On the Transferability of Feature/Level Preferences Across Competing Products Serving the Same Purposes

ABSTRACT - This paper explores implications of a change in context or goal on product evaluations. Common consumption goals can lead to quite different products being grouped together in a consumer's consideration set. Previous research had demonstrated this, but did not provide an explanation for the mechanism behind it. Using categorization research and research on importance of goals and contextual determinants of human knowledge, this paper suggests that goal-derived categories and the process of their evaluation might be at the heart of the transferability of preferences across competing products serving the same purpose.

Citation:

Milos D. Graonic and Allan D. Shocker (1993) ,"On the Transferability of Feature/Level Preferences Across Competing Products Serving the Same Purposes", in NA - Advances in Consumer Research Volume 20, eds. Leigh McAlister and Michael L. Rothschild, Provo, UT : Association for Consumer Research, Pages: 389-394.

Advances in Consumer Research Volume 20, 1993      Pages 389-394

ON THE TRANSFERABILITY OF FEATURE/LEVEL PREFERENCES ACROSS COMPETING PRODUCTS SERVING THE SAME PURPOSES

Milos D. Graonic, University of Minnesota

Allan D. Shocker, University of Minnesota

ABSTRACT -

This paper explores implications of a change in context or goal on product evaluations. Common consumption goals can lead to quite different products being grouped together in a consumer's consideration set. Previous research had demonstrated this, but did not provide an explanation for the mechanism behind it. Using categorization research and research on importance of goals and contextual determinants of human knowledge, this paper suggests that goal-derived categories and the process of their evaluation might be at the heart of the transferability of preferences across competing products serving the same purpose.

INTRODUCTION

A number of recent studies have indicated the importance of taking into account consideration set influences on consumer decision-making and choice (Srivastava et al. 1984; Nedungadi 1990; Ratneshwar and Shocker 1991). This research has demonstrated that prior to choice, a consumer can sometimes consider a set of products from different nominal product categories (termed "non-comparables" by Johnson [1984]). These studies do not suggest a mechanism behind the phenomenon. A reasonable speculation regarding such a mechanism is that all the products in the consideration set may offer the same benefits in the given context and could therefore compete. An especially interesting issue for research arises from the logical conclusion that preferences for (i.e., judgments about) at least some attributes or characteristics of such alternative products may be transferable from one product to another, since competing products would be evaluated on their ability to serve the same purpose(s).

This study is exploratory and is the first in a series of studies aimed at explaining the mechanism behind the formation of consideration sets containing noncomparables, and how preferences for different products serving the same purpose could be transferable from one product category to another. A potential explanation and theoretical foundation are suggested by recent research on the role of goals and context in people's decision making and categorization research.

PREVIOUS RESEARCH AND THEORETICAL FOUNDATION

Research in Consumer Behavior

There has been much literature which directly or indirectly addresses issues of concern in this research. As we have already mentioned, consideration set research (for an overview see Shocker, Ben-Akiva, Boccara, and Nedungadi [1991]) and Johnson's (1984) research on noncomparables are two important streams of such research. Two findings from consideration set research are central for the present study. First, this research indicates that people often recall from memory a set of products to serve as a basis for eventual choice. Second, it seems that the process of choice is characterized by the movement of a product set (the composition of which could be modified), not simply an individual product, through different stages of thought before the decision is made. Johnson's (1984) research indicated that comparison of different product categories necessitates a certain level of abstraction to compare physically different products.

The Role of Goals and Context in Consumer Decision-Making

It has been demonstrated in many different areas of social sciences that human knowledge is structured and organized in different ways and by various means. For purposes of our inquiry we are interested in the role of context (situation) and goals in human knowledge and, therefore, consumer decision making. We argue that context and goals impose important constraints on human knowledge and therefore must be included in the study of consumer choice process and choices.

First, consider the role of context. It has always been easy to criticize knowledge-related experiments from the standpoint that testing people's skills at performing a particular task is not valid if the task is not relevant to the "real world" (Lynch 1983). In other words, laboratory experiments should desirably be relevant to the real world to serve the purpose of explaining and predicting real world behavior and determining a level of skill (Dreyfus and Dreyfus 1986). It might seem as though people make inconsistent decisions in experiments, but often they are consistent with respect to the criterion they are using, if it was properly understood by the experimenter. Human understanding is organized by perspective; a simple rewording of a question on a choice task can change the choice.

