Exploring the Relationships Between Means-End Knowledge and Involvement

Michael S. Mulvey, Pennsylvania State University
Jerry C. Olson, Pennsylvania State University
Richard L. Celsi, California State University, Long Beach
Beth A. Walker, Arizona State University
ABSTRACT - This research investigates the cognitive basis for involvement. We propose that the means-end knowledge about a product that is activated in a situation creates perceptions of personal relevance or feelings of involvement. We demonstrate the feasibility of this approach by describing the means-end knowledge of consumers at three levels of involvement.
[ to cite ]:
Michael S. Mulvey, Jerry C. Olson, Richard L. Celsi, and Beth A. Walker (1994) ,"Exploring the Relationships Between Means-End Knowledge and Involvement", in NA - Advances in Consumer Research Volume 21, eds. Chris T. Allen and Deborah Roedder John, Provo, UT : Association for Consumer Research, Pages: 51-57.

Advances in Consumer Research Volume 21, 1994      Pages 51-57

EXPLORING THE RELATIONSHIPS BETWEEN MEANS-END KNOWLEDGE AND INVOLVEMENT

Michael S. Mulvey, Pennsylvania State University

Jerry C. Olson, Pennsylvania State University

Richard L. Celsi, California State University, Long Beach

Beth A. Walker, Arizona State University

ABSTRACT -

This research investigates the cognitive basis for involvement. We propose that the means-end knowledge about a product that is activated in a situation creates perceptions of personal relevance or feelings of involvement. We demonstrate the feasibility of this approach by describing the means-end knowledge of consumers at three levels of involvement.

Why do some consumers seem to care more about some products and brands and not others? Why are some consumers highly motivated to seek information about certain products, or to buy and use these products in particular situations? These questions concern consumers' involvement, a key concept for understanding consumer behavior.

Involvement refers to consumers' perceptions of personal relevance for an object, activity, or event (cf. Bloch and Richins 1983). To date, however, most research on consumer involvement is at too high a level of abstraction to explain why people perceive a product to have personal relevance. In the typical measure of product involvement, such as Zaichkowsky's (1985) Personal Involvement Inventory (PII), consumers give direct ratings of their interest in a product, or their perceived importance of a product. Such measures indicate the overall level of consumers' overall involvement with the product, but they do not indicate the cognitive basis of that personal relevance. This would require measures of consumers' knowledge (beliefs) concerning how and why a product is personally relevant.

Celsi and Olson (1988) suggest that the perceived personal relevance of a product is determined by the activated cognitive structure of means-end associations that link people's knowledge about product attributes and benefits with their self-knowledge about important needs, goals, and values. In this paper, we describe an initial attempt to measure the cognitive basis for involvement by measuring the means-end knowledge structures that represent consumers' perceptions of personal relevance.

MEANS-END THEORY

A means-end chain is a simple knowledge structure that links product attributes to the consequences produced by these attributes (Gutman 1982). Olson and Reynolds (1983) described a means-end chain with six levels of attributes, consequences, and values, ordered from less to more abstract (see figure 1).

For example, a couple may decide to deposit money at a small, local bank because it is nearby (concrete attribute) and provides friendly, personal service (functional consequence). Personalized service is seen as important because the couple wants to establish a trusting relationship with the company responsible for managing their money (psychosocial consequence), in order to satisfy their need for security (terminal value). A means-end chain is the cognitive representation of the connection between a person's knowledge about a product (salient attributes and benefits) and their self knowledge (important psychological and social consequences and values). Thus, the end consequences of some means-end chains can be quite abstractCe.g., a person's life goals and personal values.

The means-end approach is based on the assumption that consumers see products as means to important ends. This means that the personal consequences produced by a product are more important (more self-relevant) than the characteristics of the product itself. A product is self-relevant to the extent a consumer sees it as instrumental in achieving important consequences or values. When means-end knowledge is activated from memory or formed in a situation, the person perceives the product to be personally relevant and feels involved with it (Celsi and Olson 1988). Thus, means-end knowledge structures are the cognitive basis for involvement. They also account for how and why consumers feel involved with a product in a particular situation.

