context maps
TRANSCRIPT
-
8/14/2019 Context Maps
1/24
Context effects and context maps
for positioning
Minhi Hahn*Eugene Won**Hyunmo Kang*Yong J. Hyun**KAIST Graduate School of Management**Dongduk Womens University
Received (in revised form): 29 September 2005
Context effects refer to changes in consumer preference and choice responseswhen a new alternative is added to a choice set. This paper proposes a generalscheme for classifying various context effects using newly defined share-ratiomeasures (SRM) and share-change measures (SCM). With these measures, we canalso draw context maps and preference-substitutability maps that visualise thenature of context effects and positions of competing brands. These maps allow
marketers to make positioning decisions that take advantage of positive contexteffects.
Introduction
A scene is shown of two truck drivers competing to see who will arrivefirst at a building. The brands on the trucks are shown as FedEx and UPS.
They arrive at the building almost at the same time. The drivers rush intothe building only to find a DHL man coming out with a smile. This is arecent DHL commercial, comparing it with and differentiating it fromFedEx and UPS. By giving consumers the impression that FedEx and UPSare similar, DHL may have tried to distinguish its own service as unique.
Market positioning is defined as arranging for a product to occupy aclear, distinctive, and desirable place relative to competing products in theminds of target consumers (Kotler & Armstrong 2004, p. 55). Similarly,product position means the way the product is defined by consumers onimportant attributes the place the product occupies in consumers mindsrelative to competing products (Kotler & Armstrong 2004, p. 259). Thus,
International Journal of Market Research Vol. 48 Issue 2
-
8/14/2019 Context Maps
2/24
unless competing products are clearly identified, market positioning isincomplete and a product position cannot be understood clearly.
It is becoming more difficult nowadays to identify relevant competingproducts as more products are competing in goal-derived or needs-basedproduct markets. For example, a luxury car brand may have to competenot only with other luxury car brands but also with other non-caralternatives, such as stock options or money, as rewards packages forsenior executives. Also, the main market for a ladys skin cream productmay be the Christmas present market for husbands. Aside from suchunusual examples, it is not an easy task to identify those competitors thatcompete more directly with the firms brand even in the same product
category. In this paper, we propose a new framework, called context maps,which can help marketing managers determine competing products forcomparison in making positioning decisions. The context maps showvisually how shares are gained or lost when each competing alternative isadded to consumers consideration sets.
The framework is also useful for classifying context effects (i.e. theeffects of adding an alternative to or subtracting an alternative from achoice set on changes in consumer preference and choice responses). Manypopular models used for predicting market shares, including multinomial
logit models, are based on the independence from irrelevant alternatives(IIA) principle (Luce 1959). This principle assumes that relative preferencebetween two options does not depend on the presence or absence of otheroptions. However, many consumer studies have shown the existence ofvarious context effects. For example, a new alternative product often takesdisproportionately greater share from similar than from dissimilarproducts (Debreu 1960). Tversky (1972) and Batsell and Polking (1985)are among the many who proposed choice models that incorporate suchsubstitutability effects among similar products. On the other hand, it was
observed that a new alternative product can increase shares of similar butmore powerful products. Huber et al. (1982) called such context effectsattraction effects. Simonson (1989) observed compromise effects suchthat the addition of an extreme alternative increases shares of intermediateoptions. The addition of a new alternative similar to competing productsmay increase or decrease shares of the focal product. While Kahn et al.(1987) observed lone alternative effects, where the lone alternatives lostshare, Brenner et al. (1999) observed what we call uniqueness effects,where the unique alternative gained share. Our classification frameworkmakes clear how those independently observed context effects areinterrelated.
Context effects and context maps for positioning
-
8/14/2019 Context Maps
3/24
We also present preference-substitutability maps. These maps show thecompetitive relationship between all pairs of brands in the market in terms
of their relative preference and relative substitutability with each other. Ascontext maps, preference-substitutability maps are derived from the changein market share of the existing brands due to the introduction of a third brand.
In the next section, we suggest a general scheme to classify various con-text effects when an alternative is added to a choice set. After key measuresto analyse context effects are proposed, we describe how context maps andpreference-substitutability maps can be drawn and analysed. Thenapplications are illustrated with real survey data for cell phones, digitalcameras, and movies in the Korean market. We conclude this paper with a
discussion of the strengths and weaknesses of the proposed framework.
Classifying context effects
We first introduce the situation where there is no context effect. Accordingto Luce (1959), the independence from irrelevant alternatives (IIA) axiomholds if
for i,j A T, where T= { 1 , 2 , , N} is a set ofNcompeting products,PA(i) is the probability of i being chosen when A is the set of availablealternatives. Popular choice models such as the multinominal logit (MNL)model (Guadagni & Little 1983; McFadden 1973) and ASSESSOR (Silk &Urban 1978), a new product forecasting model, are based on the IIAaxiom. When this assumption holds, the addition of a new alternative inthe choice set or the subtraction of an alternative from the choice set does
not affect the preference of consumers. Therefore, there is no context effectwhen the IIA axiom holds.For simplicity of further discussions, we will suppose there are two
alternatives, 1 and 2, in the choice set. Alternative 1 is our focal alterna-tive. Let the choice shares of alternative 1 and 2 be P(1) and P(2). Inaddition, suppose another alternative, 3, is added to the choice set. Let thechoice shares of 1 and 2 be P(1) and P(2) in the choice set with the threealternatives. With the simplified notations, IIA holds if
(1) (1)
(2) (2)
P P
P P
=
( ) ( )
( ) ( )
A T
A T
P i P i
P j P j
=
International Journal of Market Research Vol. 48 Issue 2
-
8/14/2019 Context Maps
4/24
Our classification scheme is based on the following definitions.
