Jonathan Gutman

A Means-End
Chain Model
Based on To practitioner and researcher alike, consumer
values play an important role in understanding
behavior in the marketplace. This paper presents
Consumer a model linking perceived product attributes to
values.

Categorization
Processes

V ALUES have been shown to be a powerful force

in governing the behavior of individuals in all
aspects of their lives (Rokeach 1968, Yankelovich
to the achievement of desired ends (valued states). For
example, knowing that consumers want to look well
dressed doesn't tell us much unless we know why
1981). Their use in marketing planning and strategy they want to look that way (sexual attractiveness, ac-
could be improved if the behavior of consumers could complishment, neatness, etc., which are value-level
be related to their values. Several attempts have been considerations) and what attributes in clothing they
made to provide a theoretical and conceptual structure associate with being well dressed. For the means-end
connecting consumers' values to their behavior (How- chain to have a central place in marketing planning
ard 1977, Vinson, Scott and Lamont 1977, Young and and consumer research, the nature of the elements
Feigin 1975). These attempts can be subsumed under comprising means-end chains and the nature of the
the rubric of a means-end chain. Means are objects connections between these elements must be speci-
(products) or activities in which people engage (run- fied, and the model must be operationalized so as to
ning, reading). Ends are valued states of being such facilitate research.
as happiness, security, accomplishment. A means-end
chain is a model that seeks to explain how a product
or service selection facilitates the achievement of de- Conceptual Model for Means-End
sired end states. Such a model consists of elements Chain
that represent the major consumer processes that link
values to behavior. The purpose of this paper is to present the author's
means-end chain model. The model is based on two
The means-end chain concept offers marketing
fundamental assumptions about consumer behavior:
managers a way to position products by associating
(1) that values, defined here as desirable end-states
means (the physical aspects of products) with adver-
of existence, playa dominant role in guiding choice
tising that seeks to tie the consumption of products
patterns, and (2) that people cope with the tremendous
diversity of products that are potential satisfiers of
their values by grouping them into sets or classes so
Jonathan Gutman is Associate Professor, University of Southern Cali-
fornia. as to reduce the complexity of choice. This suggests
that in addition to the product-class type of product

Journal of Marketing
60 / Journal of Marketing, Spring 1982 Vol. 46 (Spring 1982), 60-72.
categories, e.g., toothpaste, mouthwash, chewing fore, the values-consequences linkage is one of the
gum, mints, consumers are capable of creating cate- critical linkages in the model. To the extent that val-
gories based on product functions. For instance, they ues are ordered in importance (Rokeach 1973), they
could produce a category of breath fresheners that in- also give consequences importance, in that conse-
cludes products from all of these product classes quences leading to important values should be more
("fresh breath" is unlikely to be classed as a value important to a person than those leading to less im-
per se, but it might lead to favorable interpersonal portant values (a position very similar to that espoused
interactions that could lead more directly to valued by Rosenberg 1956).
states). An act of consumption must occur in order for the
In addition to these two assumptions about con- desired consequences to be realized. Thus a choice
sumer behavior that are essential to the particular form among alternative products has to be made. In order
of the model, there are two other assumptions of a to make this choice, the consumer has to learn which
more general nature. These assumptions are that all products have attributes that will produce these de-
consumer actions have consequences (although all sired consequences. Therefore, the second important
consumers wouldn't agree that the same actions in the linkage in the model is that between consequences and
same situations produce the same consequences), and product attributes.
that consumers learn to associate particular conse-
quences with particular actions. Product-Use Situation
The last concept necessary to the explication of the
Consequences model at this point is that of the product-use situation.
Consequences may be defined as any result (physio- Consumers obviously encounter many potential prod-
logical or psychological) accruing directly or indi- uct-use situations. There are many ways of defining
rectly to the consumer (sooner or later) from his/her and describing such situational variables (Belk 1975,
behavior. Consequences can be desirable or undesir- Fennell 1978). But for the level of analysis referred
able. There is a term and a literature in marketing to in this context we may simply define a product-use
dealing with desirable consequences called benefits situation as any situation that involves the use of a
(Haley 1968, Myers 1976), which are the advantages commercially available product or service.
consumers enjoy from the consumption of products. Consumers have desired consequences they are
Benefits differ from attributes in that people receive trying to achieve. Consumption situations provide
benefits whereas products have attributes. "White them with an opportunity to achieve these conse-
teeth" is a benefit that can be obtained from brushing quences. Each consumer learns over time which
your teeth with anyone of a number of toothpastes choices in a given situation produce these desirable
or powders with "teeth whiteners." consequences and which do not. Consumers evaluate
Consequences may be physiological in nature (sat- product-use situations in terms of their potential im-
isfying hunger, thirst, or other physiological needs), pact over time. The length of the time horizon the
or the nature of the consequences resulting from one consumer adopts with respect to the situation can
behavior may be psychological (self-esteem, im- modify the importances of the consequences. Thus
proved outlook for the future) or sociological (en- someone who doesn't eat rich desserts on a regular
hanced status, group membership). Direct conse- basis for health or caloric reasons may nevertheless
quences come directly from the thing consumed or indulge in a chocolate mousse on a special occasion.
from the act of consumption. Indirect consequences The means-end chain permits us to focus on the basic
can occur when other people react favorably or un- aims consumers have in life while not losing sight of
favorably to us because of our consumption behavior. how these aims influence choices in specific situa-
They can also occur when direct consequences pro- tions.
duce other consequences. For instance, buying a new
dress makes you feel better (a direct consequence); Summary of Conceptual Model
because you feel better, people react more favorably To summarize, the model may be conceptualized as
to you (an indirect consequence). In addition, some shown in Figure 1. Consuiner values give conse-
consequences occur immediately during the act of quences valence and importance. These importances
consumption (having one's thirst quenched), whereas are modified by the situation, which induces the con-
others occur later (indigestion, tooth decay). sumer to consider the consequences in light of the
The central aspect of the model is that consumers demands of the situation. The relevant consequences
choose actions that produce desired consequences and coming out of this person-situation interaction are the
minimize undesired consequences. It is suggested, basis of a functional category of products that can best
following Rokeach (1973), that values provide con- produce the relevant consequences. These products
sequences with positive or negative valences. There- are selected on the basis of attributes they possess,

