Chapter IV
Analysis of Results I
Consumer Preferences and Priorities
In thc. earlier chapter, the various tl~eoretical considerations for consumer
satisfaction and dissatisfaction were discussed. These theoretlcrll considerations have
provided a scope for identifying the important variables contrlhu~ing satisfaction to the
consumers. In any consumer satisfaction study, consumer preferences I expectations play
a vital role in deciding the level of satisfaction. The first phase of ;~nalysis in this respect
is carried out to determine the level of preferences i expectattons prevailing among the
selected sample consumers of 100 cc motor cycles and also to 1.1nd out the interrelations
anlong the various consumer preference variables. 'l'he present chapter discusses the
responses collected from the consumers pertaining to the~r preferences, by analysing thc
results with the appropriate statistical tools
The first issue in this connection is idcnl~ty~nf rhc Important product and service
attributes that are essential in influencing the purchase dcc~s~on of the consumers I1ivc
important variables contributing to consumer satisfaction 1 dissat~sfaction have been
identified. The variables identified include ( i ) Company Image ( i i ) Product Feature (i i i )
Customer Service Facilities (iv) Delivery Terms and (v) Product Price. These variables
were further classified into various attributes. These attributes were furnished in the form
of questions, requesting the consumers to give their expectations regarding the
importance of each attribute. The first pan of the questionnaire comprises of questions
relating to these attributes in order to find out the level of consumer preferences.
Factor Analysis is widely employed to study a complex product or service
attributes to Identify the major characteristics (factors) which are considered important
to the consumer. Harper W.Boyd, Jr. Ralph Westfail and Stanley F.Stasch while
explaining a case study relating to an automobile purchase e\~;iluation advocated the use
of factor analysis for evaluation'. According to them. researchers can ellcit the opinion
of a large salliple of buyers about their agreement or d~sa~reetncnt ahout varlous Prcduct
Features such as Safety, Exterior Styling, Interior Roominess o r Iicononiy of Operation.
Once this infbrmation is available, it can be used to ~dentify t11c preterence and priorities
of the customers. These informations can guide the industry as rtbgards thc characteristich
which are important to be incorporated into the product and ~Jcntify the advertising
themes with the help of Factor Analysis. Jagdish N Sheth, in hls rnultlvariate analysis in
Marketing specifies Factor Analysis as a method of reducing a het of data into a Inore
compact form while throwing certain properties of' the data tnto hold relief. Furthcr,
the uses of Factor Analysis in marketing research can be rcvlcwed by looking at the
following five uses of the technique.
In Factor Analysis the data are collected o n the ranklrlg or rat~ng of products in
terms of overall preference. Market researchers l o o k for c1uc.s lo ~nfe r the d~mensions
which latently order products or service facilities in terms of consumer preference. I t
helps to identify the characteristics for finding the overall preference. Clustering of
variables or individuals for classification and segmentation can also bc done by Factor
H. W .Boyd.Jr.Ralph Estfall and Stanley F. Stasch, Marketing Research Text and Cases, Seventh Edition, Illinois, Richard D.Irwin lnc. 1990, p.638
Jagdish N.Sheth, "Multivariate Analysis in Marketing". Jounlal of Advertising Research, Vol. 10, No. 1, Feb. 70, p. 29-38
Analysis. Another use of Factor Analysis is to isolate those variables which are potential
factors.
There are four stages in Factor Analysis. The correlation niatrlx for all attributes
are computed and attributes which are not interrelated can he identlf~ed from th; matrix
and associated statistics. The appropriateness of the usc of 1'r:ctor model can also he
evaluated. The variables must be related to each other in order to make the factor model
appropriate. If the correlations between the attributes are small, ~t is unlikely that they
share common factors.
The Kaiser-Meyer-Olkin (KMO) measure of' sampling adequacy IS an indcx ['or
comparing magnitudes of the observed correlation coefficients to the magnitudes of the
partial correlation coefficients. Small values for the K M O measure indicate that the
Factor Analysis of the attributes is not an apt method hecause correlation hetween palrs
of attributes cannot be explained by the othcr attr~butra KMO index ranges hetween 0
and 1. Kaiser characterises the measures in the 0 9 0 ' s ; i j marvelous, in the 0.80's ;is
meritorious, in the 0.70's as middling, in the 0 (10's as rncJlocre. in the 0 50's as
miserable and below 0.50's as unacceptable3.
The goal of the factor extraction step is to determine the important and essential
factors that contribute to consumer satisfaction. I n Principal Component Analysis, linear
combinations of the observed attributes are formed. The first principal component is the
combination that accounts for the largest amount of variance in the sample. The second
principal component accounts for the next largest amount of variance and uncorrelated
Kaiser, H.F., "An Index of Factorial Simplicity", Psychometrika Vo1.39, 1974, pp. 31-36.
TABLE 4.1
Frequency Distribution of Preference Scores on Five Variables
Contributing Consumer satisfaction
Attribute
-
Company Image
Brand Imagc
Quality Image
Media Image
Reference Group Image
Service Image
R & D Image
Product Feature
Product Appearance
Technology
Emergency Needs
Technical Specification
Proximity of Outlet
Range of Models
Problem Freeness
Fuel Efficiency
Pick up 1 Brake Control
Environment Safety
Seating Comfort
Citythighway Riding
h~ AI All
Inlportant
3
0
h
8
0
0
2
0
5
0
17
8
4
3
2
4
6
2
\ ( # \ rrs
1111l1vr1~111
-- 7 7
4
7- 3
2h
7
7
7
9
17
10
56
53
9
7
7
10
14
13
Ranking
C * ~ t r w h u l
I ~ u ~ M u n l
98
Ih
I h l
143
103
87
87
3 I
7 h
58
117
143
3 5
6 1
20
78
80
60
Ed.