Since researchers have not measured the product-related knowledge that underlies involvement, we conducted a simple study to demonstrate the usefulness of means-end theory in representing personal relevance. In particular, we wanted to explore two broad issues about the content and structure of consumers' product knowledge:

Content. How does the means-end knowledge about a product differ between more and less involved consumers? Are there systematic differences in knowledge about attributes, consequences, or values?

Structure. Do more involved consumers have a more complex network of knowledge compared to less involved consumers? Do the interrelationships differ at the attribute, consequence, or value levels?

METHODOLOGY

To examine these issues, we identified three groups of consumers with very different levels of involvement (low, moderate, and high) and compared their means-end knowledge structures about tennis rackets. We measured subjects' involvement with the activity of playing tennis because most people's self-relevance is focused on playing tennis, not on tennis products. We measured subjects' product knowledge about tennis rackets because this is the most critical product for playing the sport.

Subjects were 58 people from a University community, selected on a convenience basis. This group consisted of 23 women and 35 men, ranging in age from 18 to 51C90% were between the ages of 20 and 22. All but four of the subjects were college students.

We measured subjects' involvement with playing tennis using Zaichkowsky's (1985) Personal Involvement Inventory (PII scores can range from 20-140). Based on their PII scores, we assigned subjects to three, nonoverlapping involvement groupsCLow (PII=20-60; n=17), Medium (PII=70-110; n=21), and High Involvement (120-140; n=19). Most people in the low involvement group were novices who played tennis only infrequently and were uninterested in the sport. People in the medium involvement group played tennis occasionally to frequently and were moderately interested in the sport. People in the high involvement group played competitively on a daily basis and were highly committed to the sport (in fact, several were on the university tennis team). Thus, the three involvement groups represented a wide range of involvement from quite low to extremely high. Three to four weeks after completing the PII, we assembled subjects in small groups of 5 to 10 to measure their means-end knowledge about tennis rackets.

FIGURE 1

TABLE 1

LISTING OF MEANS-END CONCEPTS

We used a self-administered questionnaire developed by Walker and Olson (1991) to elicit people's means-end knowledge. Essentially, this paper-and-pencil procedure allows subjects to administer the laddering "interview" to themselves. First, subjects wrote down the product features they thought people in general consider when purchasing a tennis racket. Next, they were asked to identify the characteristics that they, personally, would consider when buying a tennis racket for their own use. Of these, they checked the four most important choice criteria in their purchase decision.

Next, subjects completed a self-administered laddering "interview" for each of the four choice criteria. The procedure began by writing the reason the first choice criteria was important to them (often this was a functional consequence of some sort). Then they wrote why that answer was important (what does this get you?). This usually produced a higher-ordered consequence. This pattern of writing why each answer was important continued until subjects could go no further (they reached the end of the means-end chain). This self-administered laddering procedure produced four means-end chains for each subject consisting of two to five levels each.

ANALYSIS

We used a computer program called "Laddermap" (Gengler and Reynolds 1992) to help analyze the laddering responses. The analysis began with a content analysis to reduce subjects' idiosyncratic responses to a common set of meanings. Two independent judges conducted the content analysis, and disagreements were resolved in discussions with a third judge. We identified 26 concepts that summarized all the attributes, consequences, and values mentioned in the laddering responses. Table 1 lists the final means-end concepts, along with examples of the verbatim responses that make up each concept.

Next, we examined the connections between the attribute, consequence, and value concepts. These linkages are often called implications. An implication is the perception of a causal or instrumental relationship between two concepts (a red car implies a fast acceleration). A means-end chain is merely a sequence of causal implicationsCan attribute implies a consequence which implies another consequence which implies a value. Two types of relationships between concepts are possibleCdirect and indirect implications. For example, a means-end chain A¦B¦C contains two direct implications (A¦B and B¦C), each of which was explicitly stated by a consumer in the laddering interview. In addition, the means-end chain includes an indirect implication between A¦C, which was not actually mentioned by a consumer, but is implied by the two direct associations. We combined both types of implications to produce a 26 x 26 implications matrix for each involvement group (available from the first author). The implications matrix accounts for all the direct and indirect relationships between a set of means-end concepts, and thus accounts for the aggregate cognitive structure of the group.

