You are on page 1of 26

LADDERING THEORY,

METHOD, ANALYSIS,
AND
INTERPRETATION
Thomas J. Reynolds and Jonathan Gutman
Journal of Advertising Research Feb/March, 1!!
ersonal values research in mar"eting has recently received a substantial amount of
attention from both academics and #ractitioners This more in$de#th #rofiling of the
consumer and his or her relationshi# to #roducts offers #otential not only for understanding
the %cognitive% #ositionings of current #roducts but also #ermits the develo#ment of #ositioning
strategies For ne& #roducts. 'ndorsing this more #sychological vie& of the mar"et#lace, (heth
)1!*+ suggests that to be com#rehensive in mar"eting #roducts in the 1!,-s both researchers
and management are going to have to, if they have not already, ado#t this consumer$based
orientation rather than one that merely focuses on #roduct characteristics.
P
The a##lication of the #ersonal values #ers#ective to the mar"eting of consumer #roducts can
be classified into t&o theoretically grounded #ers#ectives, %macro% re#resenting sociology and
%micro% re#resenting #sychology )Reynolds, 1!.+. The macro a##roach refers to standard
survey research methodology combined &ith a classification scheme to categori/e res#ondents
into #redetermined clusters or grou#s )e.g..0A1( methodology of the (tanford Research
2nstitute+. 3roducts and their #ositioning strategies are then directed to a##eal to these general
target grou#s, such as the Merrill 1ynch solitary bull a##ealing to the achiever orientation &hose
desire is to send out and 4get ahead of the #ac"5 )3lummer, 1!.+.
Reynolds )1!.+ notes, though strong on face validity these rather general classifications fail
to #rovide an understanding, s#ecifically, of ho& the concrete as#ects of the #roduct fit into the
consumer6s life. As such, the macro survey a##roach only gives #art of the ans&er, namely, the
overall value orientation of target segments &ithin the mar"et#lace. Missing are the "ey defining
com#onents of a #ositioning strategy7the lin"ages bet&een the #roduct and the #er.onally
relevant role it has in the life of the consumer.
The more #sychological #ers#ective offered by the %micro% a##roach based u#on Means$'nd
Theory )Gutman 1!8+, s#ecifically focuses on the lin"ages bet&een the attributes that e9ist in
#roducts )the %means%+, :he conse;uences for the consumer #rovided by the attributes, and the
#ersonal values )the 4ends5+ the conse;uences reinforce. The means$end #ers#ective closely
#arallels the origin of attitude research re#resented by '9#ectancy$0alue Theory )Rosenberg,
1.<+, &hich #osits that consumer actions #roduce conse;uences and that consumers 1earn to
associate #articular conse;uences &ith #articular #roduct attributes they have reinforced through
their buying behavior. The common #remise, then, is that consumers learn to choose #roducts
containing attributes &hich are instrumental to achieving their desired conse;uences. Means$'nd
Theory sim#ly s#ecifies the rationale underlying &hy conse;uences are im#ortant, namely,
#ersonal values.
The focus of this article is on detailing the s#ecifics of the in$de#th intervie&ing and analysis
methodology, termed 4laddering5 )Gutman and Reynolds, 1=> Reynolds and Gutman, 1!?a+,
for uncovering means$end hierarchies defined by these "ey elements and their lin"ages or
connections. The combination of connected elements, or ladder, re#resents the lin"age bet&een
the #roduct and the #erce#tual #rocess of consumers, &hich as #ointed out #reviously, yields a
more direct and thus more useful understanding of the consumer.
Laddering
1addering refers to an in$de#th, one$on$one intervie&ing techni;ue used to develo# an
understanding of ho& consumers translate the attributes of #roducts into meaningful associations
&ith res#ect to self, follo&ing Means$'nd Theory )Gutman, 1!8+. 1addering involves a tailored
intervie&ing format using #rimarily a series of directed #robes, ty#ified by the 4@hy is that
im#ortant to youA5 ;uestion, &ith the e9#ress goal of determining sets of lin"ages bet&een the
"ey #erce#tual elements across the range of attributes )A+, conse;uences )B+, and values )0+.
These association net&or"s, or ladders, referred to as #erce#tual orientations, re#resent
combinations of elements that serve as the basis for distinguishing bet&een and among #roducts
in a given #roduct class.
2t is these higher$order "no&ledge structures that &e use to #rocess information relative to
solving #roblems )Abelson, 1!1+, &hich, in the consumer conte9t, is re#resented by choice.
Casically, distinctions at the different levels of abstraction, re#resented by the A$B$0s, #rovide
the consumer &ith more #ersonally relevant &ays in &hich #roducts are grou#ed and cate$
gori/ed. Thus, the detailing and subse;uent understanding of these higher level distinctions
#rovides a #ers#ective on ho& the #roduct information is #rocessed from &hat could be called a
motivational #ers#ective, in that the underlying reasons &hy an attribute or a conse;uence is
im#ortant can be uncovered.
For e9am#le, the follo&ing ladder, starting &ith a basic distinction bet&een ty#es of snac"
chi#s, re#resents #art of the data collection from a single subDect in a salty$snac" studyE
)0+ self$esteem
2
)B+ better figure
2
)B+ don6t get fat
2
)B+ eat less
2
)A+ strong taste
2
)A+ flavored chi#
These elements &ere se;uentially elicited from the res#ondent as a function of the laddering
techni;ue6s ability to cause the res#ondent to thin" critically about the connections bet&een the
#roduct6s attributes and, in this case, her #ersonal motivations.
The analysis of 1addering data such as this across res#ondents first involves summari/ing the
"ey elements by standard content$analysis #rocedures )FassarDian, 1==+, &hile bearing in mind
the levels of abstraction, A$B$0, conce#tuali/ation. Then a summary table can be constructed
re#resenting the number of connections bet&een the elements. From this summary table domi$
nant connections can then be gra#hically re#resented in a tree diagram, termed a hierarchical
value ma# )G0M+. )This ty#e of cognitive ma#, unli"e those out#ut from traditional factor
analysis or multidimensional scaling methods, is structural in nature and re#resents the lin"ages
or associations across levels of abstraction Hattributes$conse;uences$valuesI &ithout reference to
s#ecific brands.+ Jnfortunately, though basically accurate, this general descri#tion of> the
analysis #rocess has not been s#ecific enough to #ermit first$time analysts )or their su#eriors+ to
feel comfortable &ith dealing &ith all the vagaries of ;ualitative data of this ty#e. Thus, a ste#$
by$ste# #rocedure, including both the analysis and the assessment of the resulting ma#, &ill be
detailed by &ay of e9am#le later.
2nter#retation of this ty#e of ;ualitative, in$de#th information #ermits an understanding of
consumers6 underlying #ersonal motivations &ith res#ect to a given #roduct class. 'ach uni;ue
#ath&ay from an attribute to a value re#resents a #ossible #erce#tual orientation &ith res#ect to
vie&ing the #roduct category. Gerein lies the o##ortunity to differentiate a s#ecific brand, not by
focusing on a #roduct attribute, but rather by communicating ho& it delivers higher level
conse;uences and ultimately ho& it is #ersonally relevant, essentially creating an 4image
#ositioning.5 This understanding ty#ically serves as the basis for the develo#ment of advertising
strategies, each re#resenting a distinct 4cognitive5 #ositioning, &hich reinforces the various
levels of abstraction for a given #erce#tual orientation )Klson and Reynolds. 1963; Reynolds and
Gutman, 1!?+.
2n sum, the e9#ress #ur#ose of the intervie&ing #rocess is to elicit attribute$conse;uence$value
associations consumers have &ith res#ect to a #roduct or service class. The general notion is to
get the res#ondent to res#ond and then to react to that res#onse. Thus, laddering consists of a
series of directed #robes based on mentioned distinctions initially obtained from #erceived
differences bet&een and among s#ecific brands of #roducts or services. Again, after the initial
distinction obtained by contrasting brands is elicited, all subse;uent higher$level elements are not
brand s#ecific. The laddering results can be used to create an G0M summari/ing all intervie&s
across consumers, &hich is inter#reted as re#resenting dominant #erce#tual orientations, or
4&ays of thin"ing,5 &ith res#ect to the #roduct or service category.
Objectives
(ince the introduction of the laddering methodology into the consumer research domain, nu$
merous a##lications, both a##lied and academic, have been e9ecuted )Gutman, 1!?> Gutman
and Alden, 1!?> Gutman and Reynolds, 1!*> Gutman, Reynolds, and Fiedler, 1!?> Klson and
Reynolds, 1!*> Reynolds and Gutman, 1!?a> Reynolds and Gutman, 1!?b> Reynolds and
Jamieson, 1!?+. Again, the #rimary a##lication has been to develo# a cognitive hierarchical
value ma# indicating the interrelation of the attributes, conse;uences, and #ersonal values for a
given #roduct or service category.
Jnfortunately, the term laddering in the mar"eting community has become a some&hat generic
term re#resenting merely a ;ualitative, in$de#th intervie&ing #rocess )Morgan, 1!?+, &ithout
reference to either its theoretical under#innings )Gutman, 1!8+ or the rather critical distinction
bet&een the intervie&ing #rocess and analytical methods used to derive meaning from the re$
sulting data )Lurgee, 1!.+. Mot only have these critical distinctions been overloo"ed, but even
the standard definition of laddering as an intervie&ing methodology, to date, has not been
addressed in the academic literature. Given the value of this ty#e of in$de#th understanding of
the consumer, in #articular, the #otential &ith res#ect to the s#ecification of more accurate and
a##ro#riate #ositioning strategies, a com#rehensive documentation of this research a##roach is
needed.
Thus, it is the #rimary obDective of this article to detail the intervie&ing techni;ues that #ertain to
laddering in order to #rovide a foundation for both its a##lication as &ell as subse;uent method
evaluation. A secondary obDective is to #rovide a detailed descri#tion of ho& the analysis of this
s#ecific ty#e of ;ualitative data is #erformed. The third and final obDective is to demonstrate ho&
the laddering results are inter#reted &ith res#ect to develo#ing and understanding #erce#tual
orientations and #roduct #ositionings.
Interview Environment
General Considerations.
An intervie&ing environment must be created such that the res#ondents are not threatened and
are thus &iling to be intros#ective and loo" inside themselves for the underlying motivations be$
hind their #erce#tions of a given #roduct class. This #rocess can be enhanced by suggesting in
the introductory comments that there are no right or &rong ans&ers, thus rela9ing the res#on$
dent, and further reinforcing the notion that the entire #ur#ose of the intervie& is sim#ly to
under$
stand the &ays in &hich the res#ondent sees this #articular set of consumer #roducts. 3ut sim#ly,
the res#ondent is #ositioned as the e9#ert. The goal of the ;uestioning is to understand the &ay in
&hich the res#ondent sees the &orld, &here the &orld is the #roduct domain com#rised of
relevant actors, behaviors, and conte9ts. The a##roaches and techni;ues discussed in this article
are designed to assist the res#ondent in critically e9amining the assum#tions underlying their
everyday common#lace behaviors. @ic"er )1!.+ discusses ho& researchers might use some of
these same devices in brea"ing out of their traditional modes of thin"ing.