~mponanl
155
254
54
47
120
142
162
239
150
204
103
8 1
275
24 1
279
21 1
118
163
V r n
h p n a n l
22 1
225
255
275
260
263
24 I
220
25 1
227
206
2 14
176
187
101
1 96
28 1
26 1
58
123
123
143
0 5
1 06
124
8 1
126
55
178
252
270
116
78
83
172
172
172
158
9
7
46
5 2
15
24
20
I I
20
2
I2
I h
19
3
6
4 1
5 3
I 00
87
1 0
4
7
7
4
5
7
1
3
3
0
3
5
2
( )
1
1 1
14
66
56
6
239
243
214
187
245
249
191
235
257
252
256
1 90
1x7
254
254
233
184
116
151
226
Customer Service
Dependable Service
Employee Behaviour
Service CentreAppearance
Record maintenance
Problem Appraisal
Complaint Register
Working Hours
Delivery Scl~edule
Maintenance Awareness
Delivery Terms
Prompt Dellvery
Colour Cho~ce
Choice of Outlet
User Guidance
Easy Availability
Product Price
Competitive Pricing
Price Alteration
Pollution Awareness
Loan Facility
Cost of R & D
Resale Value
189
119
109
113
139
113
154
169
93
190
50
30
22
126
160
131
76
45
33
99
Weighted average scores are calculated for different attributes by assigning weights as
5.4,3,2, and 1 to the 'Extremely Important', 'Very Important', 'Somewhat I~nponant', 'Not Very
Important' and 'Not At All Important' respectively. The same calculated welghted average
preference scores are going to be used for comparison with satisfaction scores in the next chapter
Rleasure of Adequacy for the Factor Analysis
Any data in general should be appropriate for f'actor analytical n~tdel that may vary from
one situation to another. Before proceeding with the Factor Analys~s of the present data, the
adequacy test must be applied. The rule of thumb index provided by tia~ser Measure of Sampling
Adequacy car1 be applied to acquire a rough idea ot whether the data lire adequate for 111c
techn~que or not.
The following table 4.2 provides the KMO index for the five variables chosen for the
present study.
TABLE 4.2
Kaiser Meyer Olkin Measure of Adequacy on Fivr ('otlsl~mcr Satislaction Variables
Measure
Mediocre
Middling
Meritorious
Mediocre
Mediocre
Variable
Company Image Variable
Product Feature Variable
Customer Service Variable
Delivery Terms Variable
Product Price Variable
IiMO
0.640
0.714
0 814
O 660
0 620
The results presented in table 4.2. indicate that the acceptable adequacy is provided for
all the variables. For Customer Support Service variable the Measure of Sa~npling Adequacy 1s
considered as nleritorious. Thus, the Factor Analysis can bc carried further for factor extraction
and also to identify the consumer preference attributes on all the five vartable chosen.
Factor Structure for the Company Image Preferences
Comprehensive models of consumer behaviour !lave suggested that consun~rr 's brand
tnformation and Company lmage of the product is an Irnportant input into 111s product purcliase
decisions. This variable is generally thought to be one of'the several Ley v;~rtables that Intervenes
hetween a dectsion maker's perceptions and his subsequent product sclcc~lon.
Six items in Company Image were framed to measure the d~verstty of Company Image
variable. A pretest with a convenience sample of 30 respondents lndtcatcd that these attrtbutes
provided satisfactory results to identify the consumers' preference or1 Company I~nage var~ablc
In the absence of any other suitable guidelines, thr hc\l \tr:itegy would he to Itlvesttgate tile
nature and structure of the Company Image attributrs t o L I I O M 111e COIISUIIICT'S preferences on tlit!,
variable.
Factor Analysis is used to find out different groups of' attributes wl~lch are Important In
taking the purchase decision of a product. Thc prlnclpal component ar~alys~s finds a linear
combination of attributes explaining the variability, contrtbuted by indiv~dual variables.
As a first step, the correlation matrix for six attrtbutes on C'ompany Image variables is
examined. (Appendix. B ) These results clearly ~ndtcate that all the correlation values are
positive. However, wide variation is present in the observed correlation (0.39 to 0.04). The
Kaiser-Mayer-Olkin measure of sampling adequacy is equal to 0.64 and the Barlett's test of
sphericity ( = 270.79, p = 0.0000). It may be concluded that in the present investigation there
is some shared variance among the six Company lmage attributes.
Factor extraction procedure helps to decide the number of' factors that are needed to
represent the data. It is helpful to examine the percentage of total varlance explained by each
factor. The total variance is the sum of the variances of each variable Several lnethtds have been
proposed for determining the number of factors to be used In the given model. One criterion
suggests that orlly factors that account for the eigenvalue greater than I 0 sliould he included in
the analysis."
In the factor structure for the Compdny Image. three factors have been extracted. The
factors extracted for this variable explain a total variance of 67 9 perccrlt. ;~lthough in only three
factors the eigenvalue is more than 1.0. So, only three factors Iln\,e t>cc~l studled. Slnce the
factors which have eigenval~ies less than 1.0 are of little importance, tlie!l are not taken into
i~ccount for the interpretation purpose of this study. The cumulatl\,e percentage of varlance and
percentage of variance accounted for each factor with eigenvalues ot C'o~npany Inlagc varlahle
are summarised in the Table 4.3
TABLE 4.3
Cumulative Percentage of Variance arid I'erccl~tagc of' Variance
Accounted for Company Image Attributes I)! Each Factor wit11 Ilige~~viilues.
Tucker, Koopman and Linn "Evaluation of factor analytic research procedures by means of simulated correlation matrices" Ysychometnka, Vo1.34, pp.427-436.
Eigenvalues
1 60
1.35
1.12
0.82
0.61
0.49
Attribute
Brand image
Quality Image
Media Image
Reference Group Image
Service Image
R & D Image
Cumulative
Percent of
Variance
26.7
49.2
67.9
81.6
91.8
100.0
Percenl of Variance
Accountetl for
Ry Each Factor
26 7
22.5
18.7
13.7
10.2
8.2
The above results shown in table 4.3 show that 67.9 percentage of total variance is
attributable to the first three factors. The remaining three factors together account for only 32.1
percentage of variance. Thus, it is evident that a model wlth three factors may be adequate to
represent the data.