Although some subjects mentioned the same concept or implication more than once, we counted each unique concept and association mentioned by a subject only once. This avoided overweighting associations mentioned by verbose respondents who might mention an association between two concepts several times in the four ladders. Thus, each unique link in a means-end chain received equal weight in the implication matrix.

The aggregate implications matrix for each involvement group was converted to a cognitive structure map by the Laddermap program. The program produces a graphic representation of the key attribute, consequence, and value concepts and the associations between them. Figures 2, 3, and 4 present the cognitive structure maps for the low, medium and high involvement groups, respectively. To simplify the maps, we present only those connections mentioned by at least two consumers.

The format for the cognitive structure maps was suggested by Gengler, Klenosky and Mulvey (1992). Each means-end concept is represented by a circle. White circles represent attributes, grey circles represent consequences, and black circles represent values. The area of the circles are proportional to the percentage of subjects who mentioned the concept. The thickness of the line segments is proportional to the percentage of subjects who linked the concepts (both direct plus indirect associations).

A good cognitive structure map should account for most of the concepts and associations elicited by the laddering procedure and should be easy to interpret (a minimum of crossing lines). The cognitive structure map for low involvement consumers accounted for 70% of all direct and indirect relations, while the maps for medium and high involvement consumers account for 82% and 80% of the relationships, respectively.

RESULTS AND DISCUSSION

The purpose of this research was to explore differences in means-end knowledge structures for consumers with different levels of involvement. We examined both the content and structure of means-end knowledge. Content refers to the specific means-end concepts mentioned by subjects, while structure refers to the interrelationships between concepts. Both are necessary to understand the cognitive structure basis for involvement.

Content of Means-End Knowledge

First, we examined whether the total number of unique concepts mentioned across the four ladders varied with level of involvement. An analysis of variance revealed a significant main effect of involvement (Hi=9.8, Med=8.1, Lo=8.0; F[2,55]=3.61, p<.05). A Scheffe test revealed significant differences between high involvement and both low and medium involvement, but no differences between low and medium levels of involvement. This result suggests that high involvement consumers activate more product-related knowledge during decision-making than do medium and low involvement consumers.

This difference in knowledge content was reflected in the average length of the means-end chains. High involvement subjects had somewhat longer chains on average than did low and medium involvement subjects (Hi=3.4, Med=2.8, Lo=2.8; F[2,55]=8.18, p<.01). This suggests that high involvement subjects have better defined (more complex or more complete) means-end chains than less involved consumers do.

Next, we looked for differences in the types of knowledge concepts used to make decisions about tennis racquets. We found no overall differences in the average number of attributes mentioned as choice criteria by subjects. Neither were there any differences in the average number of values linked to those product attributes. Perhaps because tennis racquets have a limited number of observable attributes, tennis players, regardless of their involvement, "know" use about the same amount of attribute information. Likewise, perhaps there is a limited number of values which one can satisfy in playing tennis, whether one is an involved or casual player.

Consumers did mention differing numbers of intermediate level, consequence concepts in their laddering "interviews." High involvement subjects mentioned more unique consequences than the medium and low involvement consumers who did not differ (Hi=4.1, Med=2.8, Lo=3.1; F[2,55]=4.98, p<.01). This suggests that highly involved subjects may have a better understanding of how specific product attributes influence their playing (functional consequences), and how these consequences will help them achieve their end goals (values). Essentially, the product knowledge structure for the highly involved subjects (see Figure 4) is more "interconnected" and "complex" as reflected in the greater number of linked consequences and the longer means-end chains. The more complex means-end chains of highly involved consumers may be due to their greater "product familiarity," as they play more frequently and thereby learn which product attributes provide personally relevant consequences.

Finally, we examined whether specific means-end concepts at the attribute, consequence, and value levels were mentioned more often by a specific involvement group. Logistic regression was used to model the probability that a person would mention a certain concept, given their level of involvement. Model fit was evaluated using the -2 Log L statistic. Differences between the levels of involvement were evaluated using the Maximum Likelihood Estimates (see SAS 1992).