2m#ortantly, intervie&ers must #osition themselves as merely trained facilitators of this dis$
covery #rocess. 2n addition, due to the rather #ersonal nature of the later #robing #rocess, it is
advisable to create a slight sense of vulnerability on the #art of the intervie&er. This can be
accom#lished by initially stating that many of the ;uestions may seem some&hat obvious and
#ossibly even stu#id, associating this #redicament &ith the intervie&ing #rocess, &hich re;uires
the intervie&er to follo& certain s#ecific guidelines.
Kbviously, as &ith all ;ualitative research, the intervie&er must maintain control of the in$
tervie&, &hich is some&hat more difficult in this conte9t due to the more abstract conce#ts that
are the focus of the discussion. This can be best accom#lished by minimi/ing the res#onse
o#tions, in essence being as direct as #ossible &ith the ;uestioning, &hile still follo&ing &hat
a##ears to be an 4unstructured5 format. Cy continually as"ing the Why is that im#ortant to
youA5 ;uestion, the intervie&er reinforces the #erce#tion of being genuinely interested and thus
tends to command the res#ect and control of the dialogue.
Cy creating a sense of involvement and caring in the intervie&, the intervie&er is able to get
belo& the res#ondent6s surface reasons and rationali/ations to discover the more fundamental
reasons underlying the res#ondent6s #erce#tions and behavior. Jnderstanding the res#ondent
involves #utting aside all internal references and biases &hile #utting oneself in the res#ondent6s
#lace. 2t is critical that ra##ort be established before the actual in$de#th #robing is initiated as
&ell as maintained during the course of the intervie&. Casically, the intervie&er must instill
confidence in the res#ondent so the o#inions e9#ressed are #erceived as sim#ly being recorded
rather than Dudged.
Also critical to the intervie&ing #rocess is the ability of the intervie&er to identify the
elements brought forth by the res#ondent in terms of the levels of abstraction frame&or". Thus, a
thorough familiarity &ith the Means$'nd theory is essential.
(ensitive areas &ill fre;uently #roduce su#erficial res#onses created by the res#ondent to
avoid intros#ection about the real reasons underlying the res#ondent6s behavior. A clinical sensi$
tivity is further re;uired of the intervie&er to both identify and deal &ith these fre;uent and #o$
tentially most informative ty#es of dialogue.
As in all intervie& situations, since the res#ondents &ill react directly in accordance &ith the
intervie&er6s reactions7both verbal and nonverbal7it is vital to ma"e the res#ondent feel at
ease. Kne should carefully avoid #otentially antagonistic or aggressive actions. Moreover, to
avoid any 4intervie& demand characteristics,5 nonverbal cues such as a##roval, disa##roval,
sur#rise or hostility, or im#lying reDection should be avoided. 3ut sim#ly, the intervie&er should
be #erceived as a very interested yet neutral recorder of information.
Laddering Methods
Eliciting Distinctions. 1addering #robes begin &ith distinctions made by the individual re$
s#ondent concerning #erceived, meaningful differences bet&een brands of #roducts. Gaving
made a distinction the intervie&er first ma"es sure it is bi#olar, re;uiring the res#ondent to
s#ecify each #ole. The res#ondent is then as"ed &hich #ole of the distinction is #referred. The
#referred #ole then serves as the basis for as"ing some version of the 4@hy is that im#ortant to
youA5 ;uestion. The follo&ing overvie& identifies three general methods of eliciting distinctions
that have #roven satisfactory. The intervie& outline generally includes at least t&o distinct
methods of eliciting distinctions to ma"e sure no "ey element is overloo"ed.
1. Triadic Sorting )Felly,1..+.
3roviding the res#ondent &ith sets of three #roducts as in the Re#ertory Grid #rocedure is one
&ay to elicit res#onses from a res#ondent. Follo&ing are instructions for a &ine cooler study
&hich used triads to elicit initial distinctions.
2nstructions for Triads
Nou &ill be #resented &ith five grou#s of three different &ine coolers. For each grou# of
three you &ill have the o##ortunity to tell me ho& you thin" about the differences among the
coolers. For e9am#le, if you &ere given a grou# of three carsE
1incoln Bontinental7
Mustang7Badillac
you might say 4car ma"er5 as a &ay of thin"ing about them. T&o are made by Ford and one is
made by General Motors. Another &ay to thin" about them is si/e7big versus small. Kf
course, there are many different &ays that you could thin" about the cars, for e9am#leE
O high styling versus ordinary styling
O economy versus lu9ury
O s#orty versus traditional
There are no right or &rong ans&ers. As 2 #resent you &ith each grou#, ta"e a moment to
thin" about the three &ine coolers.
(#ecifically, 2 &ant you to tell me some im#ortant &ay in &hich t&o of the three &ine
coolers mentioned are the same and thereby different from the third. Again, &hen 2 sho& you
the names of the three &ine coolers, thin" of some overall &ay in &hich t&o of the coolers are
the same and yet different from the third. 2f your res#onse for one grou# of &ine coolers is the
same as for a #revious grou#, try to thin" of another &ay in &hich they differ.
2. Preference-Consumtion !ifferences.
3reference differences can also be a useful device for eliciting distinctions. Res#ondents, after
#roviding a #reference order for, say, brands of coolers, might be as"ed to tell &hy they #refer
their most #referred brand to their second most #referred brand, or more sim#ly to say &hy one
#articular brand is their most #referred )or second most #referred, least #referred. etc.+ brand.
To illustrateE
Nou said your most #referred brand is Balifornia Booler and your second most &as Cartles and
Jaymes. @hat is it, s#ecifically, that ma"es Balifornia Booler more desirableA
Along these same lines, one might as" about #reference and usage and ;uery instances &here
li"ed brands are used infre;uently or less &ell$li"ed brands are used more fre;uently. This device
&or"ed &ell in a #ro#rietary study of snac" chi#s. Lifferences bet&een &hat #eo#le li"e and
&hat they actually used o#ened u# the discussion to include strategies to limit or control the
consum#tion of snac"s.
3. !ifferences "y #ccasion.
2n most cases it is desirable to #resent the res#ondent &ith a #ersonally meaningful conte9t
&ithin &hich to ma"e the distinctions. This contributes to more im#ortant distinctions being
elicited as res#ondents6 distinctions are being e9amined in the conte9t of the setting in &hich
they naturally occur )Car"er, 1<!> Run"el and McGrath, 1=8+. Attention to the conte9t of
consumer behavior #rovides a more meaningful conte9t for laddering to #roceed. 3eo#le do not
use or consume #roducts in general> they do so in #articular conte9ts. A study done in the
convenience restaurant category )Gutman, Reynolds, and Fiedler, 1!?+ used triads bet&een
various convenience restaurants as a starting #oint. 2t &as soon discovered that the distinctions
elicited re#resented such obvious #hysical characteristics of the #laces com#ared )namely,
hamburgers versus chic"en+ that they did not #ermit movement to higher, more #ersonally
meaningful areas from this starting #oint.
Res#ondents &ere then ;uestioned about their usage of various convenience restaurants and
the occasion )day$#art, &ho &ith, concomitant activities+ in &hich they fre;uented them. Jsing
this information to #rovide a relevant conte9t relating to fre;uent usage of the category, re$
s#ondents &ere given the same triads but &ith a conte9t for ma"ing a com#arison. For e9am#le,
it might be suggested to a mother &ith young children that she has been out sho##ing &ith her
children, and it being lunch time, she &ants to sto# for lunch on the &ay home. Three conve$
nience restaurants could be com#ared for their suitability &ith res#ect to this usage situation. Re$
s#ondents could res#ond to triads using their t&o or three most fre;uent usage occasions as a
conte9t for res#onding.
@hat is im#ortant is to #rovide a meaningful basis for the res#ondent to "ee# in mind &hen
thin"ing about differences among the stimuli. 2n this manner their distinctions are more li"ely to
lead to a meaningful consideration of outcomes accruing to the res#ondent, &hich relate to
ma"ing distinctions among the #roducts.
Selecting Ke Distinctions to 1adder. Ty#ically, a res#ondent can only mention 1, to 18 dif$
ferent distinctions for a given #roduct category. Knce a satisfactory number of distinctions have
been mentioned, the intervie&er has basically t&o o#tions on ho& to select &hich ones &ill serve
as the basis for building ladders. 'ither the intervie&er can Dudgmentally select &hich
distinctions are to be used on the basis of #rior "no&ledge of the category or &ith res#ect to the
s#ecific research issue at hand. Kr, the intervie&er can #resent a card &ith all the mentioned dis$
tinctions on it and have the res#ondent rate the relative im#ortance of each, then select those &ith
the highest ratings.
!he !wo "asic Problems o# Laddering. 3rior to the detailing of the s#ecific intervie&ing
techni;ues, t&o of the most common #roblems encountered in laddering and the general ty#e of
tactics re;uired to counter the situation &ill be revie&ed. An understanding of these basic issues
&ill #rovide a necessary basis for learning the more detailed techni;ues to be #resented later in
the article.
1. The $esondent $ea%%y !oes &ot 'no() the *ns(er. @hen as"ed &hy a #articular
attribute or conse;uence is im#ortant to them, the res#ondent often cannot articulate a 4ready5
reason. This lac" of #revious thin"ing of the reason underlying &hy the lo&er level construct is
im#ortant can be dealt &ith by as"ing &hat &ould ha##en if the attribute or conse;uence &as not
delivered. 'ssentially this is negative laddering. The 4nonconscious5 reason )#referred in the
Mean$'nd a##roach to the #sychoanalytic 4subconscious5+ is then ty#ically discovered by the
res#ondent imagining the negative, resulting from the absence of the given construct, and then
relating that bac" to &hat must be delivered if that negative is to be avoided.
Another general class of #robing to avoid bloc"s on the #art of the res#ondent is to change or
re#hrase the ;uestion in a situational conte9t, much li"e the more concrete method illustrated
earlier for initially eliciting distinctions. Cy discussing the issue in this manner, an ans&er is
ty#ically 4discovered5 due to the ability to concreti/e the issue at hand and deal &ith s#ecific
circumstances.