A Means-End Chain Model Based on Consumer Categorization Processes / 61

 initial grouping level based primarily on their physical
FIGURE 1 characteristics. The usage situation (regular lunch,
Conceptual Model for Means-End Chain occasional lunch, at a baseball game, etc.) would
modify the importances of the consequences, produc-
ing a set of relevant consequences for that consump-
tion situation. In addition, this set of relevant conse-
quences would be applied to the product class groupings
(Produce)
.. that comprise the categories at the initial grouping
level. At this point a final choice could be made, if
more than one choice remained.
+ (Results In)

DESIRED (Benefits) AND Categorization Processes

I
UNDESIRED CONSEQUENCES

Produce-use Si t.ua t i on
The categorization process is proposed as the method
of studying how consumers organize their thinking
1
cons e que nce s ) Perceptions Modify "'SITUATIONS
Serve as Salience of Consequences
(
Criteria for
(Produces)
about specific product alternatives. It is hypothesized
that consumers create arrays of products that will be
,...-- -, ....'!RE-LE-VAN-T-'C'---ON-SE-QUE-N-CE-'SI
instrumental in helping them achieve their desired
Identity Groupings Attributes Imply consequences, which in tum move consumers toward
of Products Based on
Product Attributes
Products' Ability
to Produce Consequences valued end states. If this connection can be made,
t (Produce)
marketing will be in a better position to understand
FUNCTIONAL GROUPINGS
OF PRODUCTS how personal values influence everyday consumer
I (Comparison
choices.
+ Processes)
ICHOICE I Categorization processes represent the way in
which consumers segment their environments into
meaningful groups by creating equivalences among
nonidentical stimuli. "The world consists of an infi-
which imply their ability to produce the desired con- nite number of potentially different stimuli. Thus a
sequences and avoid the undesired consequences. basic task of all organisms. . . is a segmentation of
Over time, consumers learn to distinguish between the environment into classifications by means of
products they wouldn't use and those they would use which nonidentical stimuli can be treated as equiva-
and in which types of situations they would use those lent" (Rosch 1978, p. 1). Categorizing is the way
products. To understand the linkages between the lev- consumers separate the world into smaller units to
els in the means-end chain, the differences as well as make it manageable, to limit it to the point where it
the similarities between the products in the functional matches their individual capacities.
categories have to be studied. The ways in which con- The marketplace is full of many more objects than
sumers describe the diversity of these functionally individuals have values. Therefore, "ends" are few,
equivalent products can tell us much about their and "means" are many (Vinson, Scott and Lamont
means-end chains. This approach also allows us to 1977, p. 46). It is essential for consumers to reduce
determine why certain products are not chosen or con- the complexity inherent in the multitude of alterna-
sidered and the consequences that would be produced tives with which they are faced. "When faced with
if they were chosen. a set of stimuli people often organize them in clusters
Consider, as an example, the purchase of a hot to reduce information load and facilitate further pro-
dog. What consequences are produced by the con- cessing" (Tversky and Gati 1978, p. 91). To enable
. sumption of a hot dog and how do these consequences a person to achieve his/her values, groups or cate-
relate to values? There are many potential conse- gories of products (product classes) have to be related
quences depending on who is doing the consuming: systematically to the higher-level ends if the chain is
taste experience, type of meat, nutrition obtained to serve its instrumental purpose.
from meat, preparation consideration, consumption of This occurs because although grouping is deter-
additives, salt, fat, and so forth. mined by the object's properties, the choice of prop-
Relevant values with respect to pleasure, living a erties to be focused on is influenced by values. Con-
comfortable life, religion, and good health, among sumers group products in different categories depending
others, could playa role in attaching valences and on which features they emphasize and which features
importances to these consequences. Different types of they ignore. In this manner, the ways in which prod-
hot dogs, luncheon meats or sandwich fillings- ucts are identified or described by consumers fit their
depending on the breadth of alternatives the consumer subsequent grouping of these products into functional
might wish to consider-would be categorized at an groupings (for example, one consumer might have a