Definition 1
There is a context effectif
Definition 2
The context effect ispositive for alternative 1 if
The context effect is negative for alternative 1 if
Definition 3
Suppose the context effect is positive. The positive context effect is definedto be strong ifP(1) > P(1). The positive context effect is defined to beweak otherwise.
Table 1 shows our classification scheme. The context effect is defined to bepositive if the focal alternative gets a greater share than expected from the
IIA principle (i.e. when there is no context effect). The context effect isdefined to be strongly positive if the share of the focal alternative isincreased as the third alternative is included in the choice set. Note that theregularity assumption in choice states that choice shares cannot increasewhen a new alternative is added to the choice set. Thus, the situation withthe strong context effect is the one that violates the regularity assumption.Table 1 also relates each category to previously studied context effects.Also, there are context effects newly defined in this paper.
When the added alternative (alternative 3) takes away more share fromthe focal alternative 1 than from 2, we define that this is a negative contexteffect for alternative 1. If alternative 3 is more similar to alternative 1, it is
(1) (1)
(2) (2)
P P
P P
(1) (1)
(2) (2)
P P
P P
Context effects and context maps for positioning
-
8/14/2019 Context Maps
5/24
International Journal of Market Research Vol. 48 Issue 2
Tab
le
1
Choiceoutcomesandconte
xteffects
*
Classification
[P(1)/P(1)]*
Remark
Nocontexteffect
Equa
ltoone
Lessthanorequaltoone
Independencefromi
rrelevantalternatives(IIA)(Luce1959)
Neg
ativecontexteffect
Lessthanone
Substitutability(similarity)effect(negative)(Debreu1960;
Tversky1972)
Lonealternative
effect(Kahnetal.1987)
Weakpositivecontexteffect
Greaterthanone
Weakattraction
effect**
Weakcompromiseeffect**
Weakuniquenesseffect**
Substitutabilityeffect(positive)
Stro
ngpositivecontexteffect
Greaterthano
ne
Attractioneffect(Huberetal.1982)
Compromiseeff
ect(Simonson1989)
Uniquenesseffe
ct**
*P(1)andP(2)arechoiceshareswithonly1and2inthechoiceset.P(1)andP(2)arechoiceshareswith1,2andth
eaddedoptioninthechoiceset.
**N
ewlydefined.
1
2
1
2
[
(
)/
(
)]
[
(
)/
(
)]
P
P
P
P
-
8/14/2019 Context Maps
6/24
the substitutability effect (Debreu 1960). For example, suppose there aretwo animation movies in a choice set for viewing at the regular price (i.e.
The Incredibles and Shrek II). Also, suppose a new option is added ofviewing Shrek IIwith a $1 discount. The additional option is likely to takeaway most of the share from viewing Shrek IIat its regular price but notmuch from viewing The Incredibles. If alternative 3 is more similar toalternative 2, it is consistent with the lone alternative effect. Kahn et al.(1987) suggested that lone alternatives can be in a disadvantageousposition when competing with a group of alternatives that are similar toone another. If the addition of alternative 3 enhances the attractiveness ofthe category characterised by alternatives 2 and 3, it can reduce the share
of alternative 1 more than expected under the IIA condition.When the added alternative takes away more share from the competing
alternative 2 than 1, there is positive context effect for alternative 1. At thesame time, if the share of alternative 1 is increased, there is a strongpositive context effect. The attraction effect (Huber et al. 1982) andcompromise effect (Simonson 1989) are in this category. Also, theadvantage of a unique alternative over a group of similar alternatives,observed by Brenner et al. (1999), is consistent with the strong positivecontext effect. Simonson and Nowlis (2000) suggest that consumers need
for uniqueness enhances the choice of unique alternatives. We will call ituniqueness effect if alternative 1 shows a strong positive context effectwhen a third alternative (3) is added that is more similar to alternative 2.
With the addition of alternative 3, the market share of alternative 1 maybecome lower than its original share but it may still achieve a greater sharethan expected under the IIA condition. Such weak positive context effectswere not discussed in previous studies of consumer choices. If there is apositive context effect, but not strong enough to be classified as theattraction effect, compromise effect or uniqueness effect, because the
market share is not increased with the addition of the third alternative, wewill call it weak attraction effect, weak compromise effect and weakuniqueness effect, respectively. Weak positive context effects can arisefrom the substitutability effect as well. When the added alternative is moresubstitutable with alternative 2 than 1, it may take a disproportionatelygreater share from alternative 2.