After identi@ing the number of factors, the ;~ttributes ~ncluded In each factor are
distinguished. Factor loadings are considered as the k s t measure to lr~cludc attributes into a
factor. The factor loading ass~ciated with a specific factor and a speclfic state IS simply the
correlation between that factor and that statement's standardlsed response cores
The table 4.4 shows the factor loading on the C'umpany Imagc v;lrlable (SIX attributes)
TABLE 4.4
Factor Loading Structure for the Company In~agr Attril)utes
The variables with high loadings here are In effect the dornlnant variables determining
the consumer preferences of the people on the Company Image. The first factor with the
eigenvalue of 1.60 accounts for 26.7 percent of the total variance. The factor matrix pattern
reflects a reasonably clear loading structure. The first factor, representing Media Image and
Interpersonal Image created by friends and relatives signifies the two important attributes showing
Attribute
Brand Image
Quality Image
Media Image
Reference Group Image
Service Image
R Br D Image
Factor I
0.088
0.103
0.853
0.851
-0.030
0.129
Factor I1
0.01 I
0 010
0 002
0.03 1
0.802
Factor I11
(,O
0 Oh2
- 0 049
0.080
0.014
consumer preference in Company Image variable. Reference groups can be very potent also very
influential o n consumer behaviour. Consumer researchers have investigated the role of' reference
group influence on product and brand choice for several product categories'. I t comblnes the
concept of public and private consumption goods in relation to luxury and necesslty items
According to their findings for the publicly consumed necesslty Items, the ~ntluence of reference
groups on the brand of the product is strong. In keeping w ~ t h this, the present study also confirms
this particular phenomenon. The influence of reference group on Product and Brand Image is
found to be very high with a high factor loading of 0.851 among the consumer Company I~nage
attributes.
This reference group influence, further supplemented by the MLYII;I Image, with a factor
loading of 0.853, signifies that the Media Image is also Important In purchase prefrrence. l'lie
consumer f o r m certain expectations on the Product influenced by the relerence groups
complemented by media image. Thus, it is obvlous that the Refrrer~cc Ciroup and Med~a lrnage
influence the consumer purchase decision w ~ t h regard to the purchase of I00 cc rnotor cycles
Marketing specialists of this product have to recognlse ;111c! rcspect these sentiments and provide
better product and services to the reference groups (gt8ncr.;~l puhl~c) By d o ~ n p so they can create
a better Company Image. Therefore, this factor can be referred as rhc "KcScrcnce Ciroup Image
Factor" influencing the purchase decision and preference
The second factor is'identified in Service Image and Research and r)evelopment Image
contributing 0.804 and 0.802 factor loading respectively T h ~ s can be rel'cred as "Facility lrnage
Factor". The R & D work is identified as "art for art's sake" and 1s called "scientists isolated
from the realities of the business world". Generally, the K Kr D people lose their deali ism slowly
Their main desire is only to find a scientific solutlon to the marketing problem. The Organisation
William O.Bearden and Micheal J . Etzel, "Reference group influence o n product and brand purchase decisions", Journal of Consurtter Research. Vol . 9, Sept. 1982, p. 185.
should integrate the R & D ' ~ e ~ a r t m e n t with other departments such as sales, marketing and
production and data processing. Hence, the present study stresses the Importance on R LYr D
Image. Further, the respondents' preference is towards the product de\,eloprnent and the creation
of Company Image through better service facilities.
The [hlrd factor identified from the Company Image 1s Brarid Irnage and Qual~ty lmagc
with an eigenvalue of 1.12 and variance of 18.7 percent Thr fac~or s t rumre shows the high
factor loading of 0.81 and 0.76 on Brand Image and Quality lmagr 'l 'h~s tactor, therefore. can
be considered as "Quality lmage Factor". In order to form Impressions ot products. consumers
process additional stimuli that are not the actual physlcal characterr\t~c\ of' the product ~tself
These features, often called 'extrinsic cues', could be packag~ng ~ll;~r:~cter~.rtlcs, advertising
messages, statements of friends and many other piecc of informarlor1 from a w ~ d e variety of'
sources." Thus, the information from the Brand lrnage and Qual~t!, Irrllrgc I S Identified asthc third
factor in the Company Image variable.
To summarise consumer preferences, the Comp;~ny Irnage r l~a~nly depends on word of
mouth from the reference groups of friends and relat~\c~s \uppleniented t>y thc advertlsemeot
through varlous media. Since automobiles require furthrr service\ ;lttisr \airs. the next prlorlty
of the customer goes to the Service lmage and R K: D 1111agc of the company ( 'u~torner
preference on Company image is comparatively low for Brand Image and Quality Image.
Factor Structure for the Product Feature Preferences
Consumers often judge the quality of a product or service on the basis o f a variety of cues
that they associate with the product. Some of thew cues are Intrinsic to the prcduct or services;
Donald F.Cox, "The Sorting Rule Model of the Consumer product Evaluation Process," in Donald F.Cox (Eds) Risk taking and Infom~afion handling in consumer behaviour, Cambridge, M A , Harvard Business School, 1967, pp 167-179.
others are extrinsic, such as price, store image, service environment, brand image and
promotional message. Such cues provide the basis for perceptions of the product and service
quality either s~ngly or collectively.
Consun~ers' attempts t o evaluate directly the products' physical attrihutes are otikn called
intrinsic cues, \uch as size, shape and grade, quality and perfonllancc of tlie product. Some
evidences suggest that consumers' product perceptions are Inore Ilkel!. to be ~nflucrlced by
extrinsic cues when the product is complex in nature However, little is known about how
consumers select such cues to form interpretations, or what condit~ons ~ntluence this process.