62 / Journal of Marketing, Spring 1982

category of luncheon meats, another of pork products, Having presented a conceptual model for a means-
another of meat with and without additives). This end chain, other means-end approaches that have ap-
means that values are translated from their context at peared in the literature will be reviewed. It will be
the more inclusive levels of the chain to the less in- shown that existing approaches have fallen short of
clusive levels where products are categorized into meeting one or another of the criteria of specifying
product classes. Consumers are more likely to be in the links in the chain, of showing how they are con-
agreement on what physical characteristics products nected to one another, or of being sufficiently oper-
possess than at high levels of inclusiveness where cat- ationalized to guide research. In most cases the central
egories are based on the function of value-producing consideration in developing these models is not the
consequences. presentation of a fully developed means-end chain
This categorization process takes place at each model, either because the focus is broader (Howard
level of the means-end chain as categories of greater 1977), narrower (Vinson, Scott and Lamont 1977),
inclusiveness are formed at higher and higher levels or wholly pragmatic (Young and Feigen 1975).
of the chain (inclusiveness refers to the degree of sim-
ilarity among objects in a category). Brands of paper
towels grouped together would represent a category Review of Means-End Model
of low inclusiveness. If consumers distinguish be- Literature
tween plain paper towels and those with a pattern, two
categories of lower inclusiveness would be created. Values Orientations for Means-End Chains
If paper towels were grouped with paper napkins, Rokeach (1968, 1973) has provided much of the im-
cups, and tissues, a more inclusive category of paper petus for research on values. "A value is an enduring
products would be created. And depending on what belief that a specific mode of conduct or end-state of
other items are grouped with paper products, people existence is personally or socially preferable to an
might form any of a variety of functionally labeled opposite or converse mode of conduct or end-state of
categories, such as "saves me time and energy," existence" (1973, p. 5). Thus values are special kinds
"keeps me healthy," "uses up scarce resources." It of preferences for modes of conduct or end-states of
is in this way that products are categorized and create existence. All behavior is not guided by values, only
groupings that are optimal in providing desired con- that behavior that is related to maintaining and en-
sequences that will provide value satisfaction. hancing self-esteem (Rokeach 1973, p. 14). A per-
It will be argued that these categories are based son's values are assumed to come from "culture, so-
on a cognitive element called a distinction. Distinc- ciety, and its institutions, and personality" (1973, p.
tions are dichotomies that represent the end points of 3). After a value is learned it is integrated into an
dimensions along which objects may be compared. If organized system of values in which all of a person's
two products can be compared using a hot/cold dis- values are ordered in importance with respect to one
tinction, this indicates that they may be compared by another.
temperature, although they may differ. An array of Two kinds of values have been defined by Ro-
products can be created by applying this distinction keach: terminal and instrumental. Terminal values are
to any of a limitless number of products. However, concerned with preferred end-states of existence (i.e.,
many products may not be comparable using this dis- happiness, security, accomplishment), and instrumen-
tinction. If a consumer doesn't think of a chain saw, tal values are related to modes of behavior (i.e., hon-
a dry cleaner or a box of paperclips as being either est, courageous, broad-minded) which are instrumen-
hot or cold, then none of these objects would be part tal in achieving these end-states.
of the array. Other products such as orange juice, cof- The interrelation between these "means" values
fee, tea, milk, water, com flakes or oatmeal could be and these "ends" values is referred to as a value sys-
compared on this basis. tem. Rokeach' s interest in values has not led to his
The resulting array can contain as many cate- developing mechanisms for translating instrumental
gories as there are products or just one category (if values into choices of objects to satisfy those values.
all objects were thought to be the same temperature). Therefore, from a marketing perspective more needs
The essential point is that the distinction being dichot- to be done to demonstrate how preferred end-states
omous has nothing to do with the nature of the array of existence are translated into specific choices in spe-
of products created by its application. A distinction cific situations. The author's model does not make use
is not an array of products; it is a dichotomous cog- of instrumental values but uses consequences more
nitive element in which the two parts of the dichot- akin to terminal values than instrumental values. A
omy have some meaningful relation to each other. consequence is a state of being produced by an act of
What is meaningful is a personal decision for each consumption. It is not an end state, but its relation or
consumer to make. ability to move the consumer toward an end state is

A Means-End Chain Model Based on Consumer Categorization Processes / 63

what gives the consequence a meaningful role in the juices are types of breakfast beverages and regular,
model. instant, and freeze-dried are subsumed by coffee). As
these categories are based on semantic properties
Vinson, Scott and Lamont's Centrality of (what things are called) rather than consumers' eval-
Beliefs uations of them, there is likely to be high agreement
The nature of the various means-end chains appearing among people as to the meaning of a brand.
in the literature is a function of the perspectives of However, choice is more a matter of evaluation
their authors. Vinson, Scott and Lamont (1977) have than merely being based on semantic meaning, so
stayed closest to Rokeach in providing a model of the Howard posits a hierarchical evaluative structure hav-
consumer's value-attitude system. Their model has ing two levels, one headed by terminal values and one
global values (a redefinition of terminal values) at the headed by instrumental values (Howard 1977, pp. 94,
most central level (closest to a person's self-concept). 163):
Domain-specific values, preferences for something
Terminal choice crite- belief about attitude
less general than end-states of being, are at an inter- Values ria for prod- product class toward
mediate level. And evaluative beliefs about products uct class product
are at the least central level. Domain-specific values class
are hypothesized to bridge "the gap between the tra- Instrumental choice crite- belief about attitude
ditional conception of closely held but general global Values ria for brand brand toward
brand
values and the less closely held descriptive and eval-
uative beliefs about products' attributes" (Vinson, Terminal and instrumental values are defined in the
Scott and Lamont 1977, p. 45). same way that Rokeach defines them. Howard defines
Values are defined as "centrally held cognitive choice criteria as "mental counterparts of the attri-
elements that stimulate motivation for behavioral re- butes by which a consumer judges a brand" (1977,
sponse" (Vinson, Scott and Lamont 1977, p. 45), that p. 28). Beliefs indicate where a product lies with re-
"exist in an interconnected hierarchical structure in spect to a given choice criterion. Attitude is concep-
which global values are related and connected to gen- tualized as beliefs weighted by the saliences given
eralized consumption-related values, which are, in them by choice criteria.
turn, similarly associated with product attributes" What Howard is suggesting is that grouping and
(Vinson, Scott and Lamont 1977, p. 49). However, distinguishing will result in semantic categories, and
although various levels of a means-end chain are ar- that choice criteria generated by terminal values en-
ticulated, no means are provided by which values at able consumers to choose appropriate product classes,
global and domain-specific levels are related to each while choice criteria generated by instrumental values
other nor how either level of values is tied to evalu- enable consumers to pick appropriate brands.
ations of product attributes. The author's model differs from Howard's model
in that the concept of instrumental values is not used.
Howard's Means-End Chain Model Values, as preferred end-states of being, are a type
Howard suggests that the total meaning of a brand is of consequence for which a person has no further
called its semantic structure (see Figure 2), which re- (higher) reason for preference. Values and conse-
fers to a hierarchically organized set of categories into quences are bases for categorization along with initial
which a brand is placed by a consumer (Howard grouping categories based on physical characteristics
1977, p. 42). The structure is hierarchical because (Howard's semantic meaning). Another key differ-
categories at the top (viz., breakfast beverages) sub- ence is the author's reliance on categorization as the
sume categories at lower levels (tea, coffee, milk and basis for levels in the means-end chain. There is only
one chain as opposed to hypothesizing parallel se-
mantic and evaluative chains. Value and consequence
level categories subsume grouping level categories,
creating a single hierarchy of categories across all lev-
FIGURE 2
Hypothetical Semantic Structure for Coffee as a
els of the means-end chain.
Breakfast Beverage
The Grey Benefit Chain
l1ilk Young and Feigin (1975) discuss benefit chain anal-
ysis as a method for linking emotional or psycholog-
I
Regular
ical benefits to product claims or product attributes.
I I
Brand A Brand B
I
Brand c
I
Brand A
I I
Brand B Brand C
I
Brand A
I
Brand B
I
Brand C
A benefit chain begins with a product description in-
cluding a specific product attribute. The consumer
provides two benefits that are derived from this at-