Thinking is intrinsically interwoven with the context of the problem to be solved. Context includes a problem's physical structure, it's conceptual structure, the purpose of the activity, and a social milieu (Rogoff 1984). This implies that even an experimental setting, which is intended to be "neutral" and context free, provides a particular context. Thinking, as a practical activity, is adjusted to meet demands of the situation. Therefore, logical problem-solving in one environment may not be logical problem-solving in another (e.g., see the results of research on the Wason selection task in Cosmides [1989]).

A very good example for the contextual dependency of skills in consumer behavior is presented by Lave, Murtaugh, and Rocha (1984). The authors studied grocery shopping as an example of everyday activity in context. They compared results on an identical arithmetic task (selection of the least expensive alternative based on comparisons of price and quantity) during an actual shopping process and in a laboratory situation. Supermarket calculations were much more accurate than laboratory calculations (98% error free compared to 59% error free). Not only were the shoppers more precise, but they also carried out many more calculations (2.5 calculations on average) for each grocery item. The authors hypothesized that this happened because the supermarket situation juxtaposes problem, solution, and checking activities, i.e., a realistic setting for a problem creates better solutions.

A very broad picture of the role of goals in consumer behavior was presented by Gutman (1982) who used means-end analysis to explain how a product or service selection facilitates the achievement of a desired end state. His model assumes that people are aware of product benefits and their consequences, and their goal is to choose the product whose benefits can satisfy values the consumer cherishes. Indeed, other experimental research has also demonstrated that consumer goals are drivers in the choice process (Park and Smith 1989; Bettman and Sujan 1987; Huffman and Houston 1992).

The cognitive psychology literature has also indicated a role for goals as a basis for categorization (Barsalou 1983). In the course of engaging in goal-directed behavior people often create specialized concepts called goal-derived categories. For example, a goal to lose weight can create the category of "foods not to eat on a diet." Since human behavior is believed to be largely goal-driven, this suggests that goal-derived categories might prove useful in the explanation of preferences and choices.

Theories about Concepts and Goal-based Categorization

The dominant approaches to human categorization (i.e., prototypicality and exemplar) have been criticized in the cognitive psychology literature and appear unsatisfactory for understanding substitution among products. They do not adequately deal with the question of how categories initially are formed, but rather they accept a category as given and seek to describe its composition and membership. A promising new approach is offered by Medin and his associates (for a more complete conceptualization, see Medin 1989). They claim that the complexity of human categorization could be explained by a "theory-driven" approach.

Our concepts are often embedded in "theories" about the way the world functions. We often find a deeper essence and use that to provide meaning when other information is not available. That is, our "theories" help us to construct meaning which can take us beyond the explicit information given. Those "theories" are very often of causal (cause-effect) relationships and are of an explanatory nature. A goal-derived category involves "knowing" a set of dynamic causal relations. Therefore concepts are dependent upon a network of relations in which they are embedded.

It is possible to find numerous examples of people using some kinds of personal theory when categorizing by looking at published experimental work. For example, when asked to list the attributes of a particular category, respondents would most often list particularly salient attributes. In addition, respondents would also list some non-salient attributes, but ones which are diagnostic in people's background knowledge (Murphy and Medin 1985). As another example, assumed correlations between the attributes in a concept have been shown not to be random ideas, but the result of thought processes caused by very different stimuli at different times (Malt and Smith 1984). Even more, people often see illusory correlations which can not be based on facts, but only upon their "theories" (Medin and Shoben 1988).