2. +ssues That ,ecome Too Sensiti-e. As the res#ondent is ta"en through the laddering #ro$
cess, that is, moved u#&ard through the levels of abstraction, the dynamics of the intervie&
become more and more #ersonal. Reaction to the continued #robing 4@hy is that im#ortant to
youA5 ;uestion about sensitive issues can vary from 4&affling5 )redefining the ;uestion at an
e;ual or lo&er level+ to stating 42 don6t "no&,5 silence, or even formulating e9traneous argu$
ments as an attem#t to tal" around the issue. Also, the res#ondent can manifest avoidance
behavior by attaching negative or adverse characteristics to the intervie&ing #rocess or to the
intervie&er.
Casically, three techni;ues can be em#loyed to deal &ith res#ondent bloc"s due to sensitive
issues. The first involves moving the conversation into a third #erson format, creating a role$
#laying e9ercise. The second, and most dangerous o#tion, is for the intervie&er to reveal a
relevant #ersonal fact )ty#ically fabricated+ about him/herself that ma"es the res#ondent feel less
inhibited by com#arison. The third, and most common, is to ma"e a note of the #roblem area and
come bac" to the issue &hen other relevant information is uncovered later in the intervie&.
Techni;ues. 'ach of the follo&ing techni;ues &ill be illustrated by using one common #roduct
class, &ine coolers, for #ur#oses of sim#licity. A short definition of each techni;ue &ill be
#resented. Then verbatim transcri#tions are sho&n to give a more com#lete e9am#le of the
laddering #rocess. (ummary ladders are detailed to illustrate the content classification by level of
abstraction )A/B/0+. Mote that each ladder is contained &ithin the G0M de#icted in Figure 1.
1. .-o/ing the Situationa% Conte0t )P+. 1addering &or"s beet &hen res#ondents are #roviding
associations &hile thin"ing of a realistic occasion in &hich they &ould use the #roduct. 2t is the
#erson that is the focus of study, not the #roduct. Therefore, it is essential to elicit from
res#ondents the most relevant occasions for #roduct consum#tion and to use these as the focus of
the intervie&.
InterviewerE Nou indicated that you &ould be more li"ely to drin" a &ine cooler at a #arty on
the &ee"end &ith friends, &hy is thatA
$es%ondent& @ell, &ine coolers have %ess a%coho% than a mi9ed drin" and because they are so
fi%%ing 2 tend to drin" fe&er and more slo&ly.
Interviewer& @hat is the benefit of having less alcohol &hen you are around your friendsA
$es%ondent& 2 never really have thought about it. 2 don6t "no&.
InterviewerE Try to thin" about it in relation to the #arty situation. '() @hen &as the last time
you had a &ine cooler in this #arty &ith friends situationA
$es%ondentE 1ast &ee"end. 2ntervie&erE K"ay, &hy coolers last &ee"endA Res#ondentE @ell,
2 "ne& 2 &ould be drin"ing a long time and + didn1t (ant to get (asted.
InterviewerE @hy &as it im#ortant to not get &asted at the #arty last &ee"endA Res#ondentE
@hen 26m at a #arty 2 li"e to socia%i2e3 tal" to my friends, and ho#efully ma"e some ne& friends.
2f 2 get &asted 26m afraid 26d ma"e an ass of myself and #eo#le &on6t invite me ne9t time. 2t6s
im#ortant for me to be art of the grou.
The summary ladder for )1+ isE
0 sense of belonging )#art of the grou#+
B sociali/e
B avoid getting drun" )&asted+
A less alcohol/filling
2. Postu%ating the *"sence of an #"4ect or a State of ,eing '().
Kne &ay of 4unbloc"ing5 res#ondents &hen they cannot move beyond a certain level is to
encourage them to consider &hat it &ould be li"e to lac" an obDect or to nut feel a certain &ay.
This device often enables res#ondents to verbali/e meaningful associations.
Interviewer& Nou said you #refer a cooler &hen you get home after &or" because of the fu%%-
"odied taste. @hat6s so good about a full$bodied taste after &or"A
$es%ondent& 2 Dust li"e it. 2 &or"ed hard and it feels good to drin" something satisfying.
2ntervie&erE @hy is a satisfying drin" im#ortant to you after &or"A
$es%ondent& Cecause it is. 2 Dust enDoy it.
InterviewerE @hat &ould you drin" if you didn6t have a cooler available to youA )P+
$es%ondentE 3robably a light beer.
Interviewer& @hat6s better about a &ine cooler as o##osed to a light beer &hen you get home
after &or"A
$es%ondent& @ell, if 2 start drin"ing beer, 2 have a hard time sto##ing. 2 Dust continue on into
the night. Cut &ith coolers 2 get fi%%ed u and it6s easy to sto. 3lus, 2 tend to not eat as much
dinner. 2ntervie&erE (o &hy is continuing to drin" into the evening something you don6t &ant to
doA
$es%ondentE @ell, if 2 "ee# drin"ing 2 generally fa%% as%ee #retty early and 2 don6t get a chance
to ta%/ to my (ife after the "ids go to bed. (he &or"s hard &ith the house and the "ids all day7
and it6s really im#ortant that 2 tal" to her so &e can "ee# our good relationshi#, our fami%y %ife3
going.
The summary ladder for )8+ isE
0 good family life
B able to tal" to my &ife
B don6t fall aslee#
B )consume less alcohol+
A filled u#/easy to sto#
A full$bodied taste/ less alcohol
3. &egati-e 5addering '().
For the most #art, the laddering #rocedure #roceeds by #robing the things res#ondents do and
the &ay res#ondents feel. Go&ever, much can be learned by in;uiring into the reasons &hy
res#ondents do not do certain things or do not &ant to feel certain &ays. This techni;ue is
#articularly relevant &hen res#ondents cannot articulate &hy they do the things they do.
'9#loring hidden assum#tions in this manner and using the device of ma"ing the o##osite
assum#tion have #roven to be useful devices in ma"ing res#ondents a&are of im#lications of
common behaviors )Lavis, 1=1+.
InterviewerENou indicated a distinction bet&een 18 ounce and 1< ounce bottles. @hat si/e
bottle do you #referA
$es%ondentE 2 al&ays buy a 18 ounce bottle.
InterviewerE@hat6s the benefit of buying a 18 ounce bottleA $es%ondentE 2 Dust buy it out of
habit.
InterviewerE@hy &ouldn6t you buy a 1< ounceA )P+
$es%ondentE 2t6s too much for me to drin/ and it gets (arm before 2 can finish it all.
Then 2 have to thro( it a(ay.
InterviewerE (o ho& do you feel &hen you have to thro& it a&ayA
$es%ondentE 2t ma"es me mad because 26m (asting my money. InterviewerE @hat6s the im$
#ortance of money to youA Res#ondentE 26m in charge of the family budget, so it6s my
resonsi"i%ity to ma"e sure it6s s#ent right.
The summary ladder for )*+ isE
0 res#onsibility to family
B &aste money
B thro& it a&ay )don6t drin" all of it+
B gets &arm
B too much to drin"
A larger si/e
6. *ge-$egression Contrast Pro"e 789.
Moving res#ondents bac"&ard in time is another effective device for encouraging res#ondents
to thin" critically about and be able to verbali/e their feelings and behavior.
InterviewerENou said you most often drin" coolers at the bar. @hy is thatA
$es%ondentE 26ve never really thought about it. 2 Dust order them.
InterviewerE2s there a difference in your drin"ing habits com#ared to a cou#le of years agoA
)P+
$es%ondentE Nes, 2 drin" different ty#es of drin"s no&.
InterviewerE@hy is thatA Res#ondentE @ell, before 2 used to be in college, and the only thing
around seemed to be beer.
InterviewerE(o &hy do you drin" coolers no&A Res#ondentE @ell, no& 2 have a career and
&hen 2 do go out 2 go &ith co&or"ers. Lrin"ing a &ine cooler loo"s better than drin"ing a
beer.
InterviewerE@hy is thatA Res#ondentE The "ott%e shae and the fancy %a"e% loo" more
feminine than drin"ing a beer.
InterviewerE @hy is that im#ortant to youA
$es%ondentE 2t6s im#ortant to me to have a sohisticated image no& that 26m in the
&or" force. 2 &ant to be Dust %i/e my co(or/ers.
The summary ladder for )?+ isE
0 li"e my co&or"ers )belonging+
B so#histicated image
B more feminine
A bottle sha#e
A fancy label
:. Third-erson Pro"e )P+.
Another device for eliciting res#onses from res#ondents &hen they find it difficult to identify
their o&n motives or to articulate them is to as" ho& others they "no& might feel in similar
circumstances.
InterviewerENou mentioned you drin" &ine coolers at #arties at your friend6s house. @hy do
you drin" them thereA
$es%ondentE Just because they have them.
InterviewerE@hy not drin" something elseA
$es%ondentE 2 Dust li"e drin"ing coolers.
InterviewerE Why do you thin" your friends have them at #artiesA )P+
$es%ondentE 2 guess they &ant to imress us because &ine coolers are e0ensi-e. They
relate ;uality to ho& e0ensi-e it is.
InterviewerE@hy do they &ant to im#ress othersA
$es%ondentE (ince coolers are ne&, they are almost li"e a status sym"o%.
Interviewer (o &hat is the value to them of having a status symbolA
$es%ondentE My friends al&ays li"e to do one better than anyone else. 2t6s #robably
related to their se%f-esteem.
The summary ladder for ).+ isE
0 self$esteem
B status symbol
B im#ress )others+
B ;uality
A e9#ensive
*. $edirecting Techni;ues< Si%ence 789=Communication Chec/ )P+
(ilence on the #art of the intervie&er can be used to ma"e the res#ondent "ee# trying to loo" for
a more a##ro#riate or definite ans&er &hen either the res#ondent is not &illing to thin" critically
about the ;uestion as"ed or &hen the res#ondent feels uncomfortable &ith &hat he or she is
learning about themselves.
A communication chec" sim#ly refers to re#eating bac" &hat the res#ondent has said and
as"ing for clarification, essentially as"ing for a more #recise e9#ression of the conce#t.
InterviewerE Nou mentioned you li"e the carbonation in a cooler. @hat6s the benefit of itA
$es%ondentE 2 don6t thin" there6s any benefit to carbonation.
InterviewerE @hy do you li"e it in a coolerA
$es%ondentE Mo #articular reason.
InterviewerE )silence+ )P+
$es%ondentE Bome to thin" of it, carbonation ma"es it cris and refreshing.
InterviewerE @hy is that im#ortantA
$es%ondentE 2t ma"es it thirst ;uenching3 es#ecially after mo&ing the la&n and is a #ic"$
me$u#.
InterviewerE 1et me see if 2 understand &hat you6re saying. '(() @hat do you mean by
saying a #ic"$me$u#A
$es%ondentE 2 mean after 2 finish it6s li"e a re(ard for com%eting a chore 2 disli"e.