Context maps
Previous studies on context effects suggest that IIA situations are quiteideal and do not often occur. Consumer preferences for the focal product
Context effects and context maps for positioning
-
8/14/2019 Context Maps
7/24
depend on the competitors considered in the choice set. Marketers maystrategically influence consumers in selecting potential competitors
compared in the choice set. The context maps discussed in the followingenable marketers to visually investigate the context effects originated fromthe existence of competitors considered in the choice set.
We suggest two key measures for the analysis of context effects based onTable 1. The first of these is the share-ratio measure (SRM), defined as:
SRM(1, 2|3)
= and
SRM(1, 2|None)
=
where 1, 2 and 3 denote that alternative 3 is added in the choice set of 1and 2, and None means there is no additional alternative other thanalternatives 1 and 2 in the choice set. SRM(1, 2) can be understood simply
as a measure of relative preference between alternatives 1 and 2. Logtransformation is applied to make the SRM value zero when the shares arethe same for alternatives 1 and 2.
Comparing the value of SRM(1, 2|3) with that of SRM(1, 2|None), wecan identify whether there is any context effect caused by alternative 3 ornot and, if there is, whether the context effect is positive or negative. If thevalue of SRM(1, 2|3) is greater than SRM(1, 2|None), the context effect ispositive, meaning that the share of alternative 1 is at least greater than thatunder the IIA situation when alternative 3 is included in the choice set. If
the value of SRM(1, 2|3) is less than SRM(1, 2|None), the context effect isnegative, meaning that the share of alternative is smaller than that underthe IIA situation. If SRM(1, 2|3) = SRM(1, 2|None), the IIA conditionholds, implying that there is no context effect.
The second measure is the share-change measure (SCM) defined as:
SCM(1, 2|3)
= and
SCM(1, 2|None) = 0
(1)log
(1)
P
P
(1)log
(2)
P
P
(1)log
(2)
P
P
International Journal of Market Research Vol. 48 Issue 2
-
8/14/2019 Context Maps
8/24
where 3 denotes that alternative 3 is included in the choice set. SCMsimply represents the change of the focal brands market share after the
introduction of alternative 3. Again, log transformation is applied to makethe value of SCM(1, 2|3) zero when there is no change in the share ofalternative 1 even if alternative 3 is included in the choice set. We defineSCM(1, 2|None) = 0 to denote that there is no change in share when noalternative is added to the choice set. From Table 1, we see that, assumingSRM(1, 2|3) > 0, the positive context effect is strong if SCM(1, 2|3) > 0.Otherwise, it is weak.
Using the two measures, SRM as the x-axis and SCM as the y-axis, wecan draw a context map for any pair of brands in the market, as in
Figure 1. Any point in the map has the coordinate of (SRM, SCM).Suppose we have collected data from consumers asking their choiceintentions in various choice sets with two alternatives (e.g. alternative 1and 2) or with three alternatives (e.g. alternatives 1, 2 and one otheralternative). We can easily identify what kind of context effect is occurringin each situation by calculating values of SRM and SCM for each of thecases when a third alternative is added and then drawing context maps likethat shown in Figure 1. All the points, {SRM(1, 2|k), SCM(1, 2|k)}, aredirected by the arrows starting from the point {SRM(1, 2|None),
SCM(1, 2|None)} which is {log (P(1)/P(2)), 0}. The directions of the arrowsshow the types of context effect that arise from the introduction of a thirdbrand, k, into the choice set {1,2}.
Context effects and context maps for positioning
Figure 1 Identifying the types of context effect with context maps
SRM(1,2|k)= log [P(1)/P(2)]
SCM(1,2|k)= log [P(1)/P(1)]
Case 3
Case 2
Case 1
Case 4
Case 5
(Log [P(1)/P(2)], 0)
Weakpositivecontexteffect
Negativecontexteffect
No contexteffect (IIA)
Strongpositivecontexteffect
-
8/14/2019 Context Maps
9/24
From context maps, we can identify the nature of context effects,observing the direction of the arrows. In Figure 1, context effects are
positive for product 1 in cases 1, 2 and 3. The arrows are directed to theright from the point of {SRM(1, 2|None), SCM(1, 2|None)}. In such cases,the market share for product 1 is at least more than expected under the IIAcondition. Case 1 is the most desirable situation, showing a strong positivecontext effect. Case 2 is a special example of a weak positive context effectsuch that market share of product 1 stays constant after the addition of thethird alternative. In Case 3, the context is positive but weak. The additionof an alternative decreases the share, but not more than expected under theIIA condition. The added alternative takes a disproportionately greater
share from alternative 2 than 1. Case 4 is the situation where IIA holds.The added alternative draws proportionally equal shares from alternatives1 and 2. In Case 5, addition of an alternative not only decreases the shareof the product but also damages the focal product more than its directcompetitor.
The illustration shows that we can analyse the advantages anddisadvantages of including an additional alternative in the choice set bydrawing context maps. Typically, the context map is drawn from theperspective of a focal product. With a fixed main competitor always in the
choice set, it shows the effect of adding a different product in the choiceset as the third alternative. Of course, for further analysis, marketers canchange the main competitor or even the focal product, and can analyse theeffects of changing the product. If we use similarity-among-alternativesdata in addition to context maps, we can identify the exact nature of thecontext effects described in Table 1.