Certain factors such as Product Quality. Product Pcrformancc. IJrotluct Appearance and
Product Efficiency can mitigate the strength of the perce~ved product quality relationsh~p and
actually overshadow it for some products in some situations 111 orrlcr to arrive at sornt:
conclusions the Factor Analysis is carried out for thc 100 cc motorcj.cles with 12 product
Characteristics: Product Appearance, Technology, tmergency Necd bulfilment, Teclin~cal
Specification, Proximity of Outlet, Range of Models. I'rohlcnl Frccnc.ss, l.uel Lftlc~ency, P~ckup
and Brake Control, Seating Comfort and Citylhighwa! I { I L I I I I ~
The correlation matrix given in Appendix- H for the twclvc prcduct attrihutes clcarly
indicates the positive correlation among the variables w ~ t h an exception of Fuel Effic~ency and
Range of Models. However, there is a wide variat~on ranglng from 0 01 to 0 498 in the observed
correlation among the 12 variables. The Kaiser-Mayer-Olkin measure of sampling adequacy
(0.71) which is meritorious for further Factor Analys~s, and Barlett's test of sphericity ( x2 =
963.54, p = 0.0000) indicate the presence of some sharrd varlance among the twelve items.
The uses of Factor Analysis are mainly exploratory depending on the major objectives
of the study. The extraction step involved in the Factor Analysis is to determine the minimum
number of common factors that would satisfactorily produce the correlations among the observed
variables. One of the most poplar cfiterir for addressing the number of factors in analysis is to
retain factors with eigenvdues greater than 1 when the correlation matrix is decomposed. This
simple criterion seems to work well1. In Table 4.5. the results of factor analysis on the reduced
correlation matrix are given for Product Feature variable.
TABLE 4.5
Cumulative Percentage of Variance and Percentage of Variance
Accounted for Product Feature Attributes by Each Factor with Eigenvalues.
' Jae On Kim, C k k s W.Mueller, Factor analysis, Statistical methods and pmctical issue$, California, Sage Publications, 1982, p.43.
Attribute
Product Appearance
Technology
Emergency Needs
Technical Specification
Proximity of Outlet
Range of Models
Problem Freeness
Fuel Efficiency
Pickup / Brake Control
Environment Safety
Seating Comfort
Citythighway Riding
Cumulative Percent
of Variance
23.1
37.4
48.1
57.7
65.8
73.1
79.0
84.6
89.7
93.7
97.0
100.0
Eigenvalue
2.779
1.706
1.280
1.160
0.960
0.872
0.71 1
0.668
0.619
0.478
0.397
0.354
Percent of
Variance
23.1
14.2
10.7
9.7
8 1
7 3
5.9
5.6
5.2
4.0
3.3
3.0
Four factors have been extracted by using the eigenvalues. Four eigenvalues are greater
than 1.0 and together explain 57.7 percent of total variance (eigenvalues 2.777, 1.706. 1.280
and 1.162). A scree plot (Appenix - B) of the eigenvalues also seems to support the four factor
solution. The varimax rotated factor pattern also reflects a reasonably clear loading structure
with the four factor structure.
The following table 4.6 provides the factor matrix and their corresponding factor loading
in order to make the interpretation easy.
TABLE 4.6.
Factor Loading Matrix Structure for the Product Feature Attributes
The varimax rotated factor for the same variables after sorting into the factors are given
in the table 4.7.
Attribute
Product Appearance
Technology
Emergency Needs
Technical Specification
Proximity of Outlet
Range of Models
Problem Freeness
Fuel Efficiency
Pickup / Brake Control
Environment Safety
Seating Comfort
Citythighway Riding
Factor 1
0.46308
0.45242
0.55299
0.48983
0.26547
0.33947
0.41623
0.52448
0.42225
0.49619
0.57314
0.65254
Factor 2
0.38785
0.15483
-0.17647
-0.21203
0. 60890
O.Oh134
0.49095
-0.34663
-0.49733
0.06604
0.18546
0.03459
Factor 3
0.5167
0.58b49
0.48266
0 05998
-0.21793
-0.10401
-0.02245
-0.21733
-0.032%
-0.31387
-0.33446
-0.34188
Factor 4
0.20602
-0.12531
-0.08252
0.56152
0.37686
0.29640
0.46552
0.301 70
0.25816
0.09683
-0.36927
-0.17850
TABLE 4.7.
Varimax Rotated Factor h d b g Structure for Product Feature Attributes (Sorted)
Product quality does not begin in the factory; i t only ends there. The person who is
making a buying decision takes quality into account quite naturally. But he rarely makes the
decision influences by quality alone. Utility of the product is also an important factor for the
consumers in their product choice. The quality and utility of a product should not be confused
in sales presentations. Utility comes first in the eyes of the consumers. They are concerned about
the use soon after they purchase the product. Quality is the innate value of the product. I t may
enhance the value of an article compared to another which might be equally useful.
Attribute
Seating Comfort
CitylHighway Riding
Technical Specification
Environrrrent Safety
Problem Freeness
Fuel Efficiency
Pickup / Brake Control
Product Appearance
Technology
Emergency Needs
Proximity of Outlet
Range of Models
Factor 3
0.06659
0.065 18
0.29698
0.00905
0.08664
0.01252
0.09 107
0.74466
0.73986
0.66384
0.03743
0. 18434
Factor 4
0.12856
0.14346
-0.39471
0.17453
0.00800
0.07620
0.12859
0.19012
0.1865 1
-0.18219
0.78354
0.76425
Factor 1
0.76800
0.70576
0.599.18
0.54370
0.00960
0.25393
0.11467
0.17423
-0.07092
0.17872
0.12437
0.17040
Factor 2
0.01799
0.22983
0.02790
0.17973
0.78964
0.67843
0.67483
-0.16313
0.16690
0.26620
-0.00473
-0.64710
The changing environments of the consumers create many superior needs and it is
:ontrolled by many other determinants. The odds are that they no longer meet the highest quality
specifications. Satisfying the emergency needs of the consumer is a must. Fresh prospective can
redefine quality as there are more efficient and effective alternative ways of meeting customer
needs. The factor analysis on the Product Feature also confirms this state that the first factor
identified with Product Feature are attributes of Seating Comfort, Citylhighway Riding Comfort,
Matching to Technical Specifications and Problem Freeness. These variables have high factor
loadings of 0.76800, 0.70576, 0.59918 and 0.54370. These attributes can be grouped in factor
I and termed as "Product Comfort Factor". Thus. Comfort andlor Utili ty Factor is identified as
important purchase preference among the Product Feature variables.
