by

Scott M. Smith

and

William R. Swinyard

Internet Text January 1999

May Not Reproduced without the Permission of the Authors

Cougar Visa: Developing a Means-End Chain for a Credit Card

WHAT IS A MEANS-END CHAIN?
Excerpts from "Applying Laddering Data to Communications Strategy and Advertising Practice" by Tom Reynolds and David Whitlark. Journal of, Advertising Research - July/August 1995

Means-End Mapping of Consumer Decision-Making
One way to understand a means-end chain is to think of consumer decision making as a problem solving process. In making decisions consumers select a course of action or means to reach an objective or end. While a means can be an end, an end can also be a means. Using the map shown below as an example, "on-time delivery," an end, is obtained through the "reliability" provided by an express mail delivery service, a means. However, "on-time delivery" itself is a means to reach other ends, such as "less worry about on-the-job unknowns," "feeling more personal control," and "peace of mind."

A means-end map arranges means and ends into a network of attributes, physical and emotional consequences, and personal values or life goals. The means-end framework for viewing consumer decision making leads to a means-end theory of communications strategy. Generally speaking, the theory posits that communications are the most personally relevant and compelling when they make a strong link between the right set of attributes, consequences, and values.

Means-End Chains and Laddering
Laddering refers to an interviewing technique that can be used to elicit means-end connections and attribute-consequence-value networks people use when making decisions about what brand to buy, what store to shop at, what issue to support, or even who to vote for.

Consider, for example, a consumer that says an important distinction between express delivery services is that one service has "package tracking software" and another service does not. The researcher would then ladder the key distinction by asking what the respondent receives by having package tracking software" (means) i.e., why is it important to the respondent. Then the respondent is asked about the consequence "makes me look good" (end) i.e., why it is important to the respondent and so forth until the attribute-consequence-value chain is exhausted.

The laddering interview reveals the linkages between attributes, consequences, and values used by respondents to justify their beliefs and/or behavior.

LEARNING ABOUT LADDERING
Excerpts from "Advancements in Laddering" by Tom Reynolds, Clay Dethloff, and Steve Westberg.

The Laddering Interview

A laddering interview is an in-depth, one-on-one structured dialog that draws out the connections people make between product attributes, the consequences of those attributes, and the human values linked with those consequences (means-end chain). Interviews last between forty-five minutes to two hours and generally recorded so that interviewers do not lose the detail of what is said.

Getting respondents to reveal their true beliefs, feelings, and goals requires warm-up questions to put the respondent at ease with the interviewer and to help them start thinking about the product of interest. Laddering differs from other types of qualitative research in that it aims to uncover personally motivating reasons behind brand choice and link them to product attributes and their consequences.

Laddering begins by identifying the most important distinguishing characteristics of the brand for a given usage situation and then moving up and down the means-end chain to get a complete picture of attribute-consequence-value identities and linkages. Moving up and down the means-end chain is done by asking a form of the question: why is that important to you?

Attitudes toward a product do not often predict choice behavior, but attitudes toward choice behavior concerning a product typically will. Laddering asks questions regarding the reasons people have for making the choices they do. It requires respondents to justify their buying behavior by explaining the distinctions they make between choice alternatives.

Eliciting Product Distinctions
Laddering research is designed to uncover the reasons underlying purchase decisions. Both the positive reasons for choosing a brand and the negative reasons for rejecting a brand should be determined. Simply asking what is good about a brand or product is not enough. The marketer needs to understand how a brand is believed to be better than others, or conversely how a brand is believed to be worse than others.

There are several methods for eliciting distinctions between brands. These methods are (1) top-of-mind imaging, (2) grouping similar brands, (3) contextual environment, (4) preference, usage, and preference-usage differences, (5) timing of purchase or consumption, (6) usage trends, (7) product or brand substitution, and (8) alternative usage occasions.

Top-of-mind imaging. The respondent is asked to give one or more first-thought associations for each of several brands or product types. Polarity (positive or negative) for each association is also determined. Then, the respondent is asked why the characteristic is -a positive or -a negative and the responses are further probed to uncover the ladder. Top-of-mind imaging identifies the most conspicuous characteristics of a brand, but not always the characteristics that differentiate it from a close competitor.

