. Innis and H. Rao Unnava (1991) ,"The Usefulness of Product Warranties For Reputable and New Brands", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 317-322.

Advances in Consumer Research Volume 18, 1991      Pages 317-322

THE USEFULNESS OF PRODUCT WARRANTIES FOR REPUTABLE AND NEW BRANDS

Daniel E. Innis, Ohio State University

H. Rao Unnava, Ohio State University

[The authors thank Paul Miniard and Peter Dickson for their valuable comments on an earlier version of this manuscript.]

This paper investigates the impact that product warranties have on consumer product evaluations for new and established high-performance brands. Results indicate that strong warranties affect product evaluations positively for new brands but have little effect for established reputable brands.

The role of product warranties in the marketing of products has undergone a significant change in the past several years. Used initially as an indicator of the limit of a manufacturer's liability in the event of product failure, warranty information is presently being used as a competitive weapon to differentiate one's product from competition. Contemporary examples include the 7/70 protection plan offered by Chrysler Corporation, and the now well-known claim by Victor Kiam that Remington shaving system "shaves as close as a blade or your money back."

Despite the increased importance attained bywarranty information in the marketing of products, literature is sparse on the effects of warranty information on consumers' evaluations of products. Several studies have, in the past, examined the importance of warranty information to consumers (e.g., Olson and Jacoby 1972; Ross 1975; Lehmann and Ostlund 1973), and how product warranties affect consumers' perceptions of risk (Ross 1975; Bearden and Shimp 1982; Shimp and Bearden 1982). The conditions under which warranty information may or may not be effective, however, have received little attention. Specifically, if a relatively unknown manufacturer launches a new brand in the market, and provides attractive warranties that are superior to those provided by existing brands, should the existing brands match these warranties? In other words, does brand image interact with warranty information? The objective of the research reported here is to examine the interaction between brand image and warranty information. It is proposed that warranty information will be more effective in influencing evaluations of new brands than it will be in influencing evaluations of brands that hold positive images in consumers' minds.

BRAND NAME AS A MODERATOR OF WARRANTY INFORMATION

Consumers appear to be very sensitive to brand names and use them as risk relievers in their product choices (Shimp and Bearden 1982; Levitt 1967, Montgomery 1975). Research has shown that warranty information is not considered as important as brand name information (Olson and Jacoby 1972; Roselius 1971). However, consumers do seem to rely on warranty information to reduce risk and form product evaluations (Bearden and Shimp 1982). The positive effects of both brand name and warranty information on consumers' product evaluations should not be interpreted to mean that these effects are additive, however (cf. Olson and Jacoby 1972). These cues may interact with each other such that warranty information is effective only under certain conditions.

The influence of brand name on consumers' attitudes has received unequivocal support in the literature. In a classic study, Simon (1970) had subjects evaluate three brands in the home furnishings product category, based on advertisements for the three brands with their names masked. It was found that the evaluations given the three brands by the subjects were completely reversed when the brand names were revealed. The market leader, which was rated third in the masked condition, was rated highest when the brand names were available. Simon concluded that brand leaders, based on their reputation, will elicit more favorable response to their advertising than less-known manufacturers.

In a somewhat similar study, Jacoby, Olson and Haddock (1971) found that the evaluations given various brands of beer by their subjects were influenced by the presence of brand name information. Specifically, the difference between subjects' ratings of an ultra-premium brand and an inexpensive brand with brand names withheld widened when the brand names were revealed. The idea that an established brand name reduces perceived risk and enhances product evaluations has been supported by a number of other researchers as well (e.g., Shimp and Bearden 1982; Levitt 1967; Montgomery 1975; see also Rao and Monroe 1989).