There are already two proposals for prospective theories of product categorization in the literature of consumer behavior. First, Boush and Loken (1991) proposed categorization based on brand. All products that have the same brand could be considered as one category regardless of their parallel membership in different product classes. Second, Bettman and Sujan (1987) and Park and Smith (1989) demonstrated that Barsalou's idea (1983) of goal-related categories could very usefully be applied to group products by their ability to satisfy the same goal. This last idea represents the approach we use in the present research. Since goal-derived categories are not already given common taxonomic categorization, for example as fruit or animals, to be able to consider different nominal products as members of the same category, a reasonable conjecture is that consumers theorize how the different products could fulfill the same goal. Creating a category using members from different product classes thus becomes an example of a theory-driven approach to categorization. Only personal cause-effect theories could help people understand how a set of quite different attributes embedded in different products could offer the same benefit(s).

RESEARCH FRAMEWORK AND HYPOTHESIS

The research framework encompasses several elements, as shown in Figure 1. Context is a general representation of the situation in which a consumer might be involved or expect to be and which is presumed to impose constraints upon his or her decision. A context could have several dimensions as, for example, social (there might be other participants), temporal (particular time of day or night), spatial (at home, at work, in a mall), etc. See Belk (1975) for such a concept of the situational context.

A context relevant for decision-making could suggest particular goals to the decision-maker or the decision-maker seeking a particular goal might choose a particular context. Let's say, if somebody is in McDonald's only a limited number of goals would be plausible - for example, to eat fast food oneself, to accompany friends or family who came to eat fast food, etc. Choice of goals is delimited by the particular context. In the present research we do not distinguish between goals and context and specify both together as part of our instructions to respondents. Context is, of course, a broader construct than a goal. It can encompass several goals (and the same goal could be relevant to several different contexts). But, there is frequently a strong and plausible relationship between context and goal so that specification of one may suggest the other.

Given a goal there may be a set of products which can satisfy and fulfill the goal to an acceptable degree. For example, if a student lives only a few miles from school and wants to ride there on a pleasant sunny day, s/he could use a bicycle, in-line skates, car, bus, etc. for this purpose (assuming the availability of all these alternatives and the skill to use each). Although all these products could satisfy the goal of "getting to school" they don't have identical physical characteristics. So, the different products could represent members of a goal-derived category. In spite of their different physical characteristics, all products could offer the benefits needed to achieve the goal. The benefits or choice criteria could be, for example, "amount of comfort", "length of time to get to school", "enjoyment level of the physical activity", etc. Therefore, instead of comparing products on physical characteristics, consumers may compare "non-comparables" on the benefits required for the particular goal.

Trade-offs among benefits and costs can be thought to be a natural outcome of goal-driven decision-making - whether comparables or non-comparables are involved as choice objects (see the substantial literature on conjoint analysis and other multi-attribute models of decision-making for support for this contention). Consumption purpose (goal) proscribes criteria (benefits) which the alternatives considered (e.g., products/services) must meet.

An obvious practical problem lies in the definition of what are benefits and what are attributes. There may be no precise line which would separate benefits (which are generally more abstract) from attributes (generally more concrete). The abstractness-concreteness dimension (Johnson and Fornell [1987]) is not always unambiguous in distinguishing benefits from attributes (e.g., low cost may be regarded as a benefit, yet may be quite concrete).

In considering how attributes may relate to benefits, it is important to recognize that there may be more than one way to create a given benefit. For example, if the goal and context is to "get to school rapidly (goal) on a pleasant sunny day (context)" and if the products considered are a bicycle and in-line skates, it is possible that a benefit such as "comfort" could be generated for both products in two very different ways:

Bicycle:                                       In-line skates:

Seat cushioning                           Weight

Size of frame                               Lacing/Closure

Type of bicycle                           Design of shoe

Curvature of handle bar               Fit of boots

Pedal type/style                           Padding in shoe

FIGURE 1

FRAMEWORK FOR PRODUCT EVALUATIONS IN CONTEXT

Attributes unique to each given product category can plausibly give rise to a similar benefit. Thus, a product can be represented in our framework by both its attributes (possibly unique) and the levels of benefits it provides. In the same product category, attributes are likely to be encoded into benefits similarly; while in the case of "non-comparables," alternatives may be comparable only in terms of their benefits. After considering the levels of benefits and/or attributes provided by each alternative, consumers make their choice; therefore the final element in our framework is the decision.