Grouping similar brands. This method uncovers the way respondents group products together and the reasons they use for forming product groups. Respondents are asked to group brands and/or products into like categories. Then the primary reason for group membership, either a positive or negative characteristic, can be elicited and laddered. Additionally, the respondent can be asked to identify the brand or product that best represents the group. Important traits and trait performance for the most representative brand can be identified and laddered as well.

Contextual environment. the contextual environment includes predetermined physical or need-state occasions of brand purchase or use. Physical occasions are generally described by time, place, and people when usage occurs. A need-state occasion is a mental need or inner desire that can span many physical occasions. For example, need-states include occasions such as relaxing, rejuvenating, building relationships, feeling powerful, reducing stress, and getting organized. The method asks respondents to associate a product or brand with a context such as "those times when you want to relax," or "after you have just completed a tough job or accomplished something that is important to you."

Preference, usage-e. and preference-usage differences. Comparing brand preference and brand usage is one of the most direct and commonly used methods for eliciting brand distinctions. Brands can be ranked with respect to (1) preference and (2) frequency of use. Then, brands can be directly compared against each other based on these rankings using such questions as, "why did you rank Brand A higher than Brand X," and/or "why do you use Brand B more often than Brand A." Also, the interviewer may ask why a brand ranks lower on preference than on usage. Often, respondents use price as the key reason for ranking one brand over another. Avoid the problem by selecting brands to compare that are price competitive.

Timing of purchase or consumption. Timing issues can influence product choice and usage. For example, a respondent might be asked to break a sickness such as the common cold into several stages like onset, full-blown, and on-the-mend. Then the respondent would relate which brands were preferred for each time-related stage. It is also common for consumers to use one brand of product during the day (Coca-Cola) and a different brand in the evening (sprite).

Usage trends. Respondents are asked to quantify their beliefs about past and future usage of a brand. For example, an interviewer might ask, "over the next five years, do you expect to use this brand more often, less often, or about the same as you have in the past?" Then, reasons for increased, decreased, or unchanged usage are elicited.

Product or brand substitution. Distinctions between brands can be directly assessed based on the ability of one brand to be substituted for another. The brand to be substituted can be a brand currently used by the respondent if a goal of the research is to increase use (identify what-attribute or consequence needs to be added or removed) or it can be a brand not currently used if a goal is to increase trial (identify what attributes or consequences need to be promoted). For an unfamiliar brand, the respondent first can sample or be given a description of the brand. Follow-on questions might include, "how likely would you be to substitute this new brand for your current brand for this occasion -- why is that?"
Alternative usage occasions. Another method is to alter or add new usage occasions for the respondent to consider. Alternative occasions can be either predetermined or provided by the respondent. For example, one might ask, "think of a new situation or occasion in which you might use Brand A, but that you currently don't -- why would you consider using Brand A for this occasion -- what is keeping you from using Brand A for this occasion now?" Both positive reasons why a brand fits a new occasion and negative reasons why it does not fit can be elicited and laddered.

In practice, several different elicitation techniques are used in a laddering study to capture a full range of meaningful distinctions between a brand and its competitors. For example, a laddering interview might start with "top-of-mind imaging" to understand general product-category beliefs, then increase in brand-related specificity through using "contextual environment" and "alternative usage occasions."

Getting the Ladders
Once distinctions are elicited, they must be laddered to move the respondent up and down the means-end chain to uncover salient attributes, consequences, and values together with their linkages.

Salient attributes are uncovered by asking questions such as, "what is it about the brand that makes it that way," "what is it about the brand that gives you that benefit," "how can the brand deliver that benefit," "what is the brand missing to give it that defect." Attributes are linked to higher level constructs (consequences and values) by asking questions such as "why is that important to you," "how does that help you out," "what do you get from that," "why do you want that," and "what happens to you as a result of that." Higher level psycho social consequences and values are most often feelings or personal beliefs, so asking "how does that make you feel" is an appropriate question.

Negative ladders begin, of course, in negative terms. At the consequence levels, the interviewer may want to ask, "why do you want to avoid that." This effectively turns the discussion from negative to positive. Most respondents are better able to discuss feelings about obtaining a value rather than avoiding one, so laddering is facilitated by talking in positive terms before reaching the value level. More examples of negative questioning are, "why is that negative to you...... how does that interfere with what you are doing," "what's wrong with that."

Often the respondent cannot answer a question or will not move to a higher level on the means-end chain. In the circumstances, one can try one of the following techniques to help the respondent move on.