Some of the earlier research on product warranties sought to assess the importance consumers attach to warranty information in their evaluation of products. For example, in one study, Olson and Jacoby (1972) asked housewives to indicate the factors they consider most important when they were contemplating the purchase of five different products (hairdryers, living room rug, ground coffee, shampoo and aspirin tablets). A list of over 12 factors was made available for each product. It was found that except in the case of hair dryers, housewives did not consider warranty information to be an important factor in the purchase of the experimental products. Brand name, on the other hand, was consistently indicated to be an important consideration in buying four of the five products.

Olson and Jacoby (1972), in discussing these results, proposed the distinction between intrinsic and extrinsic product quality cues. Intrinsic cues are those product attributes that, when changed, will result in a change in the composition of the product itself. Extrinsic cues, on the other hand, can be changed without affecting the composition of the product (e.g., warranties, brand name). In the Olson and Jacoby study, subjects were found to use intrinsic cues more often than extrinsic cues to infer product quality. Based on this result, Olson and Jacoby suggested that when intrinsic cues are available to consumers, the effect of extrinsic cues on product quality perceptions will be reduced. More important, Olson and Jacoby argued that the effect of multiple cues will not be additive, but interactive (e.g., Jacoby, Olson and Haddock 1971). The limited use of warranty information by consumers as an aid to reduce their perceived risk has also been reported by Roselius (1971).

More recently, however, Bearden and Shimp (1982) varied the quality of warranty information, the reputation of the manufacturer offering the warranty, and the price of the product, in a 2 X 2 X 2 experimental design. Two products - plastic tires and an exercise jogging system - were used. Unlike prior research, Bearden and Shimp provided subjects with other information on the products, apart from the warranty, brand name and price information. Thus, the effect of these extrinsic cues was examined when they were embedded in other intrinsic cues. Perceived quality of and attitudes toward the products were used as the dependent measures. It was found that both warranty information and manufacturer reputation had significant effects on reducing consumers' perceived risk, and enhancing their attitude toward the products. The interaction between these two variables, however, was not examined.

The positive impact of warranty information on perceived product quality has also been reported by Perry and Perry (1976) and Shimp and Bearden (1982). It appears, therefore, that even in the presence of other intrinsic cues, consumers rely on warranty information to make inferences about product quality and form attitudes.

In terms of the possible interaction between brand name information and warranty information, Shimp and Bearden (1982) hypothesized an interaction between reputation of the warrantor and the quality of warranty. They argued that for reputable manufacturers, perceived risk should monotonically decrease with an increase in the quality of warranty. For an unknown manufacturer, however, Shimp and Bearden (1982) proposed a boomerang effect with better warranties actually resulting in negative attitudes. This argument was based on their assumption that when warranties that are very attractive are offered by unknown manufacturers, consumers will respond negatively because they think the warranty is 'too-good-to-be true." In the Shimp and Bearden (1982) study, however, this interaction hypothesis was not supported.

Past research, however, indicates that both warranty information and brand name are used as risk reducers by consumers (Roselius 1971; Bearden and Shimp 1982; Shimp and Bearden 1982). Between brand name and warranty information, the former appears to be a more reliable risk reducer (Roselius 1971; Olson and Jacoby 1972). This suggests that an established brand na ne, by virtue of its record of good performance, is warrantying its performance to a prospective consumer. That is, consumers appear to be inferring a high probability of performance of an established brand based on its record of good performance in the past. In such conditions, the role of warranty information should be minimal, because a brand's past performance is acting as an implied warranty or as a signal of reliability. Thus, when a brand is well established in its product class, the role of warranty information in affecting consumers' evaluations of the brand should be minimal.

On the other hand, when a new brand is launched by a relatively unknown manufacturer, brand name can no longer act as a risk reliever. Consumers, in this case, should rely on warranty information to reduce their perceived risk about the possibility of product failure, assuming that warranty information is available. Therefore, for an unknown brand, warranty should affect consumer evaluations positively. In fact, consumer evaluations should increase monotonically with the quality of the warranty.