The summary ladder for )<+ isE
0 com#leting a chore )accom#lishment+
B re&ard
B thirst$;uenching
B refreshing
A cris#
A carbonation
S+mmar. The reader &ill no doubt notice the similarity of these techni;ues to other ;ualita$
tive intervie&ing a##roaches. The #ur#ose here has been to demonstrate their use in laddering
and to sho& ho& the ladders er se emerge from the intervie&er$res#ondent interaction.
After s#ending a fair amount of time on one ladder &ithout closure to a higher level, it be$
comes necessary to either terminate further discussion or #roceed on to another ladder and circle
bac" later. 2f one attribute or conse;uence ceases to become mobile, it is of no benefit to con$
tinue the laddering #rocess &ith it because time is limited. The more familiar the intervie&er be$
comes &ith the techni;ues and #rocedures, the better the intervie&er is able to Dudge if an out$
come can be reached in the line of ;uestioning. Cy moving on to another subDect, the res#ondent
is given time to thin" more about the issue. The res#ondent may have a bloc" and the shift can
sometimes resolve the #roblem.
The central idea is to "ee# the focus of the discussion on the #erson rather than on the #roduct
or service. This is not an easy tas" because ty#ically at some #oint the res#ondent reali/es that
the #roduct seems to have disa##eared from the conversation. Jnfortunately, there are situations
&here techni;ues and #rocedures are unable to #roduce a means$ends chain. The res#ondent may
be inarticulate or sim#ly un&illing to ans&er. 2t also ta"es a length of time for the intervie&er to
test all the techni;ues and develo# a #ersonal style that can #roduce ladders. As &ith any
;ualitative techni;ue e9#erience becomes the "ey.
Ty#ically, t&o or three ladders can be obtained from roughly three$fourths of the res#ondents
intervie&ed. A##ro9imately one$fourth of the res#ondents, de#ending on the level of involve$
ment in the #roduct class, cannot go beyond one ladder. The time re;uired from distinctions to
final ladders varies substantially, of course, but <, to =. minutes re#resents a ty#ical standard.
,nalsis
Content ,nalsis. As over$vie&ed earlier, the initial tas" of the analysis is to content$analy/e
all of the elements from the ladders. The first ste# is to record the entire set of ladders across
res#ondents on a se#arate coding form. Gaving ins#ected them for com#leteness and having
develo#ed an overall sense of the ty#es of elements elicited, the ne9t ste# is to develo# a set of
summary codes that reflect everything that &as mentioned. This is done by first classifying all
res#onses into the three basic A/B/0 levels and then further brea"ing do&n all res#onses into
individual summary codes )see Table 2 for &ine$cooler codes+.
Kbviously, one &ants to achieve broad enough categories of meaning to get re#lications of
more than one res#ondent saying one element leads to another. Net, if the coding is too broad, too
much meaning is lost. The "ey to #roducing consistency in this stage, as in all content analysis, is
reliability chec"s across multi#le coders.
2m#ortantly, the goal at this level of the analysis is to focus on meanings central to the #ur#ose
of the study, remembering that it is the relationshi#s bet&een the elements that are the focus of
interest, not the elements themselves. For e9am#le, 4avoids the negatives of alcohol5 in Figure 1
is a summari/ation of several more detailed elements )namely, not too fired, not too drun", don6t
say dumb things, and don6t get numb+. 2f all those se#arate elements &ere given se#arate codes it
is li"ely6 that none of the relations bet&een them and other elements &ould have very high
fre;uencies, and they &ould not a##ear in the G0M.
Knce the master codes are finali/ed, numbers are assigned to each. These numbers are then
used to score each element in each ladder #roducing a matri9 &ith ro&s re#resenting an indi$
vidual res#ondent6s ladder )one res#ondent can have multi#le ladders and thus multi#le ro&s+,
&ith the se;uential elements &ithin the ladder corres#onding to the consecutive column desig$
nations. Thus the number of columns in the matri9 corres#onds to the number of elements in the
longest ladder #lus any identification or demogra#hic codes. )(ee the A##endi9 for the
hy#othetical score matri9 re#resenting one ladder for <= res#ondents from &hich the G0M in
Figure 1 &as constructed.+
2t is this 4crossing over5 from the ;ualitative nature of the intervie&s to the ;uantitative &ay of
dealing &ith the information obtained that is one of the uni;ue as#ects of laddering and clearly
the one that sets it a#art from other ;ualitative methods. This summary score matri9, then, serves
as the basis for determining the dominant #ath&ays or connections bet&een the "ey elements as
&ell as #roviding the ability to summari/e by subgrou# )e.g., men only+.
!able -
S+mmar Content Codes #or .%othetical /ine Cooler E0am%le
0alues
)8,+ Accom#lishment
)81+ Family
)88+ Celonging
)8*+ (elf$esteem
Bonse;uences
!+ Quality
+ Filling
)1,+ Refreshing
)11+ Bonsume less
)18+ Thirst$;uenching
)1*+ More feminine
)1?+ Avoid negatives
)1.+ Avoid &aste
)1<+ Re&ard
)1=+ (o#histicated
)1!+ 2m#ress others
)1+ (ociali/e
Attributes
1+ Barbonation
8+ Bris#
*+ '9#ensive
?+ 1abel
.+ Cottle sha#e
<+ 1ess alcohol
=+ (maller
!he Im%lication Matri0. T&o research issues remain> constructing hierarchical ma#s to re#re$
sent res#ondents6 ladders in the aggregate> and determining the dominant #erce#tual segments
re#resented in the overall ma# of aggregate relations. To accom#lish this, the ne9t ste# is the
straightfor&ard one of constructing a matri9 &hich dis#lays the number of times each element
leads to each other element )o#erationally defined at this level as &hich elements in a given ro&
#recede other elements in the same ro&+. (uch a matri9 &ill be a s;uare matri9 &ith a si/e
reflecting the number of elements one is trying to ma#, usually bet&een *, and .,. T&o ty#es of
relations may be re#resented in this matri9E direct relations and indirect relations.
.%othetical .ierarchical 1al+e Ma% o# /ine Cooler Bategory
(elf$esteem 8* Family 1ife 81
O feel better O maintain res#ect
about self of others
O self 2mage O better family ties
O self &orth R S
R Celonging 88 R S
R O security R S
R O camaraderie R S
R O friendshi# R S
Accom#lishment 8, R / S R S
O get most from life R / S R S
R Impress Others 18 Socialize 19 S
R O successful image )able to+ S
R / S O easier to tal" S
R / S O o#en u# S
R / S O more sociable S
Reward 16 SophistIcated Image 17 | S
O satisfying O #ersonal status R S
O com#ensation O ho& others vie& me R S
/ S / R Avoid Negatives S
/ / !ore "emi#i#e 1$ of Alcohol 1% Avoid &aste 1'
/ / O socially O not too drun" O doesn6t get
(hirst)*+e#chi#g 1, / acce#table O not too tired &arm
O relieves thirst S / R S R S
O not too sour S / R S R S
/ S / R S R S
/ S / R S R S
Re-reshi#g 1. /+alit0 8 | | 1o#s+me less 11
O feel alert, O su#erior #roduct R S R O can6t drin" more
alive 2 #roduct ;uality R S R O can si# S
/ S / S R S R R R
/ S / S R S R R R
/ S / S R S R R R
/ S / S 1abel Cottle 1ess R (maller (i/e
Barbonation Bris# '9#ensive )fancy+ )sha#e+ Alcohol Filling )1, o/.+
)T+ 1 8
)T+ * ? . < =
Lirect relations refer to im#licative relations among adDacent elements. The designations of
)A+ through )'+ for the elements refer sim#ly to the se;uential order &ithin the ladder. That is,
given our &ine cooler e9am#leE
Celonging )'+
able to sociali/e )L+
avoid negatives of alcohol )B+
consume less )C+
filling )A+
The A$C )4filling7consume less5+ relation is a direct one as is C$B, B$L, and L$'. Go&ever,
&ithin any given ladder there are many more indirect relations, A$B, A$L, A$', C$L, and so
forth. 2t is useful to e9amine both ty#es of relations in determining &hat #aths are dominant in an
aggregate ma# of relationshi#s among elements. @ithout e9amining indirect relations, a situation
might e9ist &here there are many #aths by &hich t&o elements may be indirectly connected but
&here none of the #aths are re#resented enough times to re#resent a significant connection. For
e9am#le, there may be other #aths by &hich 4avoids negatives of alcohol66 leads to 4belonging.5
Mevertheless, it is hel#ful to "ee# trac" of the number of times 4avoids negatives of alcohol5
ultimately leads to 4belonging5 &hen e9amining the strength of ladders as derived from the
aggregate matri9 of relations.
Another o#tion in constructing the overall matri9 of relations among elements is &hether to
count each mention of a relationshi# among elements that an individual res#ondent ma"es or to
count a relation only once for each res#ondent, no matter ho& many times each res#ondent
mentions it. Given the #revious ladder as an e9am#le, if 4filling 7consumes less5 leads to
several higher level associations for a given individual, do you count that indirect relation as
many times as it occurs, or Dust once #er res#ondentA The significance of an element is in #art a
function of the number of connections it has &ith other elements, &hich argues for counting all
mentions, but it does distort the construction of the ma# &here there are sur#risingly fe& )to
those not familiar &ith this research+ connections bet&een elements in the overall matri9. Kften,
of all the cells having any relations, only one$half &ill be mentioned by as many as three
res#ondents.
Table 8 #resents the ro&$column fre;uency matri9 indicating the number of times directly and
indirectly all ro& elements lead to all column elements. The numbers are e9#ressed in fractional
form &ith direct relations to the left of the decimal and indirect relations to the right of the
decimal. Thus 4carbonation5 )element 1+ leads to 4thirst$;uenching5 )element 18+ four times
directly and si9 times indirectly. More #recisely, this means that four res#ondents said
carbonation directly leads to thirst$;uenching, &hereas t&o res#ondents se;uentially related the
t&o elements &ith another element in bet&een.
Constr+cting the .ierarchical 1al+e Ma%. In filling in the im#lication matri9, individual re$
s#ondent6s ladders are decom#osed into their direct and indirect com#onents )see Table 8+. 2n
constructing the G0M, 4chains5 have to be reconstructed from the aggregate data. To avoid
confusion, the term 4ladders5 &ill refer to the elicitations from individual res#ondents> the term
4chains5 &ill be used in reference to se;uences of elements &hich emerge from the aggregate
im#lication matri9.
To construct a G0M from the matri9 of aggregate relations, one begins by considering
adDacent relations, that is, if A 7U C and C7U B and B 7U L, then a chain A$C$B$L is formed.
There doesn6t necessarily have to be an individual &ith an A$C$B$0 ladder for an A$C$B$L chain
to emerge from the analysis. A G0M is gradually built u# by connecting all the chains that are
formed by considering the lin"ages in the large matri9 of relations among elements.