As would be expected, we observed more differences between involvement levels at the consequence level. For instance, quality/durability was much more likely to be mentioned by low involvement consumers than high involvement consumers (82% versus 42%; p<.05). On the other hand, high involvement consumers were more likely to mention comfort in their means-end chains than moderate and lower involvement consumers (53% versus 18% and 23%; p<.05). High involvement consumers also mentioned feel more often than less involved consumers (58% versus 29% and 32%) and they mentioned concentrate more often (26% versus 6% and 5%; p<.15 in both cases). The attribute racket material was mentioned more often by highly involved consumers than less involved consumers (p<.05). Head size and brand name also differed in levels of mention (p<.15). Finally, at the value level, play my best was mentioned more often by highly involved consumers (52.6%) than low involved consumers (17.7%; p<.15).

FIGURE 2

HIERARCHICAL VALUE MAP - LOW INVOLVEMENT

Structure of Means-End Knowledge

Next, we compared the structural linkages between means-end concepts in the cognitive maps for the three involvement groups (see Figures 2, 3 and 4). The overall knowledge structures for the three involvement groups are similar in many respects, but there are interesting differences in meaning at the level of individual means-end chains. For example, low involvement consumers seem to have a simple means-end hierarchy: they use brand name as the main cue to infer racket quality and durability. High involvement consumers, in contrast, are less likely to use brand name as a cue for quality. This does not necessarily mean that quality is less important to highly involved consumers. Rather, unlike low involvement consumers who think about "quality" as a global, abstract attribute, highly involved consumers understand and articulate the specific meaning of what constitutes a high quality tennis racket. Thus, high involvement consumers consider the head size, grip and the material when evaluating a racket. Compared to their lower involvement counterparts, high involvement consumers link these attributes to the benefits of comfort, racket feel and ability to concentrate. These factors in turn, are seen as instrumental in achieving their key goal of playing their-best.

FIGURE 3

HIERARCHICAL VALUE MAP - MEDIUM INVOLVEMENT

FIGURE 4

HIERARCHICAL VALUE MAP - HIGH INVOLVEMENT

The content and structure of the other means-end chains vary across levels of involvement, reflecting different levels and types of expertise. For example, low involvement consumers have a simple means-end chain associated with the attribute, low weight. They view low weight as a means of playing better, so they can satisfy their end goal of fun & excitement. In contrast, high involvement consumers have a more complex chain of meaning associated with racket weight. First, low weight is seen as a means of making play more comfortable, which produces better control, which allows them to play better so they can win, the end goal or value. A second means-end chain linked weight to avoiding injury, which allows the person to play more and to experience fun & excitement. Although these same concepts also appear on the low involvement cognitive map, they are not connected to product attributes; thus the meaning is not the same. The same pattern occurs with other consequences such as confidence, feel and comfort.

As mentioned earlier, the thickness of the lines on the map represents the number of subjects who associated the linked concepts. Not only do high involvement consumers have more relationships between salient concepts, but these relationships are stronger. This phenomenon is particularly evident in the relationships between play better and the values of winning, play my best, and fun & excitement. The greater number of lines represented on the maps of high involvement consumers reflects the greater complexity of their means-end knowledge structures compared to lower involvement consumers.

Although this cross-sectional study did not explore the development of means-end chains over time, the results are suggestive. Apparently, the knowledge structures of highly involved consumers have evolved to the point where most concepts are interrelated. The rather subtle implications between many product attributes and functional consequences (feel, control, comfort) are appreciated by involved experts. In addition, end goals and values tend to be interconnected for the high involvement consumer. We could speculate that highly involved consumers form a "unit" of self-relevant goals and values associated with playing tennis such as winning, play my best, and fun & excitement. Perhaps the entire means-end structure relating product-knowledge to these personally-relevant consequences is activated for use during decision-making.

Implications for Future Research

The objective of this paper was to explore means-end knowledge structures associated with different levels of involvement. Of course, the specific results have limited generalizability due to the subject group, the sample size, and the particular product and activity we examined. Our goal is to generate interest in studying involvement from a means-end perspective and stimulate future research aimed at understanding the cognitive basis for involvement.