Preference-substitutability maps
In this section, measures of relative preference and substitutability aresuggested through the application of the measures developed in previoussections. Relative preference, in a binary choice case, can be representedsimply by the share-ratio of two compared alternatives. Suppose our focalproduct is alternative 1. We define the relative-preference measure (RPM)of alternative 1 with respect to alternative j as:
RPM (1,j) = SRM(1,j|None)=(1)
log( )
P
P j
International Journal of Market Research Vol. 48 Issue 2
-
8/14/2019 Context Maps
10/24
wherej is the main competitor selected in the two-alternative choice set.The RPM(i,j) value can be calculated assuming each brand i in the market
is the focal alternative. The overall score of RPM at the industry level isobtained by getting the average of RPM(i,j) values across all thecompeting products for i:
Overall RPM(i) =
where Nis the total number of alternatives considered including i. Notethat the average ofRPM(i) values is always zero.
Substitutability of the focal firm 1 to a competitor, k, is conceptualised
as the degree of shares alternative k takes away from 1 when k is added asthe third alternative in the choice set. Assuming that there are Ncompetingproducts in total, we have (N 2) number of two-alternative choice setsthat include 1 and one of its competitors except k. The substitutabilitymeasure for alternative 1 with respect to k is the average of the propor-tions of the shares that k takes away from 1 in the (N 2) choice sets. Wedefine the relative-substitutability measure (RSM) of alternative 1 withrespect to k as:
wherej is the main competitor selected for the two-alternative choice setand N is the total number of alternatives considered including 1 and k.The greater share alternative k takes away from 1 rather than its
competitors, on average, the higher the value ofRSM(1, k). Thus, higherscores imply that 1 is more substitutable for k.RSM(i, k) values can be calculated, assuming each brand i in the market
is the focal alternative. The overall score of RSM at the industry level is
Note that the average value of all the notations RSM(i, k) is always zeroand the average value of overall notations RSM(k) is zero as well. Thismeans that the proposed substitutability measure shows the relative
1,
1( ) ( , )
1
N
k k i
Overall RSM i RSM i kN =
=
1,
1,
1(1, ) { ( ,1 | ) ( ,1 | )}2
1 ( ) ( )log log
2 (1) (1)
N
j k
N
j k
RSM k SRM j k SRM j NoneN
P j P j
N P P
=
=
1,
1( , )
1
N
j j i
RPM i jN =
Context effects and context maps for positioning
-
8/14/2019 Context Maps
11/24
substitutability of a pair compared to other pairs in the considered market.RSM has a value greater than zero if the pair has a substitutability greater
than the average of the market and less than zero in the other case.Utilising the proposed measures, we can drawpreference-substitutability
maps either from a focal firms perspective or from the overall industryperspective. Using the values of RPM(1, k) and RSM(1, k) as twocoordinates for each competitor k, preference-substitutability maps fromthe perspective of brand 1 can be drawn. The firm-level preference-substitutability map plots all the (N 1) competitors of the focal firms interms of their relative preference to and substitutability with the focalbrand. Also, with the overall RPM(i) and RSM(k) as the coordinate
{RPM(i), RSM(k)}, industry-level preference-substitutability maps can bedrawn. The industry-level preference-substitutability map plots all Nfirmsin the market according to their overall preference and relativesubstitutability values. With these maps, marketers can visually observeoverall positions of competitors with respect to the substitutability andpreference as conceptualised in this paper.
Application of context maps and preference-substitutability maps
Design and data collection
We collected choice data for cell phones, digital cameras and movies. Forthis study, five leading brands were selected from cell phones (LG,Motorola, Pantech, Samsung and SK Teletech) and digital cameras(Canon, Nikon, Olympus, Samsung and Sony). Also, five recently releasedmovies (Clean, DMZ, Notebook, Pretty and Taxi) were selected asalternatives in the movie category. These movies were all released on thesame day. Among the movies, DMZ and Pretty are Korean while the
others are not. The most important target segments for the three productsare young consumers in their twenties. We collected data from 86 collegestudents. Among the respondents, 47 (54.7%) were male and 39 (45.3%)female; the mean age was 22.7 years; 85 (98.8%) possessed cell phonesand 37 (43.0%) had digital cameras.
Subjects were asked to choose one alternative from each of ten two-alternative choice sets for one product category and each of ten three-alternative choice sets for another product category. There were threetypes of questionnaire. Two-alternative choice sets for cell phones wereincluded in Type A, for digital cameras in Type B and for movies in Type C.On the other hand, three-alternative choice sets for movies were included
International Journal of Market Research Vol. 48 Issue 2
-
8/14/2019 Context Maps
12/24
in Type A, for cell phones in Type B and for digital cameras for Type C.Thus, no subjects answered the two-alternative questions and the three-
alternative questions for the same products. Between the two-alternativequestions and three-alternative questions, subjects answered questions ondemographics and possession of the products. The three questionnairetypes were randomly assigned to subjects. Questions on perceivedsimilarity among alternatives were presented in the final part.
