Advances in Consumer Research Volume 11, 1984 Pages 348-352
RECALL AND RECOGNITION MEMORY FOR PRODUCT ATTRIBUTES AND BENEFITS
Lorne Bozinoff, Bell Canada
Victor J. Roth, University of Toronto
The distinction between product attributes and product benefits is reviewed in light of the widely-held belief that consumers seek product benefits rather than product attributes. An empirical study is undertaken to investigate consumer recall and recognition memory for product attributes and benefits. The empirical study yielded several findings. First, benefit recall was not always superior to attribute recall. Second, benefit recognition memory was also not always superior to attribute recognition memory. Third, consumers do not automatically transform product attributes into the relevant benefits. Fourth, presentation format can enhance the linkage between product attributes and benefits. And finally, neither product class familiarity or product class interest appear to be strongly related to attribute or benefit memory. Overall, these findings contradict the widely-held belief that consumers seek product benefits and not product attributes. This study suggests that advertising effectiveness can be maximized by designing ads which provide the combination of attributes and/or benefits .hat consumers seek in each product.
It has long been held that consumers utilize multiattribute decision-making processes when evaluating products (Wilkie and Pessemier 1973; Shocker and Srinivasan 1979). More recently, it has been argued that consumers evaluate products more on the basis of product benefits than on the basis of product attributes. According to this conceptualization, consumers seek certain benefits which will fulfill their consumption needs when they evaluate products (Ratchford 1975; Green 1975; Howard 1977). Thus, consumers select products more on what needs the products will fulfill (i.e., benefits) rather than on the physical characteristics of the products (i.e., attributes). Indeed, it has been suggested that the market can be segmented on the basis of benefits sought (Haley 1968). However, benefit segmentation does require continued revision because of segment instability (Calantone and Sawyer 1978).
For illustrative purposes, consider the purchase of an automobile. Under a "multi-attribute" decision-making conceptualization, a consumer can be viewed as making a purchase decision on the basis of such attributes as "mileage" and "size". In contrast, under a "benefits sought" decision-making conceptualization, a consumer would be more likely to make a purchase decision on the basis of such benefits as "inexpensive to operate" and "easy to park". This example also illustrates the relationship between product attributes and benefits. The benefit, "inexpensive to operate" is a function of the attribute, "mileage". In a likewise manner, the benefit, "easy to park" is a function of the attribute. "size".
The distinction between product attributes (i.e., physical characteristics) and product benefits has several implications for consumer research (Myers and Shocker 1981). First, it has been suggested that some attitude models, such as the "adequacy-importance" model are more appropriate for a benefit conceptualization than for an attribute conceptualization. Conversely, the "ideal point" and "vector" attitude models may be more appropriate for an attribute conceptualization. Second, some rating scales may be more appropriate for the measurement of either benefits or attributes.
The distinction between attributes and benefits also has management implications. First, it is important that the promotion of products be undertaken from the consumer's perspective. If consumers do indeed seek product benefits, then it is crucial that benefits and not attributes be emphasized in promotional materials. And second, in order to increase advertising effectiveness, it is important that the process by which consumers transform product attributes (i.e., physical characteristics) into Product benefits be understood.
THE STRUCTURE OF LONG TERM MEMORY
The question of whether consumers evaluate products in terms of attributes or benefits also impacts upon how information is stored in long term memory (LTM). An attributes conceptualization implies that information concerning products is stored in LTM in the form of attributes. Alternately, a benefits conceptualization implies that information concerning products is stored in LTM in the form of benefits.
According to most theories of long term memory, knowledge is stored as schemata (Lindsay and Norman 1977). Schemata are the large sets of well-structured cognitions that have been learned through experience over time (Norman and Bobrow 1977). Schemata consist of frameworks for organizing the information about a concept. The format of the schemata proposed by current theories of memory is a propositional representation (Lindsay and Norman 1977; Anderson and Bower 1973). A product attribute conceptualization implies that these propositions are in the form of attributes, while a product benefit conceptualization implies that these propositions are in the form of benefits. This also implies that if product information is stored in LTM in the form of product attributes, the memory for product attribute information should be superior to the memory for product benefits. Conversely, the opposite prediction could be made if consumers store product information in LTM in the form of benefits.