In sum, when consumers have available positive information about a brand's performance history, their reliance on warranty information in forming an evaluation of a new product under the same brand name would be limited. On the other hand, when no information is available to consumers about a brand's performance history, consumers will tend to rely on warranty information to form evaluations of that brand.

EXPERIMENT

The moderating effect of brand name on the effect of warranty information was tested in a 2 (brand name: established vs. unknown) X 3 (warranty quality: warranty information absent, three-month warranty, 10-year warranty) between subjects design. A total of 120 subjects, recruited from undergraduate marketing classes, participated in the study for extra credit. Subjects were randomly assigned-to treatment conditions.

Target Advertisement

A new brand of bicycle was used as the target product in this study. The choice of the target product was based on two considerations: 1) that the subjects in the sample will find the product relevant, and 2) that some degree of performance risk is associated with the product so that subjects will rely on warranty information in their product evaluations.

The information about the bicycle was presented in the form of an advertisement. The copy for the ad was derived from several existing bicycle ads in a cycling-enthusiast magazine. Seven strong and positive product claims (e.g, gel saddle to ease road shock) were presented along with a picture of the bicycle. The inclusion of these arguments rendered the test of the efficacy of warranty claims more conservative because it has been argued that the effect of extrinsic cues will be minimized in the presence of intrinsic cues (Olson and Jacoby 1972). The detection of effects of warranty information in these conditions would increase our confidence in the role of warranty information in consumer product evaluations.

Warranty Quality

The information about warranty was included as the eighth argument in the bicycle ad. In the condition where no warranty information was available, the eighth argument was not included. Thus, this condition acted as a control condition. An alternative would have been to include a phrase in the ad that there was no warranty for the bicycle. However, the presence of such a phrase was deemed to make subjects sensitive to warranty information, when in fact, they may be not. Thus, the effect of warranty information was gauged in a situation where subjects either processed it in conjunction with a substantial amount of other product-related information, or were not even primed to the existence of warranty information.

Brand Name

Two versions of the bicycle ad were developed to achieve the brand name manipulation. The first version promoted the bicycle as a new model in the Schwinn line of bicycles. Pretests revealed that the brand name Schwinn was associated with high performance and reliability. Thus, the use of the name 'Schwinn' was expected to arouse favorable impressions in subjects' minds. The second version of the target ad promoted the bicycle as a new model with a fictitious name Monarch. Pretests revealed that students were unfamiliar with Monarch bicycle. Thus, students could not rely on this brand name for performance assurance.

Procedure

The study was conducted with small groups of four to eight subjects each. Upon entering the experiment room, subjects were told that their task was to examine an ad for a new product and answer some questions about their impressions of the product. They were then given a booklet that contained some procedural instructions, the target ad, and the questionnaire.

The first question measured subjects' attitude toward the target product using five seven-point bipolar scales (anchored by "very good- very bad", "very desirable-very undesirable", "very negative-very positive", "very superior-very inferior", and "very awful-very nice"). Following that, subjects indicated their likelihood of buying the advertised brand if they were in the market for a bicycle. Two seven-point scale were used to measure buying intention (anchored by "very likely-very unlikely" and "very probable-very improbable). Finally, subjects responded to a series of eight questions that measured their product attribute beliefs. These questions were included to see if the warranty information had any effects on subjects' beliefs about the product attributes. The eight questions were derived from the eight arguments presented in the bicycle ad. Subjects indicated their agreement with each of the eight statements on a seven-point scale anchored by "strongly disagree" and "strongly agree". After completing the questionnaire, subjects were dismissed and debriefed the following day.

RESULTS

The data were coded so that higher scores meant more positive attitudes, higher buying intentions, and more positive beliefs about the product.

Manipulation checks

The awareness of the two experimental brands was first examined by comparing the number of subjects who indicated that they were aware of Schwinn and Monarch brands. All the respondents in the established brand category indicated that they had heard the Schwinn brand name before. Only three subjects (<6%) in the new brand condition indicated that they were aware of the Monarch brand name. In addition, more than 75% of the subjects rated Schwinn brand bicycles 'good'. Based on these data, the brand name manipulation was deemed successful.