Subjects were told that all the cell phones had been launched recently.They were told that all the phones featured a one-megapixel digitalcamera, an MP3 player and a video messaging function. Thus, the fivealternatives differed only in brand name and price. For digital cameras,
information regarding megapixels, size of memory and price was given foreach alternative. In the case of the movies, short introductions regardingthe genre and synopsis were presented to the subjects. We present the datafor the cellular phone category in Table 2.
Our raw data show that predicting market share based on the IIAassumption can be quite misleading. For example, SK Teletech andPantech have market share of 64% and 36% respectively in a pairedcomparison case. When LG Cyon is introduced into the choice set, it takesan overall share of 7%. According to the IIA principle, the shares for SK
Teletech and Pantech should be 60% and 33% after the new brandintroduction. However, the data show that market share for SK Teletechand Pantech is 82% and 11% respectively after the introduction. There isa 22.4% deviation from what the IIA assumption predicts. The data verifythe importance of analysing the context effects.
Analysis of key measures
From the choice data shown in Table 2, values of the share ratio measure
(SRM) and share change measure (SCM) were calculated. Table 3 showsthe SRM and SCM scores for the cellular phone category. Note that agreater value of SRM(focal, 2nd brand|3rd brand) than SRM(focal, 2ndbrand|None) means the focal brand shows a positive context effect whenthe third brand is included in the choice set, and vice versa. A positivevalue of SCM stands for an increase in share of the focal brand when thethird brand is added in the choice set. It is very difficult to interpret theSRM and SCM values directly from Table 3. The context maps allowmarketers to easily identify the context effects.
Context effects and context maps for positioning
-
8/14/2019 Context Maps
13/24
International Journal of Market Research Vol. 48 Issue 2
Tab
le
2
Sharesofcellphones
Two
-andthree-alternativechoiceset
s
Set1
Set2
Set3
Set4
Set5
Set6
Set7
Set8
Set9
Set10
i
LG
Pantech
Samsung
Motorola
Samsu
ng
SKTeletech
SKTeletech
LG
Pantech
Motorola
j
SKTeletech
Samsung
M
otorola
Pantech
LG
Pantech
Samsung
Motorola
LG
SKT
eletech
k
Samsung
LG
Pantech
SKTeletech
Motorola
LG
Pantech
SKTeletech
Motorola
Sam
sung
Sha
resintwo-andthree-alternativechoicesets(%)
Two-alternativech
oicesets
Three-alternativechoicesets
i
j
i
j
k
Set
1
14.3
85.7
10.7
35.7
53.6
Set
2
35.7
64.3
10.7
75.0
14.3
Set
3
89.3
10.7
60.7
21.4
17.9
Set
4
21.4
78.6
07.1
14.3
78.6
Set
5
89.3
10.7
71.4
10.7
17.9
Set
6
64.3
35.7
82.1
10.7
07.1
Set
7
50.0
50.0
32.1
60.7
07.1
Set
8
50.0
50.0
3.6
21.4
75.0
Set
9
67.9
32.1
32.1
32.1
35.7
Set
10
32.1
67.9
07.1
39.3
53.6
-
8/14/2019 Context Maps
14/24
Context effects and context maps for positioning
Table 3 SRM and SCM values for cell phones
Focal brand 2nd brand Added brand SRM SCM
Samsung SK Teletech Pantech 0.636 0.194LG 0.405 0.069
Motorola 0.310 0.069
Pantech SK Teletech 2.140 0.057LG 1.946 0.154
Motorola 1.224 0.057
LG SK Teletech 1.609 0.511Pantech 1.658 0.174
Motorola 1.897 0.223
Motorola SK Teletech 2.015 0.511
Pantech 1.041 0.386LG 1.386 0.223
SK Teletech Samsung Pantech 0.636 0.442LG 0.405 0.336
Motorola 0.310 0.241
Pantech Samsung 1.504 0.693LG 2.037 0.245
Motorola 1.705 0.201
LG Samsung 1.204 0.875Pantech 2.442 0.043
Motorola 3.045 0.134
Motorola Samsung 1.705 0.547Pantech 2.398 0.147
LG 1.253 0.100
Pantech Samsung SK Teletech 2.140 1.609
LG 1.946 1.204Motorola 1.224 0.693
SK Teletech Samsung 1.504 1.609
LG 2.037 1.204Motorola 1.705 0.916
LG Samsung 0.288 1.846
SK Teletech 0.405 1.846Motorola 0.000 0.747
Motorola Samsung 0.182 1.482
SK Teletech 0.693 1.705LG 0.105 0.894
LG Samsung SK Teletech 1.609 0.000
Pantech 1.658 0.288Motorola 1.897 0.000
SK Teletech Samsung 1.204 0.288
Pantech 2.442 0.693Motorola 3.045 1.386
(continued)
-
8/14/2019 Context Maps
15/24
Table 4 shows the values of the relative preference measure (RPM) aswell as the average of RPM across all the competitors (i.e. the overallrelative preference measure). Note that a positive value means the focalbrand is preferred more than the comparison brand. Naturally, the greaterthe overall RPM, the stronger the overall preference of the brand. Amongcell phones, Samsung and SK Teletech are the top two in terms ofpreference. It is interesting that Samsung and SK Teletech are equallypreferred when compared directly. However, when they are compared toLG or Motorola, Samsung is preferred much more than SK Teletech. As
expected, we found Canon and Sony to be the top two brands in the digitalcamera category. Among the movies, Notebook and Taxi were the mostpopular.