The study described below serves only as an initial, exploratory investigation of the ideas discussed above. The purpose of this study is to answer six questions concerning the recall and recognition memory for product attributes and benefits. First, in a free recall task, do consumers tend to recall more product attributes or product benefits? Second, in a recognition task, do consumers tend to recognize more product attributes or product benefits? Third, do consumers tend to transform and link product attributes to the relevant product benefits? For example, is the car attribute "mileage" automatically associated by the consumer with the benefit "inexpensive to operate"? Fourth, is it possible to strengthen the link between product attributes and the relevant products benefits? That is, if consumers do not automatically transform attributes into benefits, can this transformation be strengthened by promotional activities? Fifth, is product class familiarity related to the recall of product attributes and benefits? It is possible that more familiar consumers may have a greater recall for attributes and/or benefits than less familiar consumers. And sixth, is consumer interest in a product class related to the recall of product attributes and benefits? It is not unreasonable to expect that consumers who are interested in a product class will recall more attributes and benefits than uninterested consumers.
It is important that both recognition and recall be evaluated. Empirical studies have shown that there are several factors which differentially affect recognition and recall. These factors include frequency of stimuli occurrence, learning plans, rehearsal times and arousal levels (Bettman 1979).
One hundred and fifty-six subjects were recruited from four undergraduate marketing management classes. During regular class time, students were asked to participate in the study. Participation was voluntary, although all of the students agreed to participate. Approximately 60% of the sample were males. A convenience sample of students is defensible since the principal aim of the present research is exploratory in nature. In addition, Kruglanski (1975) has argued that homogeneous samples (such as students in particular classes at a given institution) are appropriate for theory oriented research where individual differences (i.e., demographics) are not of theoretical interest.
The subjects were given questionnaire booklets containing three new product descriptions and a series of questions concerning each product. The three products were a video phone, a convenience food and an electric car. These products were chosen because they were all relatively new to the subjects. Therefore, it was possible to assume that most of the information concerning these products was gained from the new product descriptions. Each new product description contained five attributes and five related benefits. The order of the new product description was rotated in order to eliminate any order bias.
Prior to reading each new product description, the subjects answered a Likert-type question designed to measure product class familiarity. The subjects then fully read each new product description and answered a Likert-type question designed to measure product class interest. The subjects were then given a ten-minute distractor task. The distractor task was included in order to ensure that the short term memory of the new product descriptions was eliminated.
After the distractor task, the subjects were asked to write down everything that they could remember from each of the new product descriptions. This was the free recall measure. Finally, the subjects were given a recognition test in which they had to indicate whether each of seven product attributes and seven product benefits were mentioned in each of the new product descriptions. The rating scale ranged from l (definitely mentioned) to 4 (definitely not mentioned). Mean rating scale values then made up the recognition measure. The list of attributes and benefits included all the attributes and benefits which were mentioned (five each) plus some attributes and benefits which were not mentioned (two each).
The classes were randomly assigned to one of three treatments. Subjects in the two classes which were assigned to the first treatment received new product descriptions in which the attributes and benefits were linked together, with the attribute immediately preceding the relevant benefit (AB treatment). For example, the electric car description contained the phrase, "compact size lets you park anywhere". The attribute, "compact size" immediately proceeds the relevant benefit, "lets you park anywhere". The sample size of this treatment was eighty-one.
The new product description used in the second treatment contained the same attributes as in the first treatment. However, in the second treatment, the benefits immediately preceded the relevant attributes (BA treatment). For example, the electric car description contained the phrase, "lets you park anywhere because of its compact size". The benefit, "lets you park anywhere" immediately precedes the relevant attribute, "compact size". The sample size for this treatment was fifty.
In the third treatment, the new product descriptions contained the same attributes and benefits as in the two other treatments. However, the relevant attributes and benefits were not listed together. Rather, the attributes and benefits were listed randomly throughout the new product descriptions. For example, in the electric car description, "lets you park anywhere" was not mentioned together with "compact size" but appeared elsewhere in the new product description. The sample size for this treatment was twenty-five.
Recall of Product Attributes and Benefits
The mean number of attributes and benefits recalled for each of the three new products are presented in Table 1. (The results were aggregated across all treatments.) Pairwise t-tests were calculated to test for mean differences in the number of attributes and benefits recalled. If consumers do seek product benefits rather than product attributes, then the recall of benefits should be superior to the recall of product attributes. For both the video phone and the electric car, significantly (p C .01) more benefits than attributes were recalled. However, in the case of the convenience food, significantly (p < .01) more attributes than benefits were recalled. (This same pattern of results held for each of the three treatment groups also.) These results indicate that the recall of benefits is not always superior to the recall of attributes.