The most ty#ical a##roach is to try to ma# all relations above several different cutoff levels
)usually from * to . relations, given a sam#le of ., to <, individuals+. The use of multi#le
cutoffs #ermits the researcher to evaluate several solutions, choosing the one that a##earE> to be
the most informative and most stable set of relations. 2t is ty#ical that a cutoff of ? relations &ith
., res#ondents and 18. ladders &ill account for as many as t&o$thirds of all relations among ele$
ments. 2ndeed, the number of relations ma##ed in relation to the number of relations in the
s;uare
!able 3
S+mmar Im%lication Matri0(
! 1, 11 18 1* 1? 1. 1< 1= 1! 1 8, 81 88 8*
1 Barbonation 1.,, 1,.,, ?.,< .,1 .1? .,? .,< .,? 1
8 Bris# *.,, ?.,, .,? .,? .,* .,? .,1 .,= 8
* '9#ensive 18.,, 8.,? 1.,1 1., 1.,< .,. .,. *
? 1abel 8.,, 8.,8 8.,? .,8 .,1 .,8 .,* ?
. Cottle sha#e 1.,, 1.,, 8.,8 1.,* .,8 .,* .
< 1ess alcohol 1.,, 1.,, ..,, .,1 .,1 1.,1 .,? .,1 <
= (maller 1.,, .,1 *.,, .,1 .,8 .,1 =
! Quality *,, 1.,, ?.,, ?.,* ?.,? .,1 *.,8 ., .,? !
Filling ?.,, .,? 1.,* .,* .,8
1, Refreshing 1,.,, 1.,, ..1, .,1 .,< .,? .,. .,8 1,
11 Bonsume less ..,, .,? .,8 .,* 11
18 Thirst$;uenching 1?.,, .,! .,< .,? .,? 18
1* More feminine =.,, .,8 1.,* .,? 1*
1? Avoid negative 1.,, ..,, ?.,1 .,? 1?
1. Avoid &aste 8.,, 1.
1< Re&ard 11.,, !.,, .,< 1.,. 1<
1= (o#histicated ?.,, 1.,, 1.,, ?.,8 ..,* 1=
1! 2m#ress 1.,, 1,.,, .,, 1!
1 (ociali/e *.,, ..,, 1
8, Accom#lishment 8,
81 Family 81
88 Celonging 88
8* (elf$esteem 8*
P Mo relations e9ist bet&een the attribute elements.
im#lication matri9 can be used as an inde9 of the ability of the ma# to e9#ress the aggregate
relationshi#s. There are )naturally enough+ a tremendous number of em#ty cells and ;uite a fe&
relations &hich are mentioned only once. Again, in establishing a cutoff level, one may count
only the direct lin"ages in any cell or one may count the total number of lin"ages, direct or
indirect,
To actually construct a G0M from the series of connected #airs, one must literally build u# the
ma# from the chains e9tracted from the matri9 of im#licative relations. Bonsiderable ingenuity is
needed for this tas", &ith the only guideline being that one should try at all costs to avoid
crossing lines. This disci#line #rovides a coherence to the ma# and adds considerably to its
inter#retability. The criteria for evaluating the ability of the overall ma# to re#resent the data is to
assess the #ercentage of all relations among elements accounted for by the ma##ed elements. The
reader &ill note that Figure 1 also contains both the significant direct and indirect relations
among adDacent elements.
Cefore constructing the G0M from the data in Table 8, it is necessary to #oint out the ty#es of
relations &hich might e9ist among elements. Five ty#es of relations are of noteE
A$L 'lements ma##ed as ad4acent &hich have a high number of direct relations.
M$L 'lements ma##ed as nonad4acent &hich have a high number of direct relations.
A$2 *d4acent elements &hich have a high number of indirect relations but a lo&
number of direct relations.
M$2 &onad4acent elements &hich have a lo&, non$/ero number of direct relations but a
high number of indirect relations.
M$K &onad4acent elements &hich have a lo& )or 2ero9 number of indirect relations.
An illustration of these five ty#es &ill hel# ma"e clear the consideration #rocess re;uired in the
construction of the ma#.
The first ty#e of relationshi#, A$L, is the most common and re#resents the standard basis
ty#ically used in constructing the ma#. Go&ever, even &hen only the strong #air&ise lin"ages
are summari/ed, a certain degree of sim#lification can be gained from folding in consistent
elements. For e9am#le, 1, res#ondents directly associated 4carbonation5 )1+ &ith 4refreshing5
)1,+ #roducing a strong lin"age. And, 4carbonation5 )1+ and 4thirst7 ;uenching5 )18+ have four
direct relations and si9 indirect relations #roducing a se#arate yet related lin"age. 2n this case,
one o#tion &ould be to ma# t&o lines, 1$1, and 1$18. Another o#tion &hich #ermits essentially
the same inter#retation is to ma# 1$1,$18 in &hich both are embedded. 2n effect the
4carbonation$thirst$;uenching5 )1$18+ relation is a 4M$L5 ty#e as described above, because these
elements are ma##ed nonadDacently even though they have a high number of direct relations.
The #ossibility e9ists that some relations &ould not be considered to be #ositioned adDacently
because of a lo& number of direct relations, yet because of a high number of indirect relations
this #ositioning a##ears reasonable )A$2+. To illustrate, 4fancy label5 )?+ and 4bottle sha#e5 ).+
are each lin"ed directly to 4more feminine5 )1*+ t&ice, &hich is belo& the cutoff value chosen to
construct the G0M. Go&ever, both elements have t&o indirect relations &ith 4more feminine5 in
addition to their t&o direct relations. 2t &ould seem reasonable to #osition both elements adDa$
cently to 4more feminine,5 omitting the element or elements &hich come bet&een them and
::more feminine.66 2n the case &here there are a number of diffuse #aths bet&een t&o elements
such that no #ath is dominant, as &as rather sim#ly demonstrated here, it is often useful to omit
the minor relations and Dust ma# the dominant #ath.
2f a chain is re#resentative of several individuals6 ladders, the elements in that chain &ill be
characteri/ed by a high number of indirect relations among nonadDacent relations7although
such nonadDacent elements &ill not necessarily have any direct relations )the 4M$25 relation+.
This is the ty#e of relationshi# &hich characteri/es a Guttman scale. For e9am#le, 4re&ard5 )1<+
leads to 4self$esteem5 )8*+ one time directly, but five times indirectly. 2f 4re&ard5 did not ulti$
mately lead to 4self$esteem,5 even though it does lead to 4im#ress others5 )1!+, and 4im#ress
others5 leads to 4self$esteem,5 &e &ould certainly not characteri/e the 4re&ard$im#ress others$
self$esteem5 chain )1<$1!$8*+ Vs a strong one. Thus, the 4M$25 relations, even though they are
not #lotted, are im#ortant determinants of the ;uality of the chains de#icted in the
G0M.
The last category of relations, nonadDacent relations &hich have lo& or no indirect or direct
relations )M$K+, deserves careful consideration because of an artifact in the &ay the G0M is con$
structed. As an e9am#le, 4cris#5 )8+ does not a##ear in any res#ondent6s ladder &ith either
4accom#lishment5 )8,+ or 4self$
esteem5 )8*+> ho&ever, it does have seven indirect lin"ages &ith 4belonging5 )88+. The common
as#ects of the 4carbonation5 )1+ #ath and the 4cris#5 #ath account for the G0M being dra&n in
this manner.
2n constructing the G0M in Figure 1 from the data in Table 8, the most efficient &ay is Vo start
in the first ro& for &hich there is a value at or above the arbitrary cutoff level you have chosen.
Jsing a cutoff of ?, the first significant value is 4carbonation7 refreshing5 )1, 1,+ &ith a value
of 1,.,, indicating 1, direct relations and , indirect relations bet&een these t&o elements. Me9t,
one &ould move to the tenth ro& to find the first value at or e9ceeding the cutoff value. 2t can be
seen in Table 8 that 4thirst ;uenching5 )column 18+ is the first significant value. Thus, the chain
has gro&n to 41$1,$18.5 Bontinuing in the same manner the chain &ould ne9t e9tend to 4re&ard5
)1$1,$18$1<+, then to include 4im#ress others5 )1$1,$18$1<$1!+, and, lastly, to include
4belonging5 )1$1,$18$1<$1!$88+.
Gaving reached the end of the chain, one goes bac" to the beginning to see if there are other
significant relations in the same ro&s of the matri9 &hich already have been ins#ected. For e9$
am#le, ins#ecting the first ro& indicates that 4carbonation5 is connected to 4thirst$;uenching,5
4re&ard,5 and 4im#ress others57all elements &hich are already included in the chain. 2n addi$
tion, 4carbonation5 is lin"ed to 4accom#lishment5 and 4self7esteem5 )8, and 8*+. A similar #at$
tern &ill be observed &hen lin"s &ith 4thirst$;uenching5 )18+ are ins#ected.
Go&ever, &hen 4re&ard5 )1<+ is ins#ected, it should be noted that moving across to column 8,
in ro& 1<, another significant relation is found. Thus another chain &ith common lin"s to the
original chain is #lotted )1$1,$18$1<$8,+. And, 4im#ress others5 )1!+ also is lin"ed to 4self$es$
teem5 )8*+, #roducing the family of chains sho&n belo&E
self$esteem )8*+
R
accom#lishment )8,+ R
R R
R im#ress others )1!+
R /
R /
re&ard )1<+
R
thirst$;uenching )18+
R
refreshing )1,+
R
carbonation )1+
The ne9t ste# is to move to the second ro& and start the #rocess over again. 2t &ill be seen that
4cris#5 has one set of connections &hich are identical to 4carbonation5 and thus could be #lotted
)and is so #lotted in Figure 1+ ne9t to 4carbonation.5 4Bris#5 also has connections to 4;uality5
)!+, and thus a ne& chain is started. 2t can be seen by ins#ecting Table 8 that 4e9#ensive5 )*+ has
18 direct connections &ith 4;uality.5 (tarting &ith a 4*$!5 chain, 4;uality5 )!+ is connected to
4re&ard5 )1<+ four times, so &e can include a line bet&een 4;uality5 and 4re&ard,5 thus yielding
a 4*$!$1<5 chain. 4Quality5 also leads to 4so#histicated image5 )1=+ four times directly and four
times indirectly for a total of eight connections> therefore, &e can connect these t&o elements in
the G0M. 2n scanning ro& 1= of Table 8 it can be seem that 4so#histicated image5 has 11 direct
lin"ages &ith 4im#ress others,5 so that these t&o elements can be connected in the G0M.