Table 5 shows the values of the substitutability measures RSM andoverall RSM. A positive number means the added brand takes away moreshare from the focal brand than its competitors, on average. The overallrelative substitutability measure is the average of the RSM scores. Amongthe cell phones, SK Teletech is the least substitutable and Pantech is themost substitutable overall. LG is highly substitutable with SK Teletech.Motorola is highly substitutable with Pantech. In the following sections,we interpret maps drawn from the values in Tables 3, 4 and 5.
International Journal of Market Research Vol. 48 Issue 2
Table 3 SRM and SCM values for cell phones (continued)
Focal brand 2nd brand Added brand SRM SCM
LG (continued) Pantech Samsung 0.288 0.811SK Teletech 0.405 1.504
Motorola 0.000 0.000
Motorola Samsung 0.511 1.540SK Teletech 1.792 2.639
Pantech 0.105 0.442
Motorola Samsung SK Teletech 2.015 0.405Pantech 1.386 0.511
LG 1.041 0.693
SK Teletech Samsung 1.705 1.504
Pantech 1.253 0.405LG 2.398 1.504
Pantech Samsung 0.511 1.030SK Teletech 1.792 0.847
LG 0.105 0.336
LG Samsung 0.182 0.000SK Teletech 0.693 1.099
Pantech 0.105 0.511
-
8/14/2019 Context Maps
16/24
Analysis of preference-substitutability maps
Figure 2 shows the industry-level preference-substitutability maps for thethree product categories. Every point in the map has the coordinate of{RSM(i), RPM(i)}. Among the cell phones, the Samsung brand is the most
preferred. However, it is more vulnerable in terms of substitutability thanSK Teletech. Pantech is in a highly substitutable position. In terms ofpreference, it is not among the weak brands. However, whenever a thirdbrand is considered in the choice set, Pantech is the brand that loses sharessubstantially. Nikon among the digital cameras and Clean among themovies are in similar positions to Pantech in that they are highlyvulnerable in terms of substitutability. On the other hand, DMZ in themovie category is the least substitutable movie although it is not amongthe most preferred. Apparently, those who prefer DMZ do not changetheir preferences when a third movie is presented as an option in the choiceset.
Context effects and context maps for positioning
Table 4 Relative preference and overall relative preference values
Cell phones
Comparison brand
Focal brand Samsung SK Teletech Pantech LG Motorola Overall
Samsung 0.000 0.588 2.120 2.120 1.207
SK Teletech 0.000 0.588 1.792 0.747 0.782
Pantech 0.588 0.588 0.747 1.299 0.218
LG 2.120 1.792 0.747 0.000 1.165
Motorola 2.120 0.747 1.299 0.000 1.042
Table 5 Relative substitutability and overall relative substitutability values
Cell phones
Comparison brand
Focal brand Samsung SK Teletech Pantech LG Motorola Overall
Samsung 0.312 0.302 0.343 0.241 0.149
SK Teletech 0.429 0.555 0.516 0.687 0.547
Pantech 1.144 0.833 1.404 0.833 1.054
LG 0.371 0.313 0.098 0.094 0.034
Motorola 0.345 0.834 0.156 0.544 0.392
-
8/14/2019 Context Maps
17/24
Selected preference-substitutability maps from a firms perspective arepresented in Figure 3. In the preference-substitutability map for brand i,every competitor, k, is plotted as the point with the coordinate {(RSM(i, k),RPM(i, k)}. The first three are maps from the perspective of Samsung cellphone, Canon digital camera and the movie Notebook. Each is considered
to be the leader in its respective product category. As they are the mostpreferred brands in their categories, there are no competitors in the thirdand fourth quadrants. When compared directly, SK Teletech is equallypreferred to Samsung in the cell phone category and Nikon is equallypreferred to Canon in the digital camera category. Note that Samsung issubstitutable to Pantech, implying that Pantech, not SK Teletech, could bethe most serious competitor. Similarly, for Canon, Nikon could be themost serious competitor. The movie Notebook seems to be in a stableposition. It is the most preferred and the second least substitutable movie.
Some interesting results are found for other competitors, too. Thebottom three maps are preference-substitutability maps drawn from theperspective of SK Teletech, Pantech and LG, which are active competitorsin the cell phone category. SK Teletech is as strong a brand as Samsung inthis category. Although it is less preferred to Samsung overall, it is strongerin terms of the substitutability with other brands.
On the other hand, Pantech and LG are vulnerable brands, being highlysubstitutable with other competitors. For these brands, it is best toinfluence consumers to consider only one of the competitors along withtheir brands, if possible. In particular, this may be a highly reasonablestrategy for Pantech because it shows a 35.7% share against the strongest
International Journal of Market Research Vol. 48 Issue 2
Figure 2 Preference-substitutability maps at the industry level
Cell phones Digital cameras Movies
Preference
Substitutability
SK Teletech
Samsung
Pantech
LGMotorola
1 0 1 2
2
1
0
1
2
Preference
Substitutability
DMZ
Notebook
Taxi
Clean
Pretty
1 0 1 2
2
1
0
1
2
Preference
Substitutability
Sony
Canon
Nikon
Olympus
Samsung
1 0 1 2
2
1
0
1
2
-
8/14/2019 Context Maps
18/24
brands like Samsung or SK Teletech when compared directly in the two-alternative choice sets (Table 2). Whenever a third competitor isconsidered, the added brand takes away substantial share from Pantech.LG is in a very difficult position, which requires dramatic changes in termsof both preference and substitutability.