Recognition of Product Attributes and Benefits
The mean attribute and benefit recognition scores, together with the mean false recognition scores, are presented in Table l. A series of pairwise t-tests were calculated to test for mean differences between attribute and benefit recognition scores. For the video phone, the mean recognition score for benefits was significantly (p < .01) lower than the mean recognition score for attributes. This means that recognition memory was better for benefits than attributes. However, for the convenience food, the opposite result was found. Mean recognition scores were significantly (p < .01) lower for attributes. No differences were found in the mean recognition scores for the electric car. Recognition memory appears to be better for benefits than attributes in the case of the video phone and better for attributes than benefits in the case of the convenience food.
RECALL AND RECOGNITION OF ATTRIBUTES AND BENEFITS
A series of pairwise t-tests were also calculated to test for mean differences in false recognition scores. False recognition scores were significantly (p < .05) lower (i.e., more false recognition) for benefits than attributes for both the video phone and the electric car. However, the false recognition scores were significantly (p < .01) lower (i.e., more false recognition) for attributes than benefits for the convenience food. There appears to be more false recognition of benefits than attributes for both the video phone and the electric car. On the other hand, there appears to be more false recognition of attributes than benefits for the convenience food.
A clear pattern of consumer memory findings emerge from these results. In the case of video phone, both recall and recognition memory were better for benefits than attributes. In addition, there was more false recognition of video phone benefits. This suggests that for the video phone, subjects believed that more benefits than attributes were mentioned. This pattern also tended to hold for the electric car. However, in the case of the convenience food, recall and recognition memory were better for attributes than benefits. In addition, there was more false recognition of attributes. This suggests that for the convenience food, subjects believed that more attributes than benefits were mentioned.
The recall of each attribute was cross-tabulated with the recall of the relevant benefit. The cross-tabulations were done for each of the three treatments. The results are reported in Table 2. If subjects, on their own, automatically link attributes to the relevant benefits, then there should be significant cross-tabulations of recalled attributes and relevant benefits for the RARB treatment. This treatment did not present the attributes and relevant benefits together. Rather, the attributes and benefits were randomly presented in the new product description. As Table 2 shows, in the RARB treatment, the recall of attributes was never significantly related to the recall of the relevant benefits. These results indicated that subjects on their own do not tend to link attributes to the relevant benefits.
It remains to be shown, however, that if the attributes and relevant benefits are presented together, they will be recalled together. The results for the AB and BA treatments, in which the attributes and relevant benefits were presented together, clearly show that when attributes are presented together with the relevant benefits, they are recalled together (Table 2). For both the AB and BA treatments, six of the fifteen pairs of attributes and relevant benefits tended to be recalled together by the subjects. This is an indication that while subjects do not automatically link attributes and benefits, presentation format can enhance the linkage.
Product Class Familiarity
The sample was split into two groups based on the subjects' degree of product class familiarity. Table 3 presents the mean number of attributes and benefits recalled by the low and high product class familiarity groups. While it was expected that the group with the higher degree of product class familiarity would recall more attributes and benefits than the other group, only in the case of convenience food benefits did the group with the higher degree of product class familiarity show significantly (p < .10) greater recall.
Product Class Interest
The sample was also split into two groups on the basis of subjects' degree of product class interest. Table 3 presents the mean number of attributes and benefits recalled by the low and high product class interest groups. Again, it was predicted that subjects who had a higher degree of product class interest would recall more attributes and benefits. However, only in the case of the electric car attribute was recall significantly (p < .05) higher for the group with the higher degree of product class interest.
The present study provides some tentative answers to the questions raised earlier. First, it has been shown that the superiority of benefit recall over attribute recall varies across products. It was shown that for two of the products, the video phone and the electric car, recall of benefits was superior to the recall of attributes, but that for the other product, the convenience food, the opposite relationship was found. In terms of the long term memory models discussed earlier, these results are ambiguous. Support has been found for both an attribute based and a benefit based model.