Next, the rating provided by all the subjects on the quality of warranty offered by the target brand was submitted to an ANOVA with brand name reputation and warranty quality as the independent variables. The analysis revealed only a warranty main effect (p<0.001). As the quality of warranty increased, subjects indicated more agreement with the statement that the warranty offered by the target brand was good. This finding supported the warranty quality manipulation. The mean warranty rating scores are presented in Table 1. It is interesting to note from the mean warranty quality ratings that subjects perceived the difference between the 10-year warranty and no warranty to be greater for the unknown brand ( A=2.31) than for the established brand ( A=136). This may be viewed as preliminary support for the idea that warranty information has more impact for unknown brands than for- established brands.

Attitude

The five scales composing the attitude measure were found to be highly reliable (Cronbach's A = 0.95). Therefore, the mean of these scales was computed and used in the analysis. The mean attitude scores are presented in table 1.

It was hypothesized that the quality of warranty information will interact with the brand name reputation. To test for this hypothesized interaction, an ANOVA was performed on the attitude score with warranty quality and brand name reputation as the independent variables. The ANOVA revealed no main effects (p > 0.5) for either the brand name or the warranty quality. There was, however, a significant interaction (p < 0.03) between brand name and warranty quality and the direction of mean attitude scores were in the expected direction.

TABLE 1

MEAN ATTITUDE, BUYING INTENTION AND PRODUCT ATTRIBUTE BELIEF SCORES BY EXPERIMENTAL CONDITION

Given the nature of the hypothesis, a series of a-priori multiple t-tests were conducted (Kirk 1982, p.95). First, the mean attitude score for the established brand in the no warranty condition was contrasted with the mean attitude scores for the established brand when warranty information was available. The analyses revealed no significant differences between these conditions (p > 0.2 for both t-tests) which was supportive of the hypothesis. Similar t-tests were then conducted for the new brand. The analyses revealed that the mean attitude score for the new brand was significantly higher (p < 0.01) when it was accompanied by a ten-year warranty (M = 5.89) than when it was not (M=5.35). The difference between the attitude scores when the warranty was only three months (M=5.68) and the other experimental conditions, however, was not significant.

Because it was hypothesized that the attitude scores for the new brand will monotonically increase with the quality of warranty offered by it, a linear trend analysis was performed on those attitude scores. The analysis revealed that the attitude scores exhibited a linear trend as expected (F=4.8, p < 0.03). Thus, the hypothesis that warranty quality affects only new brands, but will not affect established brands was supported.

Buying Intention

The two scales used to measure buying intention correlated highly (r=0.9). Therefore, their mean was computed and used in an ANOVA with warranty quality and brand name reputation as the independent variables. Although the direction of means mirrored the attitude data, the interaction between warranty quality and brand r.ame reputation was not significant (p<0.1).

A series of a-priori contrasts was then performed on the mean buying intention scores. Once again, no reliable differences between buying intentions were found between the various warranty quality conditions for the established brand (all p-values > 0.2). Surprisingly, no differences were exhibited by subjects in their likelihood of buying the target brand in the unknown brand condition, as a function of warranty quality. Thus, the attitude results did not extend to buying intention, although the direction of the means was in the expected direction.

Product attribute beliefs

The responses to the eight product attribute belief statements were submitted to a reliability analysis. The average inter-correlation between the responses was moderate (A=0.64). Therefore, the average of the responses to the belief questions was computed and used as an index of subjects' product attribute beliefs.

The belief index was then analyzed using ANOVA with warranty quality and brand name reputation as the independent variables. No main effects emerged (p > 0.5 for both main effects). The interaction between warranty quality and brand name reputation was, however, marginally significant (p < 0.1).