Analysis of context maps
Context maps let marketers visually identify types of context effect whendifferent competitors are added to their consumers choice sets. Figure 4shows selected context maps.
Context effects and context maps for positioning
Figure 3 Selected preference-substitutability maps from a firms perspective
Cell phones(for Samsung)
Digital cameras(for Canon)
Movies(forNotebook)
Preference
Substitutability
SK Teletech
Pantech
LG
Motorola
1 0 1
1
0
1
2
3
Preference
Substitutability
DMZTaxi
CleanPretty
2 1 0 1
1
0
1
2
Preference
Substitutability
Sony
Nikon
Olympus
Samsung
1 0 1
1
0
1
2
Cell phones(for SK Teletech)
Cell phones(for Pantech)
Cell phones(for LG)
Preference
Substitutability
Samsung
SK Teletech
SK Teletech
Samsung Samsung
Pantech
Pantech
LG
LG
Motorola
MotorolaMotorola
1 0 1
1
0
1
2
Preference
Substitutability
1 0 1
3
2
1
0
1
Preference
Substitutability
1 0 1 2
1
0
1
2
-
8/14/2019 Context Maps
19/24
International Journal of Market Research Vol. 48 Issue 2
Figure 4 Selected context maps
Motorola
LG
Pantech
SCM
SRM0.0 0.3 0.6 0.9
0.0
0.1
0.2
0.3
Product: cell phoneFocal brand: SamsungSecond brand: SK Teletech
Motorola
Pantech
SCM
SRM
0.0 0.5 1.0 1.5 2.0 2.5
0.6
0.4
0.2
0.0
0.2
Product: cell phoneFocal brand: SamsungSecond brand: LG
Motorola SK TeletechSK Teletech
LG
SCM
SRM
0 1 2 3
0.1
0.0
0.1
0.2
Product: cell phoneFocal brand: SamsungSecond brand: Pantech
Motorola
SKTeletech
Pantech
Pantech
SCM
SRM
2.5 2.0 1.5 1.0 0.5 0.00.0
0.1
0.2
0.3
0.4
Product: cell phoneFocal brand: LGSecond brand: Samsung
Motorola
SCM
SRM
0.9 0.6 0.3 0.0 0.3 0.6
2.0
1.5
1.0
0.5
0.0
Product: cell phoneFocal brand: LGSecond brand: Pantech
Motorola
Samsung Samsung
SK TeletechSCM
SRM
3.5 2.5 1.5 0.5 0.5
1.5
1.0
0.5
0.0
Product: cell phoneFocal brand: LGSecond brand: SK Teletech
Sony
Samsung
Samsung
Nikon Clean
Pretty
Taxi
Olympus
Olympus
SCM
SRM
0.0 0.5 1.0 1.5 2.0
0.2
0.0
0.2
0.4
Product: digital cameraFocal brand: CanonSecond brand: Nikon
SCM
SRM
0.6 0.3 0.0
0.2
0.1
0.0
0.1
0.2
Product: Korean moviesFocal brand: DMZSecond brand: Notebook
SCM
SRM
0.6 0.3 0.0 0.3
0.2
0.1
0.0
0.1
Product: digital cameraFocal brand: SonySecond brand: Canon
-
8/14/2019 Context Maps
20/24
The focal brand of the first three context maps (the upper row) isSamsung in the cell phone category. The first map assumes that Samsung
and SK Teletech are in the choice set. It is natural to consider SK Teletechas the main competitor for Samsung because it is almost equally preferredin the cell phone market. The context map clearly shows that it is to theadvantage of Samsung to influence consumers who have the two brands intheir choice sets to include another competitor. Strong positive contexteffects can be expected when a third alternative is included. The mapimplies that the biggest increase in share for Samsung is expected whenPantech is included in the choice set as a third alternative. The second mapshows the context effects when Samsung and Pantech are in the choice set.