If different products serve the same goal, there must be a way to relate and maybe transfer preferences between these different products. A first step in pursuing this idea would be to see whether preferences for the same products change with a change in goal. Since the present study is first in a series, our first hypothesis for the project is the one that is empirically investigated here:

Hypothesis 1: The importance of features in customer preferences for the same product will change with a change in goal or purpose.

METHOD

Overview

In order to investigate whether preferences for different and identical products change as goals change, we conducted an exploratory study. It was of particular importance to construct "natural" situations with goals and products that the particular subject population might be involved with in their everyday lives (Snyder 1981). Since we planned to sample from an undergraduate student population, we focused on products and related goals which were relevant to that population.

Through a series of (3) focus groups (each involving 2 - 6 students, who were paid for their participation) and a pretest (involving some 37 undergraduate marketing students in a large midwestern university who were given class credit for participating) we identified two product sets and two distinct goals for each which appeared to enable a test of the hypothesis. Each set consisted of two products. The sets were constructed in a way which would enable investigation of the phenomena. Two products in one set are distinctly different, while two products in the other set are similar, but "different." The first consisted of a bicycle and in-line skates and the two goals were "To travel a few miles to school on pleasant, sunny days" and "To exercise on pleasant, sunny days". These products are "non-comparables" since they don't have most or all physical attributes in common. Another set of products selected were a backpack and shoulder bag. These were considered by the students as different products, but were similar enough so that they could be represented by similar attributes, with the main differences being in styling and appearance (e.g., backpacks have two shoulder straps while a shoulder bag has one). The goals were "To carry books, notes, pencils, etc. to school" and "To carry things on an airplane trip".

In the focus groups, we had asked participants to take a given product and to suggest alternative usages for that product and other products they would consider for those same usages. The benefits suggested by the students were determined from the usages suggested and the general discussion surrounding conditions of use. From all the potential benefits that both products can offer, we selected those common to both goals, i.e., which were suggested in both contexts of use for the products.

TABLE 1

RANK ORDER CORRELATIONS OF ATTRIBUTE IMPORTANCES GIVEN CONTEXT-SHOULDER BAG AND BACKPACK

Self-administered questionnaires were completed by 74 undergraduate marketing students who participated in the research to receive course credit. Manipulation of the goals was conducted "between subjects" so that different subjects were randomly assigned to one of two different goals. Each questionnaire contained both product sets (i.e., bicycle - in-line skates and backpack - shoulder bag) and only one particular goal for each product set. We could potentially get a confound when participants evaluated more than one product for each specific context but, given the small number of respondents available and the duration of the task, that possibility seemed necessary.

Participants were asked to indicate the importance of each in a list of pre-specified benefits and then the importance of the prespecified attributes for both products, given the specified goal. Next participants indicated to which degree a particular product offers benefits. Thus there were three predictor variables: judged importances of each benefit, judged importances of each attribute, and degree to which a product offers a benefit. All three variables were measured using 5-point scales: the responses for importance questions ranged from "Not at all important" to "Extremely important" and the responses for the degree to which products offer benefits ranged from "Not at all" to "Extremely high". Finally we collected data on ownership, involvement, and individual characteristics of participants.

ANALYSIS AND RESULTS

To test the hypothesis we conducted two related sets of analyses. First, we used aggregate rank order correlations to determine whether and how rankings of importances of attributes and benefits differed between the two contexts. Second, logistic regression was applied to determine whether attribute and benefit importances differed between two contexts and, if they do, which attributes and benefits in an overall regression model could be used to predict the goal participants had in mind (the dependent variable) when evaluating attribute and benefit importances.

As shown in Tables 1 and 2, rank order correlations of attribute importances changed from one context to another. In other words, the importances of attributes (physical characteristics of the products) change as context/goal changes, although the product itself is not changed. In the case of the backpack and shoulder bag (the two products which had the same set of attributes) rank order correlations were higher for two "different" products evaluated within the same context than for two "identical" products evaluated in two different contexts (although the latter correlation was computed "between subjects," thereby confounding the interpretation). This outcome is repeated for all four situations (2 products x 2 contexts) as shown in Table 1.