Reiteration of occasion. The interviewer can remind the respondent of the occasion basis for the ladder when the respondent appears to have forgotten or lost track. It may help to have the respondent provide further information about the occasion. For example, "you were with whom, doing what, where, etc."

Alternate scenario. The interviewer can ask the respondent to think of another situation or scenario similar to the one currently being discussed in which the brand is used in a similar way for similar reasons.

Absence of product. The interviewer can ask the respondent for his or her feelings, responses, and the potential consequences if the brand were unavailable for the occasion.

Abstraction form product. Occasionally, respondents will not be able to leave the brand at the attribute level and will wonder how the brand itself can "make me feel good about myself" or can "improve my relationship with my spouse." The interviewer can ask the respondent to ignore the brand and only consider the last consequence that was mentioned.

Negative laddering. Negative laddering seeks the respondent's reasons why they do not want to do certain things or feet certain ways. The interviewer can ask the respondent what would happen if they were not able to achieve a certain positive consequence.

Age regression contrast. The age regression contrast forces the respondent to compare usage or consumption in a previous time period with now. For example, the interviewer may ask the respondent if he or she used the product five years ago then why or why not. This technique is similar to the "usage trend" method of eliciting distinctions, but is used during the actual ladder to overcome a mental block.

Third person probe. This method places the respondent in another person's shoes. The interviewer asks the respondent how others might feel in similar circumstances. The approach is useful when the respondent feels threatened or uncomfortable discussing their personal reasons underlying their behavior.

Silence. Silence and patient attention will signal to the respondent that the interviewer is waiting for a more detailed response. The respondent often will elaborate on a vague or incomplete answer.

Reiteration of "a-c-v" means-end chain. To help the respondent maintain a complete train of thought during the ladder, the interviewer can reiterate the answers given up to the point of the mental block. The complete ladder should be repeated back to the respondent after a value has been reached allowing the respondent the opportunity to verify his chain of thought. In addition, the interviewer can use the technique to refocus a rambling respondent.

Laddering Pitfalls In the course of a laddering interview, the respondent will not always provide responses that are whole or complete. The interviewer must be able to identify incomplete responses so that additional probes can draw out useful information.

Generic statements. Respondents often provide generic answers that have no specific meaning. For example, "satisfied" can be either physical (feeling full after a meal) or psychological (feeling content with oneself. Likewise, "happy" can have multiple meanings, including feeling happy about something accomplished and feeling happy for another person, and it can have varying intensity. Slang words like "cool" and "bad" in particular need to be clarified because they can hold different meanings for different people. Often, these situations can be resolved by simple asking, "what do you mean" or "could you describe that feeling."

Not brand specific. Differentiating characteristics should be brand specific and unambiguous. Distinctions that apply to many brands equally well or even to the entire category are not useful.

Multiple responses. Respondents may give more than one answer when providing distinctions or during laddering probes. In these cases, the interviewer must ask which characteristic or idea is most important for the given situation and then continue probing from there. It is possible to ladder multiple "branches" although this can confound the analysis.

Chutes and ladders. Distinctions are most often product attributes, but respondents may sometimes mention an upper level element as a basis for differentiating one brand from another. The interviewer can "chute down" by asking, "what is it about the brand that makes it that way?" Occasionally, a respondent might ladder directly from an attribute to a value or appear to leave an important element out. Again, the interviewer can ask, "I'm not sure how (lower level element) leads to (upper level element) -- is there something about the brand that makes you feel that way?"

Habit. Respondents tend to say "it's a habit" or "I've always done it that way" when they cannot think of a more rational reason for their usage or consumption behavior. The interviewer should try to uncover when and how the habit started, and what brand they would substitute if they could no longer get their favorite brand, then ladder the resulting distinctions.

I like it. Although similar to a generic statement, this phrase occurs frequently in laddering and can almost always be handled the same way. For example, the interviewer -can ask,-"could -you describe that feeling for me" or "what is it about the brand that you like."

This review has covered the most common practices, difficulties, and remedies associated with laddering interviews. A final rule of thumb for interviewers is to ask themselves, "do I understand all of the personal reasons why the respondent chooses and uses the brand?"