A priori pairwise contrasts were performed on the mean scores within each brand name condition. The contrasts revealed that the belief scores of the subjects were unaffected by the warranty quality manipulation when the brand name was well known (all p-values were > 0.2). For the unknown brand condition, however, subjects exhibited significantly more positive product attribute beliefs (p<0.01) when they were exposed to the 10-year warranty brand (M = 5.73) than when they were exposed to the no warranty brand (M = 5.23). The difference between the three-month warranty condition and the other two conditions was not significant (p>0.1). Thus, the belief index scores exhibited the same pattern as the attitude scores.

DISCUSSION

Previous research on warranty information has not examined how it interacts with the reputation of the warrantor. Our research examined the moderating effect of the reputation of the warrantor on the effect of warranty information on subjects' brand evaluations. It was hypothesized that warranty information exerts greater impact when it is offered by an unknown brand than when it is offered by a relatively established brand. This differential impact of warranty information was argued to be due to the risk-alleviating ability of an established brand name. The reduction of perceived risk by consumers that results from brand name reputation makes the role of warranty information for established brand names of minimal importance. For unknown brands, consumers do not have any past performance statistics to rely on for perceived risk reduction. Therefore, warranty information achieves greater importance when presented for an unknown brand because it performs the task of risk reduction. The perception of reduced risk, in turn, results in more positive attitudes toward the product (Bearden and Shimp 1982). This hypothesis was generally supported in our research. Subjects exhibited more favorable attitudes toward an unknown brand when a strong warranty was offered than when no warranty wa i offered. The attitude of subjects, however, was not influenced by the warranty offered by an established brand.

It is interesting that the direction of the interaction hypothesized in our research is opposite of what was hypothesized by Shimp and Bearden (1982). In the Shimp and Bearden study, warranty information was hypothesized to be more effective for established brands than for unknown brands. Their reasoning behind this expectation was that when outstanding warranties are offered with unknown brands, consumers may feel that these warranties are 'too-good-to-be-true'. The resulting counter-argumentation was expected to reduce consumers' evaluations of unknown products that are coupled with good warranties. This pattern of interaction was not supported in the Shimp and Bearden (1982) study.

We believe that the 'too-good-to-be-true' perception of consumers is plausible but only under certain limited conditions. Specifically, for consumers to conclude that the warranty offered by an unknown manufacturer is too good to be true, the warranty should diverge from the industry standard significantly so that consumers would notice the disparity. This appears to be the case in the Shimp and Bearden study, especially for the plastic tire product. The positive warranty condition offered a full money-back guarantee if the tire did not perform as promised for 5 years or 50,000 miles. Warranties for tires are normally pro-rated. Consequently, a warranty that is not pro-rated appears suspicious. Second, for a consumer to discount warranty information based on a company's reputation, the consumer must hold a negative impression about the company. In the Shimp and Bearden study, the low reputation company was portrayed as a company that did not have any experience in manufacturing tires, but only in the manufacture of plastic products (p. 231-232 of Bearden and Shimp (1982)). An unknown manufacturer who advertises its product would not make consumers aware of its inexperience. Therefore, in reality, unless consumers hold negative attitudes toward a company or the warranty offered is incongruous with competitive warranties, warranty information more likely will evoke positive response from consumers.

In our research, differences in warranty quality were found to cause statistically significant, but only small differences in subjects' evaluations of the target product. Specifically, for an unknown brand, the difference-in evaluations between the no warranty condition and the 10-year warranty condition was only 0.54 on a seven-point scale. The low sensitivity to warranty quality changes may indicate that warranty information may not be very important in consumers' product evaluations (Olson and Jacoby 1972; Roselius 1971). Another possible reason for the small differences in evaluations with changes in warranty quality may be the presence of strong product attribute information which might have overwhelmed the relatively less important warranty information. This explanation is consistent with the finding of Olson and Jacoby (1972) that people rely more on intrinsic cues than extrinsic cues to arrive at product evaluations.