2n a similar fashion, 4fancy label5 and 4bottle sha#e5 )? and .+ have t&o direct and t&o direct
lin"ages &ith 4more feminine5 )1*+, and that 4more feminine5 has seven direct lin"ages &ith
4so#histicated image5 )1=+. '9amination of ro&s <, =, , 11, and 1? )less alcohol, smaller si/e,
filling, consume less, and avoid negatives of alcohol+ have lin"ages only &ith 4able to sociali/e5
)element 1+. Thus in Figure 1, it is only 4able to sociali/e5 that lin"s u# &ith any elements on
the left side of the G0M. 2t is only at the values level, 4belonging,5 that the right side of the ma#
is connected to the elements of the left side.
The goal of ma##ing these hierarchical relations is to interconnect all the meaningful chains in
a ma# in &hich all relations are #lotted &ith no crossing lines )&hich in almost all studies is
#ossible+. This results in a ma# &hich includes all relevant relations and yet is easy to read and
inter#ret. The G0M in Figure 1 accounts for ?.. #ercent of all the direct and indirect relations
contained in the <= ladders from &hich it &as develo#ed.
Gaving #lotted all relations, it is desirable to loo" at all elements in the ma# in terms of the
numbers of direct and indirect relations they have &ith other elements, both in terms of other
elements leading into them and in terms of their connections to higher order elements. Table *
#resents the sums of the direct and indirect relations for each element. For e9am#le, 4belonging5
)88+, at the values level, is the element &hich has the most elements leading to it. Thus, it might
be seen as the core value in terms of im#ortance to the #roduct class. 2n addition, three other
elements are note&orthy for having a high fre;uency of elements leading from them as &ell as
into them, namely, 4re&ard5 )1<+, 4im#ress others5 )1!+, and 4;uality5 )!+. 2ndeed, the ;uality 4
5 re&ard 45 im#ress others 7U belonging chain can be seen to have a high number of relations
among its res#ective elements.
Determining Dominant Perce%t+al Orientations. Once a hierarchical value ma% is con$
structed, one ty#ically considers any #ath&ay from bottom to to# as a #otential chain
re#resenting a #erce#tual orientation. For e9am#le, in Figure 1 the total number of uni;ue
#ath&ays bet&een elements at the attribute level and elements at the values
!able 6
S+mmar o# Direct '77) and Indirect '88) $elations #or Each Element
'77.88)
Code !o 9rom
1 1..*. ,.,,
8 =.8* ,.,,
* 1=.*, ,.,,
? <.1? ,.,,
. ..1, ,.,,
< <.<, ,.,,
= ?.,. ,.,,
! 1.8* .,,
..18 ,.,,
1, 1<.8< 1<.,,
11 .., ..,,
18 1?.88 1..,,
1* <., <.,?
1? 1,.,. 1,.,.
1. 8.,, ?.,1
1< 8,.11 8..**
1= 1..,. 1..1.
1! 8,.,, 81.?,
1 !.,, !.11
8, ,.,, 1?.8.
81 ,.,, .18
88 ,.,, 8,..<
8* ,.,, 1..*=
level is 8*, any or all of &hich &arrant consideration. To more fully understand the strength of
the chains, the intra$chain relations can be summari/ed and evaluated. The #ortions &ithin Table
? demonstrate this #rocess. Table ? includes detailing of the relations for four chains &ithin
Figure 1 in an easier$to$read format than trac"ing them do&n in the ro&$column fre;uency
matri9 in Table 8. 3art A of Table ? sho&s the direct and indirect relations lin"ing 4carbonation5
&ith 4accom#lishment.5 2t can be seen by ins#ection that all elements are lin"ed directly or
indirectly to all other elements in the chain. 4Barbonation5 has si9 indirect lin"ages &ith
4accom#lishment,5 meaning that these t&o elements are included in si9 res#ondents6 ladders.
4Refreshing5
!able :
Partitions o# Chains b $elations
Part , ;Carbonatian4accom%lishment< chain
= 3 -= -3 -* 3= =
8 ,.,, ?.,, ,.,? ,.,? ,.,, ?.,<
1, ,.,, ,.,, 1,.,, ..1, ,.,? 1..1?
18 ,.,, ,.,, ,.,, 1?.,, ,.,< 1?.,<
1< ,.,, ,.,, ,.,, ,.,, !.,, !,,
8, ,.,, ,.,, ,.,, ,.,, ,.,, ,,,
, ,.,, ,.,, ,.,, ,.,, ,.,, ?1.8!
Part " >Carbonation4sel#?esteem@ chain
= - -= -3 -* -A 36 =
1 ,.,, 1,.,, ?.,< ,.1? ,.,? ,.,? 1?.8<
1, ,.,, ,.,, 1,.,, 1..1,. ,.,< ,.,8 1..1!
18 ,.,, ,.,, ,.,, 1?.,, ,.,! ,.,? 1?.18
1< ,.,, ,.,, ,.,, ,.,, 11.,, 1.,. 18.,.
1! ,.,, ,.,, ,.,, ,.,, ,.,, .,, .,,
8* ,.,, ,.,, ,.,, ,.,, ,.,, ,.,, ,.,,
, ,.,, ,.,, ,.,, ,.,, ,.,, ,.,, <?.<*
Part C >Less alcohol4belonging@ chain
= * -: -B 33 =
< ,.,, ..,, 1.,1 ,.,1 <.,8
1? ,.,, ,.,, ..,, ,.,? ..,?
1 ,.,, ,.,, ,.,, ..,, ..,,
88 ,.,, ,.,, ,.,, ,.,, ,.,,
, ,.,, ,.,, ,.,, ,.,, 1<.,<
Part D ;"ottle sha%e4sel#?esteem@ chain
= C -6 -D -A 36 =
. ,.,, 8.,8 1.,* ,.,, ,.,* *.,!
1* ,.,, ,.,, =.,, ,.,8 ,.,? =.,<
1= ,.,, ,.,, ,.,, ?.,, ..,* .,*
1! ,.,, ,.,, ,.,, ,.,, .,, .,,
8* ,.,, ,.,, ,.,, ,.,, ,.,, ,.,,
, ,.,, ,.,, ,.,, ,.,, ,.,, 8!.1=
and 4thirst$;uenching5 have four and si9 indirect lin"ages, res#ectively, and 4re&ard5 has eight
direct lin"ages &ith 4accom#lishment.5 2n all, the chain accounts for .1 direct relations among
elements and ?< indirect relations.
3art C of Table ? sho&s the 4carbonation7 self$esteem5 chain. This chain accounts for more
direct relations than does the chain in 3art A of Table ?. 2t is also longer, having more elements in
it. 2n general, the lin"ages among elements at the bottom of this chain have fe&er lin"ages &ith
the elements at the to# of the chain. 4Refreshing5 has only t&o indirect lin"ages &ith 4self$
esteem.5
2n 3art B of Table ?, a chain is sho&n that has fe&er elements and accounts for far fe&er rela$
tions. 2t can also be seen that 4less alcohol5 is not strongly associated &ith 4sociali/e5 or 4be$
longing.5 (uch a &ea"ness, as indicated by the lac" of associations res#ondents are ma"ing be$
t&een these elements, might re#resent an o##ortunity for a cam#aign to strengthen this tie )in the
beer category this indeed is &hat the 1.A. brand has done in its advertising in the lo&$alcohol
segment of that category+.
3art L of Table ? sho&s that, &hereas 4bottle sha#e5 and 4more feminine5 are lin"ed to
4so#histicated image,5 there is not a strong association &ith 4im#ress others.5 This may suggest
more of an internal orientation &hile the 4e9#ensive7 ;uality5 association &ith 4im#ress others5
is ;uite strong and may be reflective of an e9ternal orientation.
,%%lications
Accordingly, consideration can no& be made of the o#tions available to the researcher &ho
uses the laddering a##roach and is faced &ith the challenge of a##lying the results to the solution
of some mar"eting #roblem. The G0M obtained through the laddering #rocedure offers several
#articularly valuable ty#es of information. 2t can serve as a basis forE )1+ segmenting consumers
7 &ith res#ect to their values orientations for a #roduct class or brand> )8+ for assessing
brands or #roducts in a fashion similar to the use of more traditional ratings> )*+ evaluating
com#etitive advertising> and )?+ as a basis for develo#ing advertising strategies.
Segmentation. The goal of segmentation schemes is to classify res#ondents &ith res#ect to
some as#ect of their behavior, attitudes, or dis#ositions in a &ay that hel#s us understand them as
consumers. The values orientations in a #erson6s ladder may serve as the basis for classification,
or the researcher may grou# these values at a still higher level. it is also #ossible to include
attribute$value connections in the segmentation scheme. Knce a segmentation scheme has been
develo#ed, res#ondents6 brand$consum#tion behavior or reactions to advertising may be
assessed.
Table . includes a summary by attribute and value for res#ondents &hose ladders e9tended to
the values level. 4Celonging5 &as included in the most ladders, &ith 4self$esteem,5 4accom#lish$
ment,5 and 4family life5 follo&ing in decreasing order of fre;uency )nine ladders did not reach
the values level and thus are omitted from this analysis+. The values can be grou#ed at a higher
level using 4achievement5 and 4social5 as higher$level value orientations. An e;ual number of
subDects fall into each of these t&o values$level orientations.
Kne could also include the attribute$value connections in the segmentation scheme, assessing
them at the levels used in the G0M or in grou#ing them as sho&n in Table . into mar"eting$mi9
com#onents. 2n this e9am#le, the attributes 4less alcohol5 and 4filling5 are lin"ed to social
values, &hereas 4#rice5 is tied more closely to achievement values. 43ac"aging5 attributes are
e;ually divided, although 4si/e5 is identified &ith social values, not achievement values.
!able C
Ladder 9reE+encies #or ,ttrib+te?1al+e LinFage
Achievement (ocial
Accom#lishment (elf$esteem Total Celonging Family life Total
)1?+ )1.+ )8+ )8,+ )+ )8+
3hysical attributes < ? 1, 1, = 1=
Barbonation < ? 1, , , ,
Bris# , , , = , =
1ess alcohol , , , 1 ? .
Filling , , , 8 * .
3rice = . 18 . , .
3ac"aging 1 < = . 8 =
1abel 1 * ? 8 , 8
(ha#e , * * 8 , 8
(i/e , , , 1 8 *
Mine ladders did not reach the values level.
Res#ondent segments could be studied for brand$consum#tion differences and #references and
advertising reactions evaluated. These segmentation bases could be translated into larger scale
research on brand usage and #reference and advertising theme evaluation. That is, the findings
from this research could become the basis for more traditional #a#er$and$#encil methods that
more readily lend themselves to large$scale data collection.
3roduct/Crand Assessment 'valuation of a #roduct or brand is another im#ortant mar"eting
;uestion for &hich the results of laddering research may be of use. 2t is advantageous to allo& re$
s#ondents to use their o&n frame of reference &hen #roviding their evaluations of a brand rather
than some researcher$su##lied attributes that may not be the subDect6s o&n. For many #roduct
categories or subclasses of categories, res#ondents are much more li"ely to ma"e #reference
Dudgments at the conse;uence and values levels than at the attribute level )Reynolds, Gutman,
and Fiedler, 1!?> Reynolds and Jamieson, 1!?+.