64 / Journal of Marketing, Spring 1982

tribute and two benefits that are derived from each of Measurement of the Means-End
the two initial benefits. The process is repeated once Chain Model
more, yielding 14 benefits in all. The links in the
chain can be demonstrated by presenting one of their Overview
examples (Young and Feigin 1975, p. 73): Figure 3 presents the measurement model for the
means-end chain. The figure is divided into three
Functional Practical Emotional
The Product Benefit Benefit Pay-Off parts: the means-end chain, matrices, and inputs and
outputs. The figure shows that the means-end chain
hair spray leaves hair I don't need leaves me is comprised of three levels of distinctions-values,
that holds easier to to spend so free to do
and leaves manage much time other things I consequences and grouping. The focus of distinctions
hair soft on hair want to do at the values and consequences levels is on what prod-
ucts can do for the consumer. Distinctions at the
This approach directly attacks the problem of expli- grouping level focus on product attributes. Inputs to
cating the means-end chain (the author would define the system are products and situations; outputs are
functional and practical benefits and emotional pay- products chosen for final consideration. The matrices
offs as types of consequences). It seeks associations represent the ways in which products and situations
between surface properties of products and benefits are categorized using the distinctions at the various
that are successively closer to the core values of the levels in the means-end chain.
consumer.
The Grey Benefit Chain, although it focuses only Values-Level Distinctions
on obtaining verbalizations associated with various The values-by-consequences matrix represents the
levels in the chain, offers the best view of the power way in which the accumulated experience of the con-
of the means-end approach to provide a structure for sumer may be represented. The matrix is used to de-
solving marketing problems at several levels simulta- termine which consequences are desirable and which
neously. However, this approach to means-end chain are not. Although values influence behavior, con-
development needs an infusion of theory to give it a sumers may not be aware that this is taking place.
broader perspective in linking fundamental consumer Values, as preferred end-states of existence, are prob-
processes to marketing stimuli. ably not in the forefront of consumer awareness.

FIGURE 3
Means-End Chain Model

MEANS-END
CHAIN MATRICES
.. INPUTS OUTPUTS

c=J
VALUE LEVEL VALUES SITUATIONS SITUATIONS
DISTINCTIONS

conse-
quences
D-,,,,,,,,,,,
UNDESIRABLE
I~ CONSEQUENCES +
CONSEQUENCE LEVEL RELEVANT PRODUCTS PRODUCT

c=J ---
DISTINCTIONS CONSEQUENCES CONSIDERED CHOSEN

c=J-
or
EVOKED
HIERARCHY
OF GROUPING
RELEVANT
CONSE-
I--- SET

LEVEL DIS- QUENCES

t ..
TINCTIONS

GROUPING
LEVEL
DISTINCTIONS
GROUPING
LEVEL
DISTINCTIONS c=J
PRODUCTS PRODUCTS

A Means-End Chain Model Based on Consumer Categorization Processes / 65

Also, the author is aware that not every consumer act Situational Influences
is going to be linked to values. The area of low-in- Situations encountered by consumers are assessed in
volvement marketing (Maloney and Silverman 1979) terms of consequences that will occur in each. Re-
speaks to this issue. The critical problem with low- spondents are asked to indicate which consequences
involvement products in terms of the model is the lack are likely to occur in which situations. This results in
of linkage between consequences from causing the a set of relevant consequences emerging that will
product and a consumer's values. guide behavior in a given situation. Rarely can all of
To ascertain consumers' values and their relation a consumer's desired consequences be achieved in
to consequences and attributes, some method is one situation, and, frequently, achieving some desired
needed to force consumers to think in value-related consequences means not achieving other desired con-
terms about their choices. An approach used by Gut- sequences or having to put up with some undesired
man and Reynolds (1979) for deriving values-level consequences. This point of view is consistent with
distinctions involves the use of the repertory grid the conclusion that the situational influence is a per-
(Kelly 1955). Application of the grid requires that re- suasive factor in consumer behavior (Belk 1975, p.
spondents state how two products of a set of three are 161).
similar and how they differ from the third product
(this typically produces attribute-level distinctions). Grouping-Level Distinctions
After obtaining several distinctions at this level using The grouping level is the level at which meaning is
this technique, a laddering procedure is used to gen- ascribed to products by grouping them together with
erate values-level distinctions. Respondents are asked other similar products and giving them a group label.
which pole of their dichotomous distinctions they Physical and surface properties of products are the
most prefer. Then they are asked why they most pre- bases for grouping products at the grouping level. As
fer the pole they do. The answer to the question typ- the most immediately perceived aspects of products,
ically leads to distinctions involving product functions they serve to identify them in their most fundamental
and consequences from use that become the basis for sense. Cars are to drive; refrigerators refrigerate;
generating still higher level distinctions. The proce- stoves cook; peas, green beans, carrots are to be
dure is repeated until respondents can no longer an- eaten. It is possible to discriminate products more
swer the question. Quite frequently respondents reach finely within each of these categories. There are many
a level where they are talking about terminal values ways of classifying cars, many types of refrigerators
such as those indicated by Rokeach. An example of and many kinds of vegetables.
the output from such a procedure is shown in Figure At the lowest levels of inclusiveness, products are
4, which is based on initially asking people why they grouped together in accordance with Howard's notion
prefer crunchy cereal. of semantic structure or Lunn's (1972) organization
of ways of satisfying consumer needs by product
types, product variants and brands. Product types sat-
isfy different needs beyond generic needs (toothpaste
vs. toothpowder, hot vs. cold cereal). Product vari-
FIGURE 4 ants are variations within types (with or without flu-
Connotative Conceptual Structure for Cereals oride, presweetened vs. not presweetened). And brands
represent minor differences within variants. Thus

r:::---
Crunchy
brands as specific alternatives are at the bottom of the
Has Just Like
hierarchy and are subsumed by product variants, and

.>~I <, - - - - - Crunchy product variants, in turn, are grouped by product type.
Cereal
At the upper levels of the hierarchy, the model
Stays Aids Lasts
assumes that evaluative categories subsume product
Gives
Me More
I
With Me Digestion Longer
In Mouth class categories. An evaluative category interposes
Energy Helps
Avoid More Flavor
between consequences that are properties of the per-
Snacks Experience son and attributes that are properties of the product.
I This is a result of the need for the grouping hierarchy
Aids
to relate to relevant consequences. If "time savings"
I Can Get
More Done I
Weight Loss
is sought in a food product, something that is "easy
to use" might be sought. "Easy to use" is not a prop-

~
Improves My
Appearance
erty of a product per se (there is no "easy to use" in
I
Makes Me a product), and it is not a consequence (something
Feel Better
that happens to a person), like "time savings" is a
consequence.