When a new brand enters a market, it has to overcome steep barriers imposed by the brand images associated with existing high-performance brands. An important implication of our research is that a useful strategy in this situation is for the new brand to differentiate itself from the competition through the use of warranties. This is because it was found that the evaluations given by subjects to the new brand with a strong warranty were equivalent to the evaluations given to an established brand with a strong reputation but no warranty. Warranty information, however, did not affect subjects' evaluations of a well-known high performance brand. Therefore, in a competitive market where consumers have well-defined images of the existing brands, our results indicate that the use of warranty by a market leader as a means of differentiation may not be as effective.

Several other extrinsic cues that have been researched in the past include price, country of origin, and store image. Future research should examine the interactive nature of these cues. However, as Rao and Monroe (1989) suggest, "...Further, the interactive effect of brand name, store name, and price on quality perceptions should be investigated." (p.356). Our research underscores the interactive nature of extrinsic cues, as pointed out by Olson and Jacoby (1972), by examining the interaction between warranty quality and brand name. Future research should examine the interactive nature of other cues.

REFERENCES

Bearden, William O. and Terence A. Shimp (1982), "The Use of Extrinsic Cues to Facilitate Product Adoption," Journal of Marketing Research, 19 (May), 229-239.

Bilkey, Warren J. and Nes, Erik (1982), Country of Origin Effects on Product Evaluations," Journal of International Business Studies, Spring/Summer, 89-99.

Jacoby, Jacob, Jerry Olson and Rafael Haddock (1971), "Price, Brand Name, and Product Composition Characteristics as Determinants of Perceived Quality," Journal of Applied Psychology, 55 (6), 570-579.

Kirk, Roger E. (1982), Experimental Design: Procedures for Behavioral Sciences, Monterey, CA: Brooks/Cole.

Lehmann, Donald and Lyman Ostlund (1972), "Consumer Perceptions of Product Warranties: An Exploratory Study," in Proceedings, Third Annual Conference of Association for Consumer Research, ed. M. Venkatesan, Chicago: Association for Consumer Research, 51-65.

Levitt, Theodore (1967), "Communications and Industrial Selling," Journal of Marketing, Vol. 31, 15-21.

Montgomery, David B. (1975), "New Product Distribution: An Analysis of Supermarket Buyer Decisions," Journal of Marketing Research, Vol. 12, 255-264.

Olson, Jerry and Jacob Jacoby (1972), "Cue Utilization in the Quality Perception Process," in Proceedings of the Third Annual Conference of the Association for Consumer Research, ed. M. Venkatesan, Iowa City: Association for Consumer Research, 167-179.

Perry, Michael and Arnon Perry (1976), "Service Contract Compared to Warranty as a Means to Reduce Consumer's Risk," Journal of Retailing, 52 (Summer), 3340.

Rao, Akshay and Kent Monroe (1989), "The Effect of Price, Brand Name, and Store Name on Buyers' Perceptions of Product Quality: An Integrative Review," Journal of Marketing Research, 26 (August), 351-357.

Roselius, Ted (1971), "Consumer Rankings of Risk Reduction Methods," Journal of Marketing, 35 (January), 56-61.

Ross, Ivan (1975), "Perceived Risk and Consumer Behavior: A Critical Review," in Advances in Consumer Research, Vol. 2, ed. M. Schlinger, Cincinnati: Association for Consumer Research, 1-19.

Shimp, Terence and William Bearden (1982), "Warranty and Other Extrinsic Cue Effects on Consumers' Risk Perceptions," Journal of Consumer Research, 9 (June), 3846.

Simon, Marji (1970), "Influence of Brand Names on Attitudes," Journal of Advertising Research, 3 (June), 28-30.

Wiener, Joshua (1985), "Are Warranties Accurate Signals of Product Reliability?," Journal of Consumer Research, 12 (September), 245-250.

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