The Second factor given in the same table is designated "Product Performance Factor"
on the basis of positively loaded variables explaining the variance of 14.2 percent and has an
eigenvalue of 1.706. Three variables in this category are itnportant with high factor loading. The
data set through the factor loading indicates that among the Product Feature variable Problem
Freeness, Fuel Efficiency and Pick up and Brake Control attributes art. important attributes is this
category. Thus, performance is identified as another important factor for the purchase decision
of the consumers.
Product performance, in the form of Problem Freeness dominates in the factor structure
with a high factor loading of 0.78964. Possession of motor cycles is considered to be a
convenient mode of tramport because consumers of motor cycles want to be away from the
problems of public transportation. Thus, the result of consumers' preference of motor cycle
confirms this particular phenomenon. The second attribute included in this factor is Fuel
Efficiency with a factor loading of 0.67843. The dominance of 100 cc motor cycles in the present
market share is mainly due to this attribute. The third attribute is Pick up and Brake Control of
the motor cycle with tbe frrctor loading of 0.67483. These observations indicate the awareness
of energy c o m i o n rrd ssfcty among the consumers of 100 cc motor cycles.
The third factor, identified in table 4.7, shows a significant loading for "Appearance".
"Upto date Technology" and "Emergency Needs" which pertain to the additional necessities of
the product. This particular factor can be termad as "Additional Necessities Factor". In Managing
for Results - Peter F.Drucker identifies the marketing realities as "The customer rarely buys what
the business thinks, it sells him. One, reason for this is. that nobody pays for "~roduc;" what is
paid for is "satisfaction it can produce". But nobody can make or supply satisfactions as such -
at best, only the means to attaining them can be sold and delivered"' Thus, the product delivery
should not be confined, to the quality alone, other necessities also, have to he given due
consideration.
The fourth factor in Product Feature variable is identified w~th two attributes such as
Nearness of Supply and Range of Models with the factor loading of 0.78354 and 0.76425
respectively. Each class of customers has different needs, wants, hahits and expectations, value
concepts etc.. Yet each has to be sufficiently satisfied at least not to veto a purchase. This is
possible only by providing different range of models to suit the varied needs of the customers.
Supplementary to this, the products should be available ar rhr nearest place These two important
variables become the fourth factor in the Product Featurc category. This factor may be termed
as "Product Availability Factor".
To sum up, all the products do not have equal potential for consumer acceptance and also
there is no precise formula. by which marketers can evaluate the product acceptance, the
researchers have identified (1) relative advantage, (2 ) compatibility (3) complexity (4) trialability
and (5) observability as the five product characteristics that seem to influence the consumer
Peter F.DNclrcr, Iltaff4ging for Results, Economic Tasks and Risk Taking Decisions, MdamkIsdcas, Allied Publishers Ltd., 1970, p.94.
:ceptance of the p r ~ d u c t . ~ However, the present study provides four important Product Feature
lctors as essential for the purchase decision of the consumers. The factors include: ( I ) Comfort
nd Utility Factor (2) Product P e r f o m Factor (3) Additional Necessity Factor and (4)
'roduct Availability Factor in order of priority.
pactor Structure for the Customer Support Senice Preferences
The idea of service management as a basis for achieving the competitive edge and
~uilding business success became the theme of the 1980's.''' The secret of business is to just
:xceed what the customer expects. The first pursuit used to be customer satisfaction. Then. i t
should be customer pleasure or delight. Quality service must be recognlsed as a bottom-line issue.
However, it must not be perceived as an optional extra. For every company quality service
programmes are seen as an investment and not a cost. The starting point 1s the recognition of
what has been good about the corporate past and what is good in the present In order to identify
critical success factors for the Customer Support Service facilities among the 100 cc motor cycle
owners, nine important service facility attributes are identified and trarislated into questions to
show the customer preference on this service facilities. Based on this preferences, further analysis
is carried out.
The customer finds it very diffiarlt to evaluate the quality of services than the quality of
product. This is true because of certain distinctive characteristics of services: their intangibility,
their variability, the fact that services are simultaneously produced and consumed and their
perishability. Consumers rely on surrogate cues (extrins~c cues) to evaluate the service quality.
Evem M.Rogers, Diffusion of innovations, 3rd ed. (New York: Free Press, 1983); and Hubert. Gatignon and Thomas S. Robertson, "Innovative Decision Processes, " in Thomas S.Robertson and Harold H . Kassarirajan, eds. , Handbook of Consumer Beirariour Englewood Cliffs, NJ: Prentice Hall, 1991, pp. 3 16-348.
lo Barrie Hopson and Mike Scally , I2 steps to success through service New Delhi, Excel Books, 1994, p.15
For example, in evaluating the service quality of a doctor, the quality of the office and the
furnishing of consultation room and pleasanmess of the reception accorded - all contribute to the
patient's overall evaluation of service quality.
Various researchers have devoted themselves to the study of how consumers evaluate the
Customer Support Service quality. Conclusions have been drawn that the service qual~ty thal a
customer perceives, is a function of the magnitude and direction of the gap between the
customer's expectations of service and the customer's assessment of servlce sctually delivered.