The means-end knowledge structures provide much richer data about personal relevance or involvement than do broad measures such as PII, which fail to provide insight into the underlying reasons for involvement. It is possible for consumers to have equal levels of involvement (measured by the PII), but perceive the self-relevance of the product in completely different ways. Knowledge of the basis for perceived self-relevance is useful for many marketing applications. For instance, advertisements that communicate an entire chain of meaning, rather than isolated facts and concepts are likely to be more effective (Gengler 1990, Reynolds and Rochon 1990, Young and Feigin 1976).

The present study is cross-sectional. Further research is required to investigate how personal relevance (the relationship between consumers and products) develops over time. We need to understand the evolution of involvement in terms of how product and self knowledge become linked with experience. This will require longitudinal designs and individual-level analyses. Rummelhart and Norman's (1978) work on cognitive learning could be useful in designing such a study.

Finally, researchers should investigate how consumers use their means-end knowledge in tasks such as making decisions. Do they use individual concepts in a means-end chain as choice criteria, or do they treat the entire means-end chain as a single "unit" of meaning? Hayes-Roth (1977) suggested that complex knowledge structures may become "unitized," so that an entire means-end network of meaning could be represented by a "short-hand" representation of the meaning. If so, abstract concepts such as quality, play better, and value may be used as umbrella concepts that "stand for" the whole means-end chain of related concepts. The challenge to researchers is to describe the unitization process, uncover the conditions under which unitization occurs, and develop methods of decomposing the unitized meaning. Once this is done, we will have the means to develop a better understanding of how general concepts as quality and value are used by consumers in decision making.

REFERENCES

Bloch, Peter H. and Marsha L. Richens (1983). "A Theoretical Model of the Study of Product Importance Perceptions," Journal of Marketing, 47(Summer), 69-81.

Celsi, Richard L. and Jerry C. Olson (1988). "The Role of Involvement in Attention and Comprehension Processes," Journal of Consumer Research, 15(2), 210-224.

Gengler, Charles E. (1990). An Architectural Perspective on Advertising Strategy, unpublished doctoral dissertation, University of Texas-Dallas.

Gengler, Charles E., David B. Klenosky and Michael S. Mulvey (1992). "A Note on the Representation of Means-End Study Results," working paper, Rutgers University, Camden, NJ, 08102.

Gengler, Charles E. and Thomas J. Reynolds (1992). Laddermap: Software Tool for Analyzing Means-End Data, Version 3.11.

Gutman, Jonathan (1982), "A Means-End Chain Model Based on Consumer Categorization Processes," Journal of Marketing, 46(2), 60-72.

Hayes-Roth, Barbara (1977). "Evolution of Cognitive Structures and Processes," Psychological Review, 84(3), 260-278.

Olson, Jerry C. and Thomas J. Reynolds (1983). "Understanding Consumers' Cognitive Structures: Implications for Advertising Strategy," in Advertising and Consumer Psychology, eds. Larry Percy and Arch Woodside, Lexington, MA: Lexington Books.

Reynolds, Thomas J. and Jonathan Gutman (1988). "Laddering Theory, Method, Analyses and Interpretation," Journal of Advertising Research, 28(1), 11-31.

Reynolds, Thomas J. and John P. Rochon (1990). "Means-End Based Advertising Strategy: Copy Testing Is Not Strategy Assessment," Journal of Business Research, 22(2), 131-142.

Rummelhart, David E. and Donald A. Norman (1978). "Accretion, Tuning and Restructuring: Three Modes of Learning," in Semantic Factors in Cognition, ed, J.W. Cotton and R.L. Klatsky, Hillsdale, NJ: Lawrence Erlbaum, 37-53.

SAS (1992). SAS/STAT User's Guide, Volume 2, Version 6, 4th Edition, Cary, NC: The SAS Institute.

Walker, Beth A. and Jerry C. Olson (1991). "Means-End Chains: Connecting Products With Self," Journal of Business Research, 22(2), 111-118.

Young, Shirley & Barbara Feigin (1975). "Using the Benefit Chain for Improved Strategy Formulation," Journal of Marketing, 39(July), 72-74.

Zaichkowsky, Judith Lynne (1985). "Measuring the Involvement Construct," Journal of Consumer Research, 12(3), 341-352.

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