A statistical a##roach, Bognitive Lifferentiation Analysis )BLA+, has been develo#ed
)Reynolds, 1!*> Reynolds and (utric", 1!<+ to enable researchers to determine the level of
abstraction )attribute, conse;uence, or value+ at &hich #reference Dudgments are being made by
consumers. This a##roach #rovides indices indicating the discrimination #o&er of each of the
descri#tors &ith res#ect to a set of #air&ise discrimination bet&een stimuli. To collect data for
this ty#e of analysis, res#ondents are as"ed to sort or rate #air&ise combinations of brands in the
relevant #roduct class according to their res#ective #reference distance. Res#ondents are also
as"ed to #rovide information on the e9tent to &hich the brands #ossess or satisfy the elements at
each level of abstraction in their ladders. Kne a##ealing feature of this analytical method is that it
only re;uires ordinal data7no interval scale #ro#erties are necessary.
This information not only allo&s a determination of the levels &ithin a res#ondent6s ladder at
&hich #reference is determined, but the overall inde9 of the ladder allo&s the researcher to
determine each res#ondent6s o#timal ladder. Results from BLA analyses have sho&n that #eo#le
are not #articularly good at recogni/ing their o&n most discriminating &ay of evaluating the
brands &ithin a #roduct class, nor do they recogni/e the level of abstraction at &hich their
Dudgments are being made )see Reynolds H1!.I for a detailed summary of the method and the
results+. This suggests that researchers ought to be sus#icious of self$re#ort rating systems in$
herent in many attitude models and consumer surveys.
The out#ut from laddering, cou#led &ith the uni;ue analytical #rocedures it allo&s, #rovides
res8archers &ith a better understanding of the basis u#on &hich consumers ma"e distinctions
bet&een com#eting brands. Further, it #rovides a basis for develo#ing a #roduct s#ace that is
truly aligned &ith #reference, as such s#atial ma#s may be obtained using different levels of
abstraction as a frame of reference. Too often #roduct$#lanning decisions are based on discrimi$
nation differences and not #reference differences. Bonsumers, given the means$end frame&or",
are assumed to have multi#le orientations that are triggered by a given occasional conte9t )i.e.,
combination of situation and actors+. Thus, if the means$end #ers#ective is valid, #reference
&ould in most cases be multidimensional in nature. Therefore, the laddering a##roach #rovides a
uni;ue o##ortunity to understand the #roduct class in the consumer6s o&n conte9t. This &ould
seem to #rovide a good start for ma"ing decisions about #roducts and brands.
Assessing ,dvertising. Another im#ortant use for the results obtained through laddering
research is to uncover res#ondents6 evaluations of advertising. Advertising is vie&ed differently
&hen #erceived in the conte9t of different levels of abstraction )attribute, conse;uence, and
value+. To accom#lish this, after laddering, &hen res#ondents are sensiti/ed to the com#lete
range of their internal feelings about a #roduct class, they are sho&n a series of ads and as"ed to
rate them on the e9tent to &hich the ad communicates at each level and to #rovide some
comment on &hy it does or does not communicate at that level.
Analysis of these comments leads to the construction of a series of statements reflecting their
content. To further broaden the coverage of these statements, a model de#icting an advertising
research #aradigm can be used )see Figure 8+. This model )Reynolds and Trivedi+ indicates the
com#onents of an ad in relation to levels of involvement the consumer may have &ith the ad.
Fifty to si9ty statements can be develo#ed covering the advertising6s message elements, e9ecu$
tional frame&or"s, #erce#tions of the advertisers6 strategy and involvement &ith the ad, involve$
ment of the ad &ith the res#ondent6s #ersonal life, and the e9tent to &hich the ad ta#s into values
at a #ersonal level.
These statements can then be used to assess the relative communication at the various levels.
9ig+re 3
,dvertising $esearch Paradigm "ased on Means?End Chain Model and .ierarchical 1al+e
Str+ct+re ,nalsis
Ad 3erson
1evel Go& ad relates to #ersonal values
R
R @hat ad ma"es me thin" of
R
> Consumer ,enefit
> 3erce#tions of Involvement
of advertisers6 strategy @hat ad does to me &hile 2 &atch
R
R .0ecutiona% ?rame(or/
> Actors/situations
R
> @essage .%ements
> Attributes
R
Abstraction
This can be accom#lished, after a sensiti/ing laddering #rocedure, by sho&ing ads and as"ing 4if
the follo&ing statement a##lies5 to each res#ective ad. This #rocess can be o#erationali/ed by a
game$board a##roach )Gutman and Reynolds, 1!=+ &here a triangle is #rovided to the res#on$
dent &ith each verte9 re#resenting a se#arate ad. The use of three ads is suggested as an attem#t
to avoid the res#ondent from becoming too much of an advertising e9#ert. As each statement is
read the res#ondent can record the a##licability to one ad )recording the statement code at the
res#ective verte9+, or t&o ads )recording on the connecting line+, or all three )recording in the
middle of the triangle+. 2f the statement does not a##ly to any of the three ads, a 4not a##licable5
res#onse alternative is also #rovided.
The resulting #ercentage endorsement of each statement for each advertisement #rovides a
good indication of ho& the ad is vie&ed and the level at &hich the ad communicates. That is,
some ads may communicate &ell at the attribute level but not at the conse;uence or values level.
Bonversely, other ads may communicate &ell at the values level but be &ea" at the attribute
level. An effective ad in this conte9t is defined as one &hich communicates across all levels,
lin"ing attributes to benefits and to #ersonal values &hich often drive consumer decision$
ma"ing.
Develo%ing ,dvertising Strateg. 3erha#s the maDor benefit of laddering is the insight it
#rovides to advertising strategists. A definition of advertising communications &hich &ill #ermit
advertising strategies to be develo#ed from the G0M &ill be briefly discussed )see Reynolds and
Gutman H1!?I for a fuller discussion and illustration+. The levels of abstraction frame&or",
&hich underlie the formation of means$end chains, #rovide a basis for coordinating the results of
laddering to advertising strategy develo#ment. That is, the #erce#tual constructs de#icted in the
G0M can be used as the basis for develo#ing a strategy that &ill a##eal to consumers &ith that
#articular orientation to&ard the #roduct class.
Figure * sho&s the Means$'nds Bonce#tuali/ation of Bom#onents of Advertising (trategy
)M'BBA(+ in terms of five broad characteristics that corres#ond to the levels of abstraction
conce#tuali/ation )Klson and Reynolds, 1!*> Reynolds and Gutman, 1!?+. 4Lriving force,5
4consumer benefit,5 and 4message elements5 are directly coordinated to the values, conse$
;uences, and attributes levels of the means$end model. The e9ecutional frame&or" relates to the
scenario for the advertisement7 the 4vehicle5 by &hich the value orientation is to be communi$
cated. The s#ecification of this tone for the advertisement is a critical as#ect of strategy s#ecifi$
cation. 2t comes from an overall understanding of the &ay of #erceiving the #roduct class as indi$
cated by a #articular means$end #ath. As is a##arent &ith this s#ecification, added guidance can
be given to creatives &ithout infringing on their creativity.
The remaining and "ey as#ect of advertising strategy s#ecification is the conce#t of :leverage
#oint.5 Gaving all the other elements in mind, it is finally necessary to s#ecify the manner by
&hich the values$level focus &ill be activated for the advertisement, that is, ho& the values
considerations in the advertisement are connected to the s#ecific features of the advertisement.
)'9am#les of advertising strategy s#ecifications are not #rovided7the references cited above
#rovide am#le illustrations.+
Monetheless, the advantages of being able to s#ecify advertising strategy for all relevant
#arties7 management, creatives, and researchers7can be revie&ed. The strategy statement
itself becomes a concrete &ay of s#ecifying advertising strategy alternatives. These alternatives
are lin"ed to the chains &hich underlie them, and thus a direct connection e9ists bet&een the
strategy and the #erce#tual orientation of the consumer. Furthermore, the M'BBA( model
cou#led &ith the results from the G0M facilitate the develo#ment of several )truly different+
strategies for com#arison and revie&. 1astly, &hen a strategy has been seVO lected for e9ecution,
the M'BBA( model #rovides for a better common understanding of &hat the final #roduct
should be. This obviously leads to the use of the M'BBA( s#ecification as the basis for
evaluating the effectiveness of the advertisement.
9ig+re 6
Means?Ends Conce%t+aliGation o# Com%onents o# ,dvertising Strateg
!ri-ing ?orce The value orientation of the strategyE the end$level to be focused on in the
advertising.
5e-erage Point $The manner by &hich the advertising &ill :ta# into,5 reach, or activate the value
or end$level of focus> the s#ecific "ey &ay in &hich the value is lin"ed to the s#ecific
features of the advertising.
.0ecutiona% ?rame(or/ The overall scenario or action #lot, #lus the details of the advertising
e9ecution. The e9ecutional frame&or" #rovides the 4vehicle5 by &hich the value
orientation is communicated> es#ecially the gestalt of the advertisement> its overall
tone and style.
Consumer ,enefit The maDor #ositive conse;uences for the consumer that are e9#licitly
communicated. verbally or visually, in the advertising.
@essage .%ements The s#ecific attributes, conse;uences, or features about the #roduct that are
communicated verbally or visually.
S+mmar
This article revie&s and illustrates the techni;ue of laddering both as an intervie&ing #rocess
and through subse;uent analysis. 2t demonstrates the techni;ue6s usefulness in develo#ing an un$
derstanding of ho& consumers translate the attributes of #roducts into meaningful associations
&ith res#ect to self$defining attitudes and values. The underlying theory behind the method,
Means$'nd Theory, is discussed, as &ell as the elements of the means$end chains re#resenting
the cognitive levels of abstractionE attributes, conse;uences, and values.
The intervie& environment necessary for laddering to ta"e #lace is given s#ecial attention
along &ith the #articular #robing techni;ues em#loyed in the ;ualitative #rocess of laddering.
Casically, the res#ondent has to feel as if on a voyage of self$discovery and that the obDect of the
tri# is to revisit everyday, common#lace e9#eriences and e9amine the assum#tions and desires
driving seemingly sim#le choice behavior.
(everal s#ecific intervie&ing devices are described for eliciting #roduct distinctions from re$
s#ondents that serve to initiate the laddering #rocess, among them the use of triads, e9#loring
#reference$consum#tion differences, and e9amining ho& consum#tion differs by occasion. The
value of the occasional conte9t, #roviding a concrete frame of reference to generate meaningful
distinctions, is em#hasi/ed. Kther techni;ues Vor moving the laddering intervie& u#&ard &hen
bloc"ing occurs are also discussed and illustrated.