Thus, the findings of the consumers' expectation provide a guidelirie for co~nparison in order to
know the level of consumer satisfaction on Customer Support Service. 011 this line the present
investigation concentrates on dimensions of consumers' preferences on Custolner Support Service.
Thus, the factor analysis on Customer Support Servlce is carried out and the results are
discussed.
The correlation matrix for the nine intentions given in append~x B are cxamlned and the
results show that the values are positive. However clear variation 1s present i n the observed
correlations, which ranges from 0.151 1 to 0.453 1 . The tia~ser-Mayer-Olkin measure (KMO) of
sampling adequacy (0.81) which is considered mnerltorlous for turtller factor analysis, and
Bartlett's test of sphericity x2 = 942.98, with p significance = O.OONX) indicates the presence of
some shared variance among the nine variables of Customer Support Service facilities. Thus,
further factor analysis is carried out to identify the important factors to represent the data.
The results of the factor analysis on the reduced correlation matrix i.e. squared multiple
correlations on the diagonals in place of one are given In table.4.8.
TABLE 4.8.
Cumulative Percentage of Variance and Percentage of Variance
Accounted for Customer Support Service Attributes by Each Factor
with Eigenvalues.
The eigenvalues of the two variables are greater than I .OU and together explain 47 9
percent of the total variance (eigenvalues are 3.225 and 1.088). Further the scree plot (Append~x
- B) of the eigenvalues seems to support the two factor solution.
Attributes
Dependable Service
Employee Behaviour
Service Centre Appearance'
Record Maintenance
Problem Appraisal
Complaint Registers
Working Hours
Delivery Schedule
Maintenance Awareness
The table 4.9. provides the factor matrix for the Customer Support Service variable.
Eigenvalue
3.225
1.088
0.944
0.871
0.731
0.676
0.553
0.464
0.446
Percent of
Variance
38.6
12.6
10.5
10.0
7 .6
7 .6
5.6
4 .3
3 2
Cumulative
I'crcent of Variance
38.6
51.2
61.8
71.7
70.4
86.9
92.5
96.8
100.0
TABLE 4.9.
Factor Loading Matrix StrWwe for the Customer Support Service Attributes
The varimax rotated factor matrix is given in tablr 4.10 in order to make the conclusions
in an easy manner. The sorted varirnax rotated factor structure glves clear extraction of two
factors in the Customer Support Service facilities.
Attributes
Dependable Service
Employee Behaviour
Service Centre Appearance
Record Maintenance
Problem Appraisal
Complaint Registers
Working Hours
Delivery Schedule
Maintenance Awareness
Factor 1
0.4638
0.5444
0.5634
0.6326
0.6233
0.6406
0.6291
0.6499
0.6157
3
Factor 2
0.1575 .
0.4852
0.5272
0.1 108
0.0616
0.4197
0.4197
0.4151
0.0973
TABLE 4.10..
Varimax Rotated Factor Loading Structure for the
Customer Support Service Features (Sorted)
The Variamx Rotated Factor pattern reflects a reasonably clear load~ng structure with all
cross loadings. The first factor, representing the service attributes of Ma~nterlance of I'omplalnt
Register, Convenient Working Hours and Prompt Delivery after servlce with high factor loading
of 0.75363, 0.75024 and 0.74202 respectively. This suggests that the facilities should be prompt
and timely. This factor therefore can be termed as "Promptness in Servicc Factor".
Attributes
Delivery Schedule
Complaint Registers
Working Hours
Service Centre Appearance
Employee Behaviour
Record Maintenance
Maintaince Awareness
Problem Appraisal
Dependable Service
A customer service mission statement establishes the vision to which the organisation
should aspire for. Every organisation should realise it exists to serve its customers only. The
second factor extraction clearly identifies two factor indications. The first factor specifies that
service hours and service facilities should be prompt. The customers' preference on Customer
Suppon Service indicate this phenomenon.
Factor 1
0.75363
0.75024
0.74202
0.02803
0.04416
0.37058
0.36815
0.39869
0.21796
Factor 2 .
0.16372
0.15386
0. 14574
0.77107
0 72798
0.52459
0.50302
0.48302
0.43865
Good service is not just smiling at your customers but getting your customer to smile at
you. As business grows, it is easy for any business to lose sight of basics. The second factor
extracted clearly indicates this consideration. The second factor includes the service attributes
such as Dependable Service, Courteous Behaviour, Service Center Appearance. Record
Maintenance and Maintenance Awareness. The first factor is related to servlce facilities whereas
the second factor identified the quality of service facilities. Since. this provision of service alone
is not sufficient, the service facilities provided should be of high quality In order to make delight
customer or atleast to satisfy, their need. Providing quality service can be possible by adhering
the following rules:(l) making the people feel special. (2) managing tirst four and last two
minutes of servlce transaction, (3) demonstrating a pos~tive attitude. (4) communicatrng clear
messages and (5) showing high energy.
The present factor structure clearly indicates th~s type of behav~our phenomenon of the
100 cc motor cycle customers with a factor loading of 0.77107. 0 72798, 0 52459, 0 50302,
0.48302 and 0.43865 on Dependable Service, Courteous Service, Appealing Service Centre,
Record Maintenance and Creating an Awareness for t h t Ma~ntenance respectively. Thus, the
various quality improvement attributes of service facilit~es arc cons~dercd as the second purchase
preference evinced by the consumers of 100 cc motor cycles.
In short, the organisations should provide two important factors essential for Customer
Service in order to excel in the customer support service. The factors include the Better Customer
Support Service facilities and Timely Service.
Factor Structure for the Delivery Terms Preferences
The problems related to Delivery Terms tend to be traditional and more common in the
case of automobile industry. When one wants to investigate the superior delivery needs of the
consumrs, the following five need areas are to be fulfiled: Time need, The Specialty need, The
Quantity need, Location need'and The Hassk free need. The focus is on finding out the consumer
priorities on these need areas. In order to find out customer priorities on this part~cular aspect
the factor analysis is carried out on this particular need variable. The foregoing discussion throws
light on particular direction of identifying and grouping of Delivery Ter~lis attributes.