For Samsung, it is best to include LG as the third alternative in the choiceset as a substantial increase in share is expected. Because consumers tendto perceive LG and Pantech to be similar, Samsung may be easilydifferentiated from the two competitors. When Motorola or SK Teletech isincluded as the third alternative, Samsung can expect positive contexteffects, but not an increase in market share. The third map shows thecontext effects when Samsung and LG are included in the choice set. In thiscase, Samsung should try not to include any third alternatives in the choiceset. When a third alternative is included, Samsung will observe a negative
context effect. It is likely to face a substitution effect or a lone alternativeeffect.The three context maps in the second row are drawn from the
perspective of LG, which is one of the weakest brands in the cell phonemarket. The left context map implies that LG will be better off by addinganother competitor for consumers who have LG and Samsung in thechoice sets. In fact, LG can expect an increase in market share if Pantechis included in the choice sets. The addition of Motorola or SK Teletech tothe choice set will not change the share of LG. For those consumers who
have LG and SK Teletech or LG and Pantech in their choice sets, theaddition of a third alternative will not increase LGs share.The first two maps in the third row are selected from the digital camera
category. In this market, Canon is the leader, with Sony following closebehind. The first map shows the context map of Canon for consumers whohave Canon and Nikon in their choice sets. For these consumers, ifSamsung or Olympus is added to the choice set, strong positive contexteffects for Canon arise. As consumers tend to perceive Olympus orSamsung to be more similar to Nikon, Canon is likely to enjoy theuniqueness effect, easily differentiating it from other competitors. AddingSony shows the weak positive context effect. Canons market share will be
Context effects and context maps for positioning
-
8/14/2019 Context Maps
21/24
-
8/14/2019 Context Maps
22/24
can be visualised at the industry level with respect to preference andsubstitutability.
If the maps are analysed along with the perceived similarity data amongcompetitors, types of context effect can be identified. For example, if wehave detailed similarity data among competing products we can identifywhether it is the attraction effect, uniqueness effect or compromise effectwhen a strong positive context effect is observed. If marketers want tofurther investigate the implications of context effects, they can rely onanalysing models that incorporate context effects (Kivetz et al. 2004).
Of course, the value of analysing such maps depends on the quality ofthe data. The subjects surveyed should be representative of target segments
of the focal brand. Also, the context effects need to be compared withbasic cases where two alternatives are in the choice sets. Context effects letmarketers know how shares change from basic cases when a thirdcompetitor is included in the choice sets. The context effects should not beinterpreted beyond the basic cases on which the observed effects are based.
Our framework is developed mainly for situations where a single optionis chosen. When consumers choose more than one option, we may applyour framework repeatedly by sequentially eliminating the option that hasalready been chosen. Unless consumers choose more than one option at
the same time, our framework can still be a useful tool. For situationswhere consumers choose more than one option at the same time, we needto define alternatives as potential product mix options consumers mayselect.
There may be several directions for further research. One is studyingconditions that enhance the consistency of context effects over time.Analysing context maps will be all the more useful if the observed effectsare stable over a long period of time (Drolet 2002). Identifyingheterogeneity among consumers regarding context effects is also
important. The more homogeneous the consumer choice set in the targetmarket, the more useful will be the analysis of context maps. Studies onfurther applications of the measures may also be directed towardsmeasuring brand equity. Substitutability could be a key concept for adefensive strategy, whereas preference could be a key concept for anoffensive strategy as well as for defending brand power.
References
Batsell, R.R. & Polking, J.C. (1985) A new class of market share models. Marketing
Science, 4, Summer, pp. 177198.
Context effects and context maps for positioning
-
8/14/2019 Context Maps
23/24
Brenner, L., Rottenstreich, Y. & Sood, S. (1999) Comparison, grouping, andpreference. Psychological Science, 10, May, pp. 225229.
Debreu, G. (1960) A review of individual choice behavior: a theoretical analysis.American Economic Review, 50, December, pp. 186188.
Drolet, A. (2002) Inherent rule variability in consumer choice: changing rules forchanges sake.Journal of Consumer Research, 29, December, pp. 293305.
Guadagni, P. & Little, J.D.C. (1983) A logit model of brand choice calibrated onscanner data. Marketing Science, 2, Summer, pp. 203238.
Huber, J., Payne, J.W. & Puto, C. (1982) Adding asymmetrically dominatedalternatives: violation of regularity and similarity hypothesis.Journal of ConsumerResearch, 9, June, pp. 9098.
Kahn, B., Moore, W.L. & Glazer, R. (1987) Experiments in constrained choice.Journal of Consumer Research, 14, June, pp. 96114.
Kivetz, R., Netzer, O. & Srinivasan, V. (2004) Alternative models for capturing thecompromise effect.Journal of Marketing Research, 41, August, pp. 237257.
Kotler, P. & Armstrong, G. (2004) Principles of Marketing(10th edn). New Jersey:Prentice Hall.
Luce, R.D. (1959) Individual Choice Behavior: A Theoretical Analysis. New York:John Wiley & Sons.
McFadden, D. (1973) Conditional logit analysis of qualitative choice behavior, inZarembka, P. (ed.) Frontiers in Econometrics. New York: Academic Press,pp. 105142.
Silk, A.J. & Urban, G. (1978) Pretest market evaluation of new packaged goods:a model and measurement methodology.Journal of Marketing Research, 15, May,
pp. 171191.Simonson, I. (1989) Choice based on reasons: the case of attraction and compromise
effects.Journal of Consumer Research, 16, December, pp. 158174.Simonson, I. & Nowlis, S.M. (2000) The role of explanations and need for
uniqueness in consumer decision making: unconventional choices based onreasons.Journal of Consumer Research, 27, June, pp. 4968.
Tversky, A. (1972) Elimination by aspects: a theory of choice. Psychological Review,79, 4, pp. 281299.
International Journal of Market Research Vol. 48 Issue 2
-
8/14/2019 Context Maps
24/24