As noted in Table 2, since bicycles and in-line skates don't share attributes, rank order correlations of attribute importances could not be compared. But, again in the case of bicycles importances of attributes changed between the two contexts (the rank order correlation is .79). In the case of in-line skates, attribute importances did not change between the two contexts - an unexpected result. Three explanations occur to us. First, our manipulation of context/goal may not have been sufficiently different in the case of this product (subjects may have not seen going to school and recreation as sufficiently different in terms of what they would expect from the product). Second, in-line skates, being a relatively new product (on the market only a few years), it is possible that consumers still don't have as sophisticated an opinion about the product and its potential usages as they have for the other products used in the research (e.g., the incidence of ownership of in-line skates was markedly lower than for bicycles in our sample). Finally, it is possible that expertise could play a role here. We were able to test this final possibility (using self-reported measures of product knowledge) and found no expertise effect.

Rank order correlations of benefits were expected to change even more between two contexts than were the rank order correlations of attributes. As shown in Table 3, rank order correlation of benefits in the case of shoulder bags and backpacks reveals a change in importances between two contexts. But, rank order correlation in the case of bicycles and in-line skates reveals extreme change from one context to another. In other words, the importance of benefits almost changes completely from one context to another, in the case of bicycle and in-line skates.

TABLE 2

RANK ORDER CORRELATIONS OF ATTRIBUTE IMPORTANCES GIVE TWO DIFFERENT CONTEXTS

TABLE 3

RANK ORDER CORRELATIONS OF BENEFIT IMPORTANCES GIVEN TWO DIFFERENT CONTEXTS

TABLE 4

PERCENTAGE CHANGE IN IMPORTANCE OF ATTRIBUTES AND BENEFITS BETWEEN TWO CONTEXTS

A complementary analysis made use of logistic regression. Again, we used data on attribute and benefit importances as independent variables and context/goal as the dependent variable. The objective was to determine, from a regression model, which attributes and benefits differ between the two contexts. In other words, which attributes or benefits could better be used to account for the context.

The summary results from the logistic regressions are shown in Table 4. As we can see, in the case of all the products, the importances of attributes and benefits is different between the two contexts. It is worth noting that the importances of benefits change much more than the importances of attributes. This suggests that it may not be enough to test for differences in attribute importances between two contexts; rather it may be more important to test for differences in benefit importances.

DISCUSSION AND FUTURE RESEARCH

This exploratory research was conducted to see whether differences in product evaluations occur with a change in context/goal. Our preliminary analyses provide several interesting results. First, context changes judgments of attribute and benefit importances. Second, the data suggest the possibility that different products could be more similarly evaluated within a given context than the same product would within two different contexts. Finally, the importances of benefits appear to change more than the importances of attributes between the two different contexts.

The results could have important marketing implications. First, they demonstrate that for different purposes (goals) consumers could use different criteria when evaluating the same products. Therefore, assuring constancy of purpose (i.e., controlling for goals or usage) may be very important for modeling consumer decision-making (e.g., through conjoint analysis, discrete choice modeling, etc.). These implications could be felt not only in product design, but also in promotion (different features might be emphasized when different contexts are emphasized), pricing (the product could possess monopoly power in some contexts), display (complementary products with differential packaging emphasized), etc. Second, the idea that purpose changes evaluations could be strategically used to help a manager better understand the nature of competitive threat and more precisely define and segment "customers" (i.e., all consumers who encounter relevant purposes could be regarded as potential customers, even if they have not purchased in the category before).

Evidence is growing that goals play an important role in human decision-making. We have speculated about that role and provided limited evidence for our speculations in this exploratory research. One of the advantages of conducting such work is the opportunity to learn from these experiences. Future research in this arena must contend with several problems and issues. For example, we hope to analyze future work at the individual-level and not just at an aggregate level, as was done in the present study. Individual level analysis is in accord with the theory and eliminates aggregation bias (possibly at expense of greater demand effects). We expect to use multiple measurement techniques, possibly conjoint analysis and direct measurement (e.g., Fishbein attitude measurement or self-explicated measures). Next, it seems necessary to investigate in future research the "ease" with which subsets of product attributes or characteristics of different products can be related to the functional benefits and costs used in the study. One of the implications of the theoretical perspective we have adopted is enhanced recognition for the fact that different subsets of product characteristics may provide similar benefits (or the same characteristics are functional in different ways in different product applications). We anticipate using subjective judgments of "experts" to provide evidence in support of hypothesized attribute - benefit relationships. Finally, the theoretical basis we have posed for the research while plausible, has not been tested previously. Our first concern in this sequence of studies is to investigate support for the hypothesized effects and relationships; a second is to investigate alternative theoretical explanations for the findings.