The correlation matrix for the five intentions are thoroughly examined (Appendix- B ) .
All the values are positive excepting User Guidance with Prompt Delivery and Colour Choice.
However, clear variation is present in the observed correlations, which ranges from 0.02235 to
0.26552. The Kaiser-Mayer-Olkin measure (KMO) of sampling adequacy (0.64) which is
considered as middling for further factor analysis, and Bartlett's test of sphericity x2 = 108 28.
with p significance = 0.0000 indicates the presence of some sllared varlance among the five
variables of Delivery Terms attributes. Thus, the further factor analys~s is therefore, carried out
to identify the important factors to represent the data.
The factor structure extracted for the Delivery Terms variable consists of two factors o u t
of the five attributes taken for the study. Only two factors ha\,r eigenvalues greater than 1.0 and
hence others are not significant. These two factors accounted for 52.7 percent of total variance
which may be seen in table 4.11.
TABLE 4.11
Cumulative Percentage of Variance and Perceritage of Variance
Accounted for Delivery Terms Attributes by Each Factor with Eigenvalues.
Cumulative Percent of
Variance
30.7
52.7
70.5
86.9
100.0
Attributes
Prompt Delivery
Colour Choice
Choice of Outlet
User Guidance
Easy Availability
Eienvalue
1.53455
1.lOOOd
0.89092
0.82006
0.65387
Percent of
Variance
30.7
22.0
17.8
16.4
13.1
In order to have a detailed discussion of factor extraction and the attrihutes to be included
in each factor, the factor loadings on each variable are provided in the table 4.12.
TABLE 4.12.
Factor matrix structure on Delivery Terms attributes
TABLE 4.13.
Varimax Rotated Factor Matrix Pattern for the Delivery Terms Variallle after Sorting
Attribute
Prompt Delivery
Colour Choice
Choice of Outiet
User Guidance
Easy Availability
The first factor, which is evident from the table 4.13, Prompt Delivery, Easy Availability
of the Product and Colour Choice have got high positive loadings of 0.73334, 0.66149 and
0.58012 respectively. Delivery gives the chance to the organisation to prove their strategy's
viability impressively and quickly. In addition, delivery is a good category to learn the language
of the superior needs of the consumers, to become farnillar with the voice of the customer and
to understand how to interpret that voice and translate i t Into various tactics.
Factor 1
0.72320
0.51016
0.56481
0.09465
0.65062
Attribute
Prompt Delivery
Easy Availability
Colour Choice
User Guidance
Choice of Outlet
All customers who have purchased consumer durables wish to be familiar with the need
to have their product set up or installed immediately and want to use. The product must be made
1
Factor 2
- 0.15921
- 0.2925 1
0.49652
0.84923
- 0.14824
Factor 1
0.73334
0.66149
0.50934
- 0.20730
0.3563 1
1
Factor 2
0.10285
0.08785
0.58012
0.82896
0.66225
ready for the buyer to use it immediately. The consumer finds satisfaction when it is promptly
delivered and available for their use in time.
Another element of product delivery is the easy availab~l~ty of the prrduct. Indian
customers of automobile industry are facing a lot of difficulties on this part~cular attrihute. The
customers of automobile products have experienced a patlence waiting tinie concept. Tlius, most
of the consumers' preference showed the high loading of 0.66140. o n this particular attribute.
It often happens that a buyer of one product beconies a candidate for other sorts of
options. The buyer of one product wants to know the details of the range of ~nodels available i n
that product line, as a result the organisation should have wide range ~nodels in order to draw
customers of varied needs into the store and makes effective selling The tll~rd variable "Range
of Models" is considered as next preference priority of the IOU cc noto or cycle customers with
a positive factor loading of 0.58012. These three attributes can he termed as "Easy Availability
Factor" which is identified as the first factor in the Delivery Terms varlahle
The second factor identified includes, the Dellvery Terms attrlbures sucll as, Change of
Agency and Change of Colour in case of difficulty in supply l'he factor loadings corresponding
to these attributes are : 0.66225 and 0.58012 respectively. This particular factor signifies that the
customers are determined about their delivery from a particular outlet I f they experience any
difficulty in prompt delivery, the next priority of change of colour option and change of outlet
option are to be given to the customers. This factor may be identified as the "Change Over
Option Factor".
Thus, the Delivery Terms also influence the purchase preference of the consumers. The
two factors which are important in the Delivery Terms include "Availability Factor" and "Change
over Option Factor". Availability includes the product availability with a wide range of models
and with varied choice and colour. Change over option includes change of agency and change
of colour option in the event of difficulty in supply.
Factor Structure for the Product Price Variable
A number of research studies support the view that consumers rely on prlce as an
indicator of product quality. A comprehensive review of literature indicates that, despite mixed
findings, a positive pricelquality relationship does indeed exist . However when other cues are
available (brand name, store image etc.) they are sometin~es more ~ntluential than price 111
determining the perceived quality. In order to investigate the influence of prlce as an indicator
in consumer satisfaction, an analysis was carried out in identifying the prlce factors to find out
which are important for contributing to consumer satisfaction. The following discussion ma~nly
throws light on these lines.
The correlation matrix for the six preference intentions on the Product Price attributes
are examined from the table given in Appendix- 0. All the values are positive excepting Cost of
R & D with Notification of Price Alteration. However clcar var~ation is present in the observed
correlations, which ranges from 0.01038 to 0.37224. Thc Ka~ser-Mayer-Olk~n measure (KMO)
of sampling adequacy (0.62) which is considered as middling for further factor analysis, and
Bartlett's test of sphericity = 108.28, with p significance = 0.0000 indicate the presence of
some shared variance among the six consumer Product Price attributes. Further factor analysis
is thus needed to identify the important factors to represent the data
The factor structure extracted for the Product Prlce attributes consists of two factors out
of the six attributes taken for the study. Only two factors have eigrnvalues more than 1.0 and
hence others are not explained here. These two factors account for 5 1.7 percent of total variance
which may be seen in table 4.14.