66 / Journal of Marketing, Spring 1982

 Thus in Figure 5, which shows one person's could also be broken up using the same process in
grouping hierarchy for breakfast beverages, soft drinks reverse.)
subsumes colas, lemon-lime and other flavors. At Another approach (Reynolds 1981) begins by elic-
higher levels in the grouping hierarchy, the level of iting distinctions from respondents using the modified
a distinction is determined by its ability to separate repertory grid procedure and the ordering procedure
products into preferred and nonpreferred sets (see (both described earlier). The respondent then fills out
Tversky's theory of sequential choice, 1972). If "not a distinctions-by-products matrix, where a 1 indicates
easy to prepare" meant a product was not acceptable that a product would move a person toward that pole
for use regardless of any other properties, it would be of a distinction and a 0 indicates that it wouldn't. A
more functional to have it subsume "good for procedure described as multidimensional Guttman
you"-"not good for you." However, if a consumer scaling is used to find Guttman-like chains or relations
would be more likely not to consider a product if it among distinctions in this matrix. These chains, then,
were "not good for you" even if it were easier to would constitute the distinctions comprising the
prepare than "good for you," "not good for you" grouping hierarchy.
would be at a higher level in the hierarchy. The goal
of categorizing products at the grouping level is to Consequence Level Distinctions
identify products and array them into sets based on As discussed previously, desirable consequences
their possession of attributes that imply the ability to (benefits) derive from values. If a person values "so-
provide desired consequences for the consumer. cial recognition"--one of Rokeach's values-he/she
Two approaches seem most suitable for determin- will develop categories of objects that lead to an in-
ing the distinctions that comprise a consumer's group- crease in social recognition. Benefits such as prestige,
ing hierarchy. A direct approach may be used (Green, sex appeal, white teeth, curly hair and countless other
Fox and Camone 1969) involving hierarchical merg- benefits will act as the basis for distinctions that ca-
ing. The respondent is asked to categorize the prod- tegorize together objects that satisfy these criteria.
ucts into groups so as to make the products within One does not literally build benefits into a prod-
groups as similar as possible and different from prod- uct. Physical product attributes that provide the ben-
ucts in other groups. The respondent is asked to efits (e.g., cost, size, design) are built into a product.
merge the two groups he or she considers most similar It is essential to develop the correspondence between
and to repeat the process until only one group re- distinctions at the consequence level (e.g., prestige)
mains. Group labels at each stage provide the dis- and the product attributes that imply that the product
tinctions for the grouping hierarchy. (Large groups can provide such benefits. These relations are devel-

FIGURE 5
Grouping Hierarchy for Breakfast Beverages

Breakfast
~B'.'<'B"

Tastes Doesn't Taste

Good Good

--------~
Easy to Not too easy
I
Easy to
Prepare to Prepare Prepare

Not ~~ Harmful Not /"" Harmful

I
Not Harmful

/rm~ /m~ I
.r-, •. •.I
Cold/ " ' "Hot / "

I /""'-
Cold Hot Cold Hot Cold Hot Hot

JUi{ \ t e r I ~Jee I I
cOfLe Coffee Tea
I~k
Frozen Canned
COffje
Freeze
Drinks
Colas
~~
I
Other
Cocoa
Fresh
Cocoa Ramos
Regular Gin Cafe Aulait
I
Instant
Dried Lemon Fizz
Lime