ATTRIBUTE AND BENEFIT RECALL CROSS-TABULATIONS
RECALL OF ATTRIBUTES AND BENEFITS BY PRODUCT CLASS FAMILIARITY AND PRODUCT CLASS INTEREST
Second, it was also shown that recognition and false recognition superiority also varies across products. This suggests that consumers do not always seek benefits or look for attributes in products. Rather, this suggests that for certain kinds of products, benefits are sought while for other products, consumers look for attributes. Based on these results, it appears that consumers look for attributes in relatively-simple products (e.g., the convenience food) but seek benefits in more technical products (e.g., the video phone, the electric car).
Third, it was clearly shown that consumers do not automatically link attributes to their relevant benefits. When given a randomized presentation of attributes and benefits, the pairs of attributes and relevant benefits were not recalled together. Fourth, it was shown that this linkage can be enhanced by presenting the pairs of attributes and relevant benefits together.
Fifth, only limited support was found for the prediction that product class familiarity would be related to attribute and benefit recall. Only in the case of one product was this relationship found for benefits. And sixth, little evidence was found for the prediction that product class interest would be related to attribute and benefit recall. This relationship was only found for electric car attributes.
CONCLUSIONS AND IMPLICATIONS
Overall, this study questions the widely-held assumption that consumers seek product benefits rather than product attributes. Rather, it appears that consumers look for both attributes and benefits, depending upon the product. Furthermore, it does not appear that consumers tend to automatically transform attributes into benefits.
For consumer research, this study has an important implication. It was noted earlier that some attribute models are more appropriate for a benefit conceptualization while other attitude models are more appropriate for an attribute conceptualization. This study has demonstrated that consumers tend to process both attribute and benefit information, and that the relative amount of attribute and benefit information processed varied across product classes. This means that no single attitude model is likely to be appropriate for all product classes. Rather, it is important that each product class be investigated to determine whether attribute or benefit information is being processed by consumers.
There are also two managerial implications flowing from this study. First, advertising effectiveness can be maximized by designing ads which provide the combination of attributes and/or benefits that consumers seek in a product. Thus, ads should not necessarily stress only benefits or attributes. Second, it is equally important that the ads provide the linkage between product attributes and benefits, if the advertiser wishes consumers to make the link. Consumers will not transform attributes into benefits automatically. This study contradicts the widely-held belief that consumers seek product benefits and that they transform product attributes into benefits. This suggests that a benefit segmentation strategy may be inappropriate for some product markets.
Bettman, James R. (1979), An Information Processing Theory of Consumer Choices, Reading, MA: Addison-Wesley.
Bower, Gordon H. (1975), "Cognitive Psychology: An Introduction," in Handbook of Learning and Cognitive Processes, William R. Estes, ed., Hillsdale NJ: Erlbaum and Associates, 25-80.
Calantone, Roger 3. and Alan G. Sawyer (1978), "The Stability of Benefit Segments," Journal of Marketing Research, 15 (August), 395-404.
Green, Paul E. (1975), "Marketing Applications of MDS: Assessment and Outlook," Journal of Marketing, 39 (January), 24-31.
Haley, Russell I. (1968), "Benefit Segmentation: A Decision-Oriented Research Tool," Journal of Marketing, 32 (July), 30-35.
Howard, John A. (1977), Consumer Behavior Applications of Theory, New York: McGraw Hill.
Kruglanski, Arie (1975), "The Two Meanings of External Validity," Human Relations, 28, 653-659.
Lindsay, Peter H. and Donald E. Norman (1977), Human Information Processing, New York: Academic Press.
Myers, James H. and Allan D. Shocker (1981), "The Nature of Product-Related Attributes," in Research in Marketing, Vol. 5, Jagdish Sheth, ed., Greenwich CT: JAI Press, 211-326.
Norman, Donald A. and Daniel F. Bobrow (1975), "On The Role of Active Memory Processes in Perception and Cognition," in The Structure of Human Memory, Charles N. Cofer, ed., San Francisco CA: W.H. Freeman, 114-132.
Ratchford, Brian T. (1975), "The New Economic Theory of Consumer Behavior: An Interpretive Essay," Journal of Consumer Research, 2 (September), 65-75.
Shocker, Allan D. and V. Srinivasan (1979), "Multiattribute Approaches to Product Concept Evaluation and Generation: A Critical Review," Journal of Marketing Research, 16 (May), 159-180.
Wilkie, William L. and Edgar A. Pessemier (1973), "Issues in Marketing's Use of Multi-Attribute Attitude Models," Journal of Marketing Research, 10 (November), 423-441.