REFERENCES

Barsalou, L.W. (1983). Ad hoc categories. Memory & Cognition. 11 (3), 211-227.

Belk, Russel W. (1975), Situational Variables and Consumer Behavior, Journal of Consumer Research, Vol. 2, December, 157-236.

Bettman, James R. and Mita Sujan (1987), Effects of framing on Evaluation of Comparable and Noncomparable Alternatives by Expert and Novice Consumers, Journal of Consumer Research, Vol. 14, September, 141-154.

Boush, David M. and Barbara Loken (1991), A Process-Tracing Study of Brand Extension Evaluation, Journal of Marketing Research, 28 (February), 16-28.

Cosmides, Leda (1989), The logic of social exchange: Has natural selection shaped how humans reason? Studies with the Wason selection task, Cognition, 31, 187-276.

Dreyfus, Hubert L. and Stuart E. Dreyfus (1986), Mind Over Machine - The Power of human Intuition and Expertise in the era of Computer, New York: The Free Press.

Gutman, Jonathan (1982), A Means-End Chain Model Based on Consumer Categorization Processes, Journal of Marketing, 46 (Spring), 60-72.

Huffman, Cynthia and Michael J. Houston (1992), "Goal-Oriented Experiences and the Development of Knowledge", Working paper, University of Pennsylvania.

Johnson, Michael D. (1984), Consumer Choice Strategies for Comparing Noncomparable Alternatives, Journal of Consumer Research, Vol. 11, December, 741-753.

Johnson, Michael D. and Claes Fornell (1987), The Nature and Methodological Implications of the Cognitive Representations of Products, Journal of Consumer Research, Vol. 14, 214-228.

Lave, Jean, Michael Murtaugh, and Olivia de la Rocha (1984), In Barbara Rogoff and J. Lave (Eds.), Everyday Cognition: Its Development in Social Context, Harvard University Press, 67-94.

Lynch, John G. (1983), The role of External Validity in Theoretical Research, Journal of Consumer Research, Vol. 10, 109-111.

Nedungadi, Prakash (1990), Recall and Consumer Consideration Sets: Influencing Choice Without Altering Brand Evaluations, Journal of Consumer Research, 17 (December), 263-76.

Medin, D. (1989). Concepts and Conceptual Structure. American Psychologist, 44, 1469-1481.

Murphy, G.L and D.L. Medin (1985), The Role of Theories in Conceptual Coherence, Psychological Review, 92, 289-316.

Park, C. Whan and Daniel C. Smith (1989), Product-Level Choice: A Top-Down or Bottom-Up Process?, Journal of Consumer Research, Vol. 16, December, 289-299.

Ratneshwar, S. and Allan D. Shocker (1991), Substitution in Use and the Role of Usage Context in Product Category Structures, Journal of Marketing Research, 28 (August), 281-95.

Rogoff, Barbara (1984), Introduction: Thinking and learning in social context, In Barbara Rogoff and J. Lave (Eds.), Everyday Cognition: Its Development in Social Context, Harvard University Press, 1-8.

Shocker, Allan D., Moshe Ben-Akiva, Bruno Boccara, and Prakash Nedungadi (1991), Consideration Set Influences on Consumer Decision Making and Choice: Issues, Models, and Suggestions, Marketing Letters, 2:3, 181-197.

Snyder, Mark(1981), "On the Influence of Individuals on Situations," In N.Cantor and J.F. Kihlstrom (eds.), Personality, Cognition, and Social Interaction, Hillsdale, N.J.: L. Erlbaum, 309-329.

Srivastava, Rajendra K., Mark I. Alpert, and Allan D. Shocker (1984), "A customer oriented approach for determining market structures," Journal of Marketing, 48, 32-45.

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