A Means-End Chain Model Based on Consumer Categorization Processes / 67

oped as consumers learn about products from pack- implied by the fact that a certain brand of clothing is
aging information, advertising, discussions with oth- carried by the finest stores. There are other bases from
ers, and product trial. The product attributes that which this might be inferred, such as advertised in
imply the ability of the product to provide benefits Vogue, worn by the right people, or even intrinsic
become key defining attributes for the consumer. good taste (in the same sense one learns to identify
Many relations between attributes and conse- and appreciate artistic creations). Thus the following
quences are clear-cut and easy to determine. For beer distinctions might represent this consumer's means-
drinkers, ability to control weight is clearly tied to end chain:
caloric level in beer, resulting in the category of light
beer as opposed to regular beer. In some cases the Levels Distinctions
relationships between consequences and attributes are Value Social Recognition
Consequence Prestige
not as clear or easy to determine. The complexity of Grouping High Styling
or implicative distance between a distinction at the Grouping Carried by Finest Stores
consequence level and product categories at the
grouping level require intermediate distinctions be- The same progression may be examined from the
tween these desirable consequences and the actual perspective of levels of inclusiveness. Prestige is but
product attributes which are so much a part of dis- one of many ways to get social recognition, wearing
tinctions at the grouping level. high-styled clothing is but one of many ways to obtain
A consumer who wants good health and sees the prestige, and shopping at the finest stores is but one
connection between diet and health may form a nat- of many ways to identify high style in clothing. Such
ural-artificial distinction. Such a distinction could be behavior has potential negative consequences too.
applied to drinks by comparing drinks grouped by Shopping at the finest stores costs lots of money. The
type (fruit juice, fruit drinks, water, milk, or soft consequences associated with requiring lots of money
drinks) into an ordered array of categories along the when you don't have it are, among others, anxiety.
dimension defined by the natural-artificial distinction. And anxiety has its consequences, which are exacted
An ordered array is a related set of categories whose physiologically. These physiological consequences
boundaries are ordinally scaled with respect to the may produce substantial conflict with other desired
distinction of interest (Green & Tull 1978, p.173). end-states.
The sorting of products into categories using dis- The means-end chain is focused on the linkages
tinctions can be achieved by printing one pole of each between where a person wants to be and the means
distinction on a small card and asking respondents to chosen to get there. A product is indeed a bundle of
put products into groups that reflect equal amounts of attributes, and people do consume products for the
whatever dimension is specified by the distinction desirable consequences they get (trading off with any
(see Bourgeois, Haines and Sommers 1979; Gutman undesirable consequences). It is the implicative rela-
1977; Gutman and Reynolds 1977, 1979, for exam- tions between product attributes and consequences
ples of this methodology). and between consequences and values that are the
Each of the drink-types could be kept in its own objects of study. How these products are grouped at
category or recategorized according to its degree of various levels in the means-end chain tells us how
artificialness-naturalness, regardless of other proper- each product fits together with other products in the
ties. The manner in which this would be done is a context of the consumer's goal-oriented behavior.
function of the object properties and the values of the To create a set of products to be considered in any
consumer. In some cases such as drinks, the relations given situation, distinctions at the grouping level are
between distinctions at the grouping, consequence evaluated by the relevant distinctions at the conse-
and value levels may be clear, but there are products quence level. The output from this Relevant Conse-
where people may have great difficulty judging the quence Distinction by Grouping Level Distinction
alternatives. matrix is the determination of a path through the
Thus, in clothing, for example, where many peo- grouping level distinctions hierarchy which specifies
ple are not capable of determining quality of styling the set of products to be considered. This set of prod-
without the validation of others, surrogate indicators ucts is then itself evaluated by the Relevant Conse-
such as retail outlet are of primary importance. A con- quence Distinctions, resulting in the product or prod-
sumer placing importance on the value of social rec- ucts to be chosen in the situation.
ognition might seek the benefit of prestige from his/
her clothing selections as a means of gaining social Distinguishing Distinctions
recognition. Such a benefit might be implied by At this point in the model, the consumer has gener-
clothes judged to have high styling (an evaluation of ated a product or set of products for consideration,
a property of the clothing), and high styling might be often referred to as an evoked set, by identifying those

68 / Journal of Marketing, Spring 1982

products possessing attributes that imply the ability
to provide benefits which in tum are instrumental to FIGURE 7A
the achievement of desired ends. If this process Grouping Level Distinctions by Relevant Benefits
doesn't result in one product being preferred to all Matrix
others, further processing is still needed. However,
similarity has been established, both in terms of the Relevant Benefits Ordered
by Importance
products' physical properties at the grouping level and
their ability to provide similar advantages, which lead Grouping Level
Good
Flavor
Experience
Time
Saving
Good
For you Stimulating Affordable
toward desired values at the benefit level. The basis Distinctions

for this similarity has no further diagnostic use in dis- Tastes Good
Tastes Bad
x
tinguishing among products at this stage.
Therefore, values cannot further influence choice Easy to Prepare
Hard to Prepare
X

directly or indirectly. One is left with simple prefer-

ence (just prefer one to the other) or stochastic Not Harmful X

choice, wherein variety results by random selection Harmful

among alternatives or is based on low level choice Hot

among insignificant product differences that have no
X
Cold

stability over time.

An Illustration of the Model

For measurement purposes, the means-end chain can ment the means-end model described above: a Values-
be conceived of as a series of connected matrices. by-Consequences matrix, a Consequence-by-Situa-
Using the product category of drinks as an example, tions matrix, a Relevant Consequences-by-Grouping
the following matrices need to be generated to imple- Distinctions matrix, and a Relevant Consequences by
Product matrix.
This system is illustrated in Figures 6 and 7. Each
of the matrices is described below. The approach can
be outlined without becoming embroiled in the meth-
FIGURE 6 odology for generating the data. Only a few distinc-
Illustration of Means-End Chain
tions are shown at each level to keep the illustration
VALUES at a manageable level.
c
o Matrix 6A contains the relations between the con-
sequences deriving from consuming drinks and the
~

[ values consumers hold. The l's in the matrix are

~ '"
j .~ "signed" to indicate the relation between values and
CONSEQUENCES
... '" consequences. Some of the situations during which
Good for You
Stimulating +1 +1 drinks are consumed are shown along the top of Ma-
Affordable
Not Time Consuming
+1 +1
+1 trix 6B. Consequences to which the consumption of
Good Flavor Experience
Heartburn -I -I
+
- drinks in these situations might ultimately be related
"Nerves" - -
are shown on the side of Matrix 6B (along with their
-I
Fattening -I -I

DESIRABLE & FIGURE 7B

UNDES IRABLE
CONSEQUENCES
Freeze-Dried Coffee Products Considered by
Relevant Benefits
Good for You (+) I I I
Stimulating (+) I I 1 1
Affordable (+) 1 1
Not Time Consuming (+) 1 1 (see Figure 7) Good
Good Flavor
Experience (+) 1 1 1 1 I EX:~: :~ce Affordable *
Heartburn (-)
"Nerves It ( _ )

~
Fattening (-) Brand A

~
Product.
Chosen
_ 0'
-Evoked
Set

GROUPING HIERARCHY FOR BREAKFAST BEVERAGES

*Other benefits are presumed not to vary (have no diag-
(see Figure 5) nostic value) for within group comparisons