The analysis of laddering data is detailed noting the critical difference bet&een this method$
ology and more traditional ;ualitative research, namely, the #rimary out#ut being )structurally+
;uantitative in nature in the form of a hierarchical value ma# )G0M+. 2n this vein, the content
analysis of ladder elements is #ositioned as an im#ortant ste# in this 4crossing over5 from the
;ualitative to ;uantitative.
Letailed attention is #aid to the construction of the G0M from the im#lication matri9, &hich
re#resents the number of direct and indirect lin"ages bet&een the ;ualitative conce#ts elicited
during the laddering #rocess. Five ty#es of relations among elements are discussed, and their
res#ective im#lications for constructing a G0M are illustrated.
Gaving the G0M to &or" &ith, the ne9t ste# in transforming the out#ut of laddering into
useful information for mar"eting decision$ma"ing is to determine the dominant #erce#tual
orientations. That is, all #otential #ath&ays )connections among elements+ must be e9amined to
determine their relative strength of association. T&o #rimary considerations are s#ecified &ith
e9am#les, namely, the number of relations among elements &ithin the chain and the e9tent to
&hich all elements are interconnected.
1astly, the issue of a##lications is discussed referencing the "ey research #roblems of
#erce#tual segmentation, determining the im#ortance &eights of the various com#onents of the
ladders, and the develo#ment and subse;uent assessment of advertising from this value #er$
s#ective. All of the a##lication areas have in common that they de#end on laddering6s ability to
dra& out from the res#ondent the true basis for any meaningful connection they have to the
#roduct class.
$e#erences
Abelson, Robert. 4The 3sycho$logical (tatus of the (cri#t Bonce#t5 *merican Psycho%ogist *<
)1!1+E =1.7=8.
Car"er, R. G. 'cological 3sychologyE Bonce#ts and Methods for (tudying the 'nvironment of
Guman Cehavior. (tanford, BAE (tanford Jniversity 3ress, 1<!.
Lavis, M. (. 4That6s 2nterestingE
To&ard a 3henomenology of (ociology and a (ociology or 3heriomenology.5 Phi%osohy of the
Socia% Sciences 1 )1=1+E *,7*1?.
Lurgee, J. F. 4Le#th$2ntervie& Techni;ues for Breative Advertising.5 Aourna% of *d-ertising $e-
search 8., < )1!.+E 87*=.
Gutman, Jonathan. 4A Means$'nd Bhain Model Cased on Bonsumer Bategori/ation
3rocesses.5Aourna% of @ar/eting ?<, 8 )1!8+E<,7=8.
WWWWWW. 4Analy/ing Bonsumer Krientations To&ard Ceverages Through Means$'nd Bhain Anal$
ysis.5 Psycho%ogy and @ar/eting 1. */? )1!?+E 8*7?*.
WWWWWW, and (cott Alden. 4Adolescents6 Bognitive (tructures of Retail (tores and Fashion Bon$
sum#tionE A Means$'nd Analysis.5 2n Percei-ed Bua%ity of Products3 Ser-ices and Stores3 J. Ja$
coby and J. Klson, eds. 1e9ington, MAE 1e9ington Coo"s, 1!?.
Thomas J. Reynolds> and John Fiedler. 4The 0alue (tructure Ma#E A Me& Analytic Frame&or"
for Family Lecision$Ma"ing.5 2n The Changing Couseho%d< +ts &ature and Conse;uences3 M. 1.
Roberts and 1. @oert/el, eds. Bity, (tateE Callinger 3ublishing, 1!?.
WWWWWW andWWWWWW 4An 2nvestigation at the 1evels of Bognitive Abstraction Jtili/ed by the
Bonsumers in 3roduct Lifferentiation.5 2n *ttitude $esearch Dnder the Sun3 John 'ighmey, ed.
BhicagoE American Mar"eting Association, 1=.
WWWWW andWWWWW. 4Levelo#ing 2mages for (ervices Through Means$'nd Bhain Analysis.5 2n
Proceedings of 2nd Ser-ice @ar/eting Conference3 1!*.
,%%endi0
$aw Data #rom .%othetical /ine Cooler Data
$es%ondent Content codes
n+mber
- - -= -3 -* 3= =
3 - -= -* = = =
6 - -= -3 -* -* 36
: 6 * 3= = = =
C : -D 3= = = =
* 3 -= -3 -* -A 33
D I -3 -* 3= = =
A 6 A 3= = = =
B - -3 -* -A 36 =
-= - -= -* = = =
-- 6 A 3= = = =
-3 3 -= -3 -* -A 33
-6 - -3 -* 3= = =
-: - -3 -* -A 36 =
-C - -= -3 -* 3= =
-* 6 -* 3= = = =
-D - -= -3 -* 3= =
-A 3 -= -3 -* -A 33
-B - -= -3 -* -A 36
3= - -= -* = = =
3- 3 -= -3 -* -A 33
33 6 3= = = = =
36 - -= -3 -* 3= =
3: - -= -* = = =
3C 6 * -* 3= = =
3* 6 * -* -A 36 =
3D 6 A -A 3= = =
3A 6 -A 36 = = =
3B 6 -* 36 = = =
6= 6 A -A 33 = =
6- 6 A -D -A 36 =
63 6 D -A 36 = =
66 : -6 -D -A 36 =
6: : -6 -D -A 33 =
6C C -6 -D 36 = =
6* C -D 36 = = =
6D : -D 36 = = =
6A C -6 33 = = =
6B * -: -A 33 = =
:= * -: 3- = = =
:- * -: -A = = =
:3 * -: 3- = = =
:6 * -: 3- = = =
:: B -- -: -B 33 =
:C B -- -: -B 3- =
:* B -- -: 3- = =
:D B - -: -B 33 =
:A D -C 3- = = =
:B D -C 3- = = =
C= D -C = = = =
C- 6 A -* -A 33 =
C3 6 A -A 33 = =
C6 3 A -D 33 = =
C: 6 A -* -A 33 =
CC 6 A -A 33 = =
C* 3 A -D 33 = =
CD 3 A -D -B 33 =
CA - A -C = = =
CB * -= -* = = =
*= * -3 = = = =
*- * -B 3- = = =
*3 D -- -: -B 33 =
*6 : A -6 -D 36 =
*: : A -6 -D 33 =
*C C A -6 -D 36 =
** C -= -6 -D 33 =
*D B -B 3- = = =
WWWWWWand WWWWWW. 4Boordinating Assessment to (trategy Levelo#mentE An Advertising
Assessment 3aradigm Cased on the M'BBA( A##roach.5 2n *d-ertising and Consumer
Psycho%ogy3 J. Klson and F. (entis, eds. 3raeger, 1!=.
FassatDian, Garold. 4Bontent Analysis in Bonsumer Research.5
Aourna% of Consumer $esearch 63 1 )1==+E !71!.
Felly, George A. The Psycho%ogy of Persona% Constructs. Me& Nor"E
@. @. Morton and Bo., 2nc., 1...
Morgan, Anthony. 43oint of 0ie&E Magic To&n Revisited.5 Aourna% of *d-ertising $esearch
8?, ? )1!?+E ?7.1.
Klson, Jerry B., and Thomas J. Reynolds. 4Jnderstanding Bonsumers6 Bognitive (tructuresE
2m#lications for Advertising (trategy.5 2n *d-ertising and Consumer Psycho%ogy3 0ol. 2, 1. 3ercy
and A. @oodside, eds. 1e9ington, MAE 1e9ington Coo"s, 1!*.
3lummer, Jose#h. 4J#front Research and 'motional (trategies.5 (#eech given at Morth&estern
Jniversity (chool of Cusiness, 'vanston, 2llinois, February 1!..
Reynolds, Thomas J. 4A Mon$metric A##roach to Letermine the Lifferentiation 3o&er of At$
tribute Ratings &ith Res#ect to 3air&ise (imilarity Judgements HsicI.5 2n #roceedings of
*merican @ar/eting *ssociation .ducator1s Conference on $esearch @ethods and Causa%
@ode%ing. BhicagoE American Mar"eting Association, 1!*.
WWWWWW 42m#lications for 0alue ResearchE A Micro vs. Macro 3ers#ective.5 Psycho%ogy and
@ar/eting 8, ? )1!.+E 8=7*,..
WWWWWW and Jonathan Gutman. 41adderingE '9tending the Re#ertory Grid Methodology to Bon$
struct Attribute$Bonse;uence$0alue Gierarchies.5 2n Persona% Ea%ues and Consumer Psycho%ogy3
0ol 22., R. 3itts and A. @oodside, eds. 1e9ington, MAE 1e9ington Coo"s, 1!?a.
77 and 77. 4Advertising 2s 2mage Management.5 Aourna% of *d-ertising $esearch 8?, 1
)1!?b+E 8=7*<.
WWWWWW> WWWWWW> and John Fiedler. 4Jnderstanding Bonsumers6 Bognitive (tructuresE The Rela$
tionshi# of 1evels of Abstraction to Judgements HsicI of 3sychological Listance and 3reference.5
2n Psycho%ogica% Processes of *d-ertising .ffects< Theory3 $esearch and *%ication3 A.
Mitchell and 1. Al&itt, eds. 'rlbaum, 1!?.
WWWWWW, and 1inda Jamieson. 42mage Re#resentationsE An Analytical Frame&or".5 2n Percei-ed
Bua%ity of Products3 Ser-ices and Stores. J. Jacoby and J. Klson, eds. 1e9ington, MAE 1e9ington
Coo"s, 1!?.
WWWWWW and Fenneth (utric". 4Assessing the Borres#ondence of Kne or More 0ectors to a
(ymmetric Matri9 Jsing Krdinal Regression.5 Psychometri/a .1, 1 )1!<+E 1,17118.
77, and Mina"shi Trivedi. 4An 2nvestigation of the Relationshi# Cet&een the M'BBAs Model
and Advertising Affect.5 2n *d-ertising and Consumer Psycho%ogy. 0ol. 20, A. Tybout and 3.
Bafferata, eds. 1e9ington, MAE 1e9ington Coo"s.
Rosenberg, Milton :. 4Bognitive (tructure and Attitudinal Affect.5 Aourna% of *"norma% and
Socia% Psycho%ogy .* )1.<+E *<=7*=8.
Run"el 3. J., and *. '. McGrath. Research on Guman CehaviorE A (ystematic Guide to Method.
Me& Nor"E Golt, Rinehart X @inston, 1=8.
(heth, Jagdish. 4Mar"eting Megatrends.V6 Aourna% of Consumer @ar/eting )1!*+E .71*.
@ic"er, Alan. 4Getting out of our Bonce#tual Ruts.5 *merican Psycho%ogist ?, )1!.+E 1,?7
11,*.

You might also like