The following comments are found in the article: "A Motivational Perspective on Means-End Chains " by Joel B. Cohen, University of Florida and Luk Warlop ,Catholic University of Leuven http://www.econ.kuleuven.ac.be/tew/academic/market/members/member/courses\meansend.htm

The underlying assumption is that by making these terminal values more salient (e.g., through advertising) greater importance will be given to designated product characteristics and resulting consequences, producing the desired effect on behavior. While this is a theoretically viable approach, its success hinges on the importance of that higher order value in the consumer's choice among competing alternatives. Thus, not only must the value be a potent driving force, but distinguishing characteristics of the product and/or the consequences of owning and using it must be seen as consistent with, or furthering, the valued state.

This would appear to be a tall order for most products! "Just" determining which goals and values are especially potent and/or salient in a person's life is itself a challenging endeavor. This often involves consideration of perceived discrepancies between ideal/desired states and an assessment of one's current situation, since higher order values that are judged to be important in the abstract are not necessarily active, driving forces on a daily basis. Cantor, Marcus, Niedenthal, and Nurius (1986) hypothesize that a "working self-concept" (i.e., the subset of self-knowledge that is made salient and dominant by contextual factors) -- rather than a few generally important self-related values -- is likely to guide behavior. Context-recruited self-perceptions (i.e., the working self-concept) help identify intermediate goal regions and influence the means chosen to attain them. Walker and Olson (1991) recently expressed a similar view; that "central aspects of self" should be related to behavior only when the situation activates these aspects, and that the "particular values that influence behavior may be completely different in different situations."

Inherent Limitations of Laddering

One reason for means-end chain advocates' enthusiasm about the use of the approach to represent cognitive structures seems to be the face validity of a relatively unstructured elicitation task rather than the use of pre-specified cognitive categories. However, laddering is far from neutral in the types of responses it elicits (i.e., reasons underlying preferences). Laddering systematically probes for successively higher level goals and values, and therefore cannot be said to reflect how consumers think about products or brands. It is quite likely, for example, that there are many important associations at a given level in the hierarchy -- particularly at the attribute/benefit level -- that are ignored because of this hierarchical emphasis. In addition, in striving for useful and representative means-end chains, individuals' responses are interpreted, coded, and aggregated, thus necessarily sacrificing a certain degree of accuracy for parsimony. While such procedures may be sound as a way of focusing on predominant motivational chains, different tradeoffs would be needed if the goal were, in fact, to map consumers' cognitive structures.

A second major issue in the use of laddering to discover how consumers think about products is the leading nature of the procedure. Participants are literally "pushed up" an attribute- consequence-goal-value hierarchy in an effort to discover which of these seem to be linked hierarchically. Whether such a hierarchical arrangement exists (in any form) in the consumer's mind is not investigated; it is assumed.

Aside from respondents answering that there was no particular reason why they desired a certain attribute or felt that a certain outcome was beneficial -- and risking looking rather foolish -- laddering will produce reasons for preferences. That is what the technique is designed to do. Unfortunately, it has characteristic features of a problem-solving exercise. In a sense, riddles are posed and solved, proceeding from the shared belief that preferences must have reasons. One can only imagine the thought processes that respondents go through when they realize they have never thought about why a certain outcome is desirable! However, there is ample reason to believe that they will search for a plausible explanation, and one that speaks well of themselves (Schlenker and Weigold 1992).

Consumer Decision-Making Map: Express Mail Delivery

CHAID, Chi-square Automatic Interaction Detection, is a technique that groups data to maximize differences between segments. For example purposes, assume we identify two segments of the market, say those who were heavy purchasers and those who were light purchasers. We desire to determine how they are described demographically. CHAID would take the heavy purchasers and break them into two or more sub-groups based on the single demographic variable that best (most statistically significant) predict heavy usage. Next, each of the newly created sub-groups are divided into two or more groups based on the best predictor variable that hasn’t previously been previously used as splitting criteria. This splitting process continues until no statistically significant predictor variables remain. The process is halted once each of the sub-groups is as homogenous as possible. A decision tree diagram is then constructed, which displays the splits as well as the resulting sub-groups or segments. By tracing down the tree, we observe that we can describe the nature of the heavy users, with the variables most important in the description being at the top of the tree and the least important being at the bottom of the tree. CHAID provides an excellent model for managing segmentation problems where we have large databases consisting of many customers that have been surveyed. Large samples are important because as we continue down the tree, the sub-samples become smaller and smaller, eventually testing the limits of sample size for these analyses.