A Means-End Chain Model Based on Consumer Categorization Processes / 69

algebraic sign from Matrix 6A). The 1's in the matrix possible by drinks within the tastes good, easy to pre-
indicate which consequences are relevant in which pare, not harmful, hot portion of the grouping hier-
situations. If breakfast is the situation of interest, the archy. Comparison of specific alternatives by relevant
columns of the Situations-by-Consequences matrix for consequences shows that freeze-dried coffee is the
which there were 1's in the "breakfast" row become preferred breakfast drink. If cold as well as hot drinks
the columns of the Relevant Consequences-by-Prod- provided relevant benefits, the choice set would have
ucts matrix. The result is that the following conse- included juice, milk and water in addition to coffee.
quences are deemed relevant to the choice of a drink If more than one product emerges after this screening
in the breakfast situation: good flavor experience, process, these specific products would be compared
time saving, good for you, stimulating, affordable. for any differences in their attributes to promote the
It is unlikely that the values operating in any given desired benefits. If at the end of this process there was
situation are immediately and directly apparent. A still more than one alternative, some choice process
series of steps are exemplified by the laddering pro- would have to be made, but it wouldn't be made on
cedure which interposes between one end of the the basis of the values orientation that controls the
means-end chain and the other. The purpose of these structure of the means-end chain.
techniques would be to determine the distinctions The selection of one product or the consideration
comprising the chain and the connections between of specific products at this stage may entail some un-
them. desired consequences that come along with those
The undesirable consequences (indicated by minus products. The consumer would have to trade-off these
signs) are included in matrix 6B, although they would undesired consequences when considering the desir-
hardly be sought by consumers in these situations. able consequences.
Their absence, however, may be desired by some con-
sumers who are conscious of products providing these
undesirable consequences. In such cases, one could
put a "not" in front of the consequence and change
the undesired consequence (-) into a desired conse- Applications
quence (+).
The lower part of Figure 6 refers to Figure 5, The translation of the interest in values to actual re-
which showed a hierarchy of products at the grouping search on values has been retarded by the lack of an
level. The distinctions are shown with the products integrated theory by which values are held to mediate
categorized by them. It shows the identity (or product behavior. The model presented in this paper can serve
class categories) and evaluative categories. as a catalyst for such research. Building values into
Figure 7A shows the grouping distinctions by rel- marketing planning creates the potential to target
evant consequence matrix. Both grouping distinctions products and messages more closely to valued states
and relevant consequences are ordered in terms of im- or goals considered by consumers to be of prime im-
portance. The determination of the final set of prod- portance.
ucts to be considered is made comparing products cat- The means-end chain model offers researchers a
egorized by each side of the distinctions in terms of guide to procedures that specifically address the link-
consequences provided deriving from their consump- ages connecting values important to the consumer to
tion. Row 1 of Figure 7A shows that products that specific attributes of products. Thus consumers' per-
taste good provide the desired consequence of good ceptions and evaluations of products can be studied
flavor experience. Those products that taste good are at different levels in the means-end chain. Consumers
compared for ease of preparation. Easy-to-prepare tend to operate at the lowest level of abstraction (the
products provide the benefit of time saving. Among grouping level) when sorting products into categories
easy-to-prepare products that taste good, not harmful on whatever basis they choose (Gutman 1977). This
ones are preferred to harmful ones because they pro- model and the accompanying measurement proce-
vide more of the benefit good for you. (Some con- dures represent a way for researchers to systematically
sumers may discriminate among drinks by seeking link attribute level distinctions used to form product
those that are nutritious or actually good for you; oth- classes to distinctions at the consequence and values
ers as above, may have a less stringent criterion; still level where the basic directions of consumer choice
others may not discriminate among drinks on this ba- are determined.
sis at all.) Lastly, among not harmful, easy-to-prepare Vinson, Scott and Lamont (1977) have outlined
drinks that taste good, hot drinks are preferred at several areas where values can contribute to market-
breakfast because they provide the benefit of being ing management. It might be useful to point out how
stimulating. the means-end chain model presented herein can con-
Figure 7B shows the relevant consequences made tribute in these areas.

70 / Journal of Marketing, Spring 1982

Market Analysis and Segmentation product attributes and consumer benefits and values
Such a perspective would be valuable in determining are essential if marketing managers are to specify
product markets from a consumer point of view. Var- clearly desired products' characteristics to research
ious researchers have defined product markets ranging and development.
in breadth from interbrand competition (Butler and Promotional Strategy
Butler 1970, 1971) to competition among widely di-
vergent alternatives (Moran 1973). The ability to ex- Establishing linkages between distinctions at all levels
amine substitutability at the various levels of the in the means-end chain has much potential for adver-
means-end chain can provide a method for such con- tising. The Grey Benefit Chain was itself developed
siderations and could aid in identifying competitors to aid advertisers in generating effective ad claims.
that might otherwise be overlooked. The approach described in this paper extends the
Segmentation research seeks to find groups of scope of the central idea of the Grey Benefit Chain.
consumers defined in terms of demographics, atti- If an advertiser knew what key distinctions consumers
tude, lifestyle, geographic area or other variables that made, how these distinctions were used to establish
have a differential response to marketing control vari- product/brand groupings, and how this process was
ables (Wind 1978). The means-end approach can de- carried out at each level in the means-end chain, he/
termine the influence of values for consumer seg- she would be better able to position advertising
ments and relate these values to distinctions at the against the fundamental value orientations of target
consequence and grouping levels. consumers. (This approach also offers the possibility
Specific choices among alternatives are not deter- of increasing the involvement level of low involve-
mined at the values level. But the spectrum of avail- ment products.)
able alternatives can be divided into groups containing Research on the Grey Benefit Chain and findings
objects with the potential to satisfy values and into by Gutman and Reynolds (1979) suggest that con-
groups without value-satisfying alternatives. Subgroups sumers use surface properties of products as their ini-
of consumers can be established who categorize a tial basis for categorization of products. At the same
product not currently used in value-satisfying cate- time, if one accepts values as influencing consumer
gories. Also, a segmentation approach can separate choice, the linkages between the two levels become
current users of a product class or brand from poten- very important. The author's view of the means-end
tial users. The means-end chain approach can identify chain is that advertising research procedures often fo-
how consequences and product properties are linked cus on consumers' low level grouping distinctions
to relevant values by members of these two segments. without recognizing the hierarchical structure that
forms the basis for connecting these distinctions to
higher levels in the hierarchy. Similarities and differ-
Product Planning ences are two sides of the coin-we can better un-
The linkage among the levels of the means-end chain derstand how products and brands differ when we
ties physical product attributes to values important to understand how they are the same. The context cre-
consumers. Knowing which product attributes are ated by situation interacts with values to select dis-
used by consumers to infer the presence of desired tinctions at an appropriate level. These processes have
consequences permits clearer specifications for prod- to be understood if we are to communicate effectively
uct development by new products departments in their with consumers.
charges to research and development.
Distinctions and their role in forming categories
of products can aid managers in determining which
competitive brands/products a brand is categorized Conclusions
together with at any level in the means-end chain. The The means-end model presented in this paper is based
level in the grouping hierarchy at which a product is on the distinctions consumers use in grouping objects
first grouped together with another product will de- by virtue of their similarity and in distinguishing
termine its basic similarity to that product. At the among them by virtue of how they vary along com-
grouping level, products/brands are categorized more mon dimensions. The model is hierarchical in that it
on the basis of physical similarity. At the consequence links values that are the ends to basic feature com-
level, categorization is based on functional equiva- ponents of products/brands representing the means.
lence in terms of advantages provided to the consumer Methodology has only been touched upon, al-
by consumption of the product. Also, one can con- though sorting procedures, repertory grid techniques,
tinue up to the values level of the means-end chain and laddering were all mentioned as data-gathering
and group a product with other value-satisfying alter- techniques. To a great extent, the model has an ideo-
natives. Knowledge of linkages between physical graphic as opposed to a nomethetic orientation. One

A Means-End Chain Model Based on Consumer Categorization Processes / 71

has to be able to understand individuals before he/ ysis uniquely suited to the binary or categorical data
she can understand groups of individuals (Lewin that are typical outputs from the methodologies noted
1951). This, however, doesn't preclude use of scaling above. Among these, logical construct analysis, akin
or clustering procedures or other types of aggregate to multidimensional scaling (Gutman and Reynolds
analytical techniques. There are some forms of anal- 1977, Reynolds 1981) has shown much promise.

REFERENCES
Belk, R. W. (1975) "Situational Variables and Consumer Lewin, K. (1951), Field Theory in Social Science, New York:
Behavior," Journal of Consumer Research, 2 (December), Harper.
157-164. Lunn, T. (1972), "Segmenting and Constructing Markets,"
Bourgeois, J. D., G. H. Haines and M. S. Sommers (1979), in Consumer Market Research Handbook, R. M. Worces-
"Defining an Industry," paper presented to the TIMS/ ter, ed., Maidenhead, Berkshire: McGraw-Hill.
ORSA Special Interest Conference on Market Measurement Maloney, J. C. and B. Silverman, eds. (1979), Attitude Re-
and Analysis, Stanford, CA, March 26. search Plays for High Stakes, Chicago: American Market-
Butler, B., Jr. and D. H. Butler (1970), "Hendrodynamics: ing Association.
Fundamental Laws of Consumer Dynamics," Croton-on- Moran, W. R. (1973), "Why New Products Fail," Journal
Hudson, NY: Hendry Corp., unpublished paper, Chapter 1. of Advertising Research, 13 (April), 5-13.
- - - and (1971), "Hendrodynamics: Funda- Myers, J. M. (1976), "Benefit Structure Analysis: A New
mental Laws of Consumer Dynamics," Croton-on-Hudson, Tool for Product Planning," Journal of Marketing, 40
NY: Hendry Corp., unpublished paper, Chapter 2. (October), 23-32.
Fennell, G. (1978), "Consumer's Perceptions of the Product- Reynolds, T. J. (in press), "ERGO: A New Approach to
Use Situation," Journal of Marketing, 42 (April), 38-47. Multidimensional Item Analysis," in Educational and Psy-
Green, P., L. Fox and F. J. Carmone (1969), "Television chological Measurement.
Program Similarities: An Application of Subjective Clus- Rokeach, M. J. (1968), Beliefs, Attitudes and Values, San
tering," Journal of the Market Research Society, 11 (Jan- Francisco: Jossey Bass.
uary), 70--90. - - - - (1973), The Nature of Human Values, New York:
- - - - and D. S. Tull (1978), Research for Marketing The Free Press.
Decisions, Englewood Cliffs, NJ: Prentice-Hall. Rosch, E. (1978) "Principles of Categorization," in Cognition
Gutman, J. (1977), "Uncovering the Distinctions People and Categorization, E. Rosch and B. Lloyd, eds., Hills-
Make Versus the Use of Multi-Attribute Models: Do A dale, NJ: John Wiley and Sons, 25-49.
Number of Little Truths Make Wisdom?" in Proceedings Rosenberg, M. (1956), "Cognitive Structure and Attitudinal
of the Twenty-third Annual Conference of the Advertising Effect," Journal of Abnormal and Social Psychology, 53,
Research Foundation, New York: Advertising Research 367-372.
Foundation, 71-76. Tversky, A. (1972), "Elimination by Aspects: A Theory of
and T. J. Reynolds (1977), "A Pilot Test of a Choice," Psychological Review, 79 (July), 281-299.
Logic Model for Investigating Attitude Structure," in Mov- - - - - and I. Gati (1978) "Studies of Similarity," in Cog-
ing Ahead with Attitude Research, Y. Wind and M. Green- nition and Categorization, E. Rosch and B. Lloyd, eds.,
berg, eds., Chicago: American Marketing Association, Hillsdale, NJ: John Wiley and Sons, 81-98.
19-28. Vinson, D. E., J. E. Scott and L. M. Lamont (1977), "The
- - - - and (1979), "An Investigation of the Role of Personal Values in Marketing and Consumer Be-
Levels of Cognitive Abstraction Utilized by Consumers in havior," Journal of Marketing, 41 (April), 44-50.
Product Differentiation," in Attitude Research Under the Wind, Y. (1978), "Issues and Advances in Segmentation Re-
Sun, J. Eighmey, ed., Chicago: American Marketing As- search," Journal of Marketing Research, 15 (August),
sociation, 128-50. 317-337.
Haley, R. I. (1968), "Benefit Segmentation: A Decision Ori- Yankelovich, D. (1981), New Rules, New York: Random
ented Research Tool," Journal of Marketing, 32 (July), House.
30--35. Young, S. and B. Feigin (1975), "Using the Benefit Chain
Howard, J. A. (1977), Consumer Behavior: Application and for Improved Strategy Formulation," Journal of Market-
Theory, New York: McGraw Hill Book Company. ing, 39 (July), 72-74.
Kelly, G. A. (1955), The Psychology of Personal Constructs,
New York: Norton.

72 I Journal of Marketing, Spring 1982