TABLE 4.14.
Cumulative Percentage of Variance and Percentage of Variance
Accounted For Product Price Attributes by Each Factor with Eigenvalues.
In order to analyse in detail the factor extraction and attrihutes 10 he included in each
factor, the following tables containing the factor loadings on each variable are provided.
Attribute
Competitive Pricing
Notification of Price Alteration
Pollution Awareness
Loan Facility
Cost of R & D
Resale Value
Table 4.15.
Factor Matrix Structure on Product Price Attributes
Eigenvalue
1 A7186
1.18626
0.97495
0.78930
0.65273
0.52490
Percent of
Variancc
31.2
20.5
16.2
12.5
10.9
8.7
Factor 2
0.61691
0.66216
- 0.1 1826
- 0.19251
- 0.52251
0.20777
Attribute
Competitive Pricing
Notification of Price Alteration
Pollution Awareness
Loan Facility
Cost of R & D
Resale Value
Cuniulative
Percent of
Variance
31.2
S l 7
67.9
80.4
91.3
l(M.0
Factor 1
0.25520
0.47628
0.71940
0.74002
0.61579
0.36813
Table 4.16.
Varinmx Rotated Factor Matrix Pattern for Product Price Variable after Sorting
The first factor, which may be seen from the table 4.16, Environ~nent Awareness. R &
D Cost and Loan Facility have got high positive loadings of 0.h904h. 0 78920 and 0.74359
respectively. As Product pricing policies influence the purchase decisions of the consumers, the
companies should plan their pricing policies in a proper Inanner so as to hc more competitive in
the market. The pricing policy gives the chance to the organisatlon to prove the~r market
strategies viable. Thus, the first factor identified in the PrtKfuct Pr~ce category shows that P r~c~ng
Terms is considered as an important factor for the consumers for their purchase dec~sions In
addition to the product pricing. That is consumers are aware of the h~gher pric~ng for the
Pollution Control Equipments, further they are ready to spend additional a~nounts on the R % D
so that the cost effectiveness can be made possible in the near future. The consumers are ready
to spend some extra money for the Loan Facility.
Attribute
Cost of R & D
Loan Facility
Pollution Awareness
Notification of Price Alteration
Competitive Pricing
Resale Value
Another element of Product Price attribute is Competitive Pricing, Resale Value of the
Vehicle and Notification of the Price Change in the event of delay in delivery, Indian customers
of automobile industry give consideration to these attributes with the factor loadings of 0.80836,
0.66444 and 0.35652. In addition to pricing, some more terms of the Product Price itself are to
Factor 1
0.78920
0.74359
0.69046
0.1088 1
- 0.06500
0.227 12
Factor 2
- 0 17142.
0 17823
0.23405
0.80836
0 66444
0.35652
be given consideration. The consumers give more priority for the competitive price of the product
and resale value of the product. Further, they give importance to notificaticln of alteration in price
in case of delay in the delivery of the product. This particular factor may hc termed as "Product
Price Value Factor".
The major findings from the factor analysis of the study and the conclusions drawn are
sununarised below: The thirty eight attributes included in five variable categories arc reduced into
different groups of 'Factors'. These factors provide informations regarding consumer ?references
and priorities about the product. In addition, these factors can guide the industry as regards to
the characteristics which are important to be incorporated into the produc~ and in formulation of
advertising themes. The factor extraction process provided th~rteen factors as Irllportant purchase
preferences among the consumers of 100 cc motor cycles in Tatnil Nadu. Tlirse factors, In turn.
identified what as prominent characteristics which contribute to consumer satisfaction The
thirteen factors are : Company Image includes three, Product Featurc irlcludcs four f'iictors.
Customer Support Service comprises of two factors, Delivery Tcrm~ contributes two lniportant
factors and Product Price further contributes two factors Sor purchase preferences Among the
above thirteen factors Product feature factors are found to bc the l'rlme factors In determining
the level of satisfaction of the consumer.
Discrimination of Consumer Preferences Between Urban and Semi Urban Croups
The economic and demographic structure of the marketplace form the foundation of
consumption. Studies in consumer behaviour have always highlighted the importance of
demographic and economic variables in consumers' choice of products. Analysis of demographic
trends should therefore receive high priority for various marketing strategies. As an attempt to
find out the changing profile of the Urban and Semi Urban consumers an attempt has been made
in the second pan of this chapter to find out the difference, if any, between the urban and semi
Factor Scores
Factor scores can be obtained by multiplying standardised value of the
factor by the co~~esponding factor score co-efficients The estimation of factor scores for
each case has been done by using regression method. 'I'he factor score co-efficients for the
present data are given in Appendix. The estimated factor scores for the thirteen factors
identified earlier were saved in a separate working file and used in the subsequent analysis
Factor analysis can sometimes be useful for other analyses of dependence
structures, where the predictor are both numerous and highly correlated The
predioators are first factor analysed and the fhrther analvs~s of criterion variable can be
carried out on the full set of factor scroes." The factor scores saved in the present
study were used to rerun for the discriminant analysis The results of such analysis are
given under the discriminant analysis summary elsewhere
" Paul E.Green and Donald S Tull., Research for Marketing Decisions., Prentice Hall,lnc., Englewood Cliffs, New Jersey, 1978, p.437.
urban consumers in terms of purchase preferences on 100 cc motor cycles. For this purpose
Discriminant Analysis is used to discriminate between the two groups of populations.
Discriminant Analysis is useful in situations where a total sample is div~ded into know11 . groups based on some classificatory variable, and \\then the researcher IS interested in
understanding group differences or in predicting correct belonging to a group of new sample
based on the information on a set of predicator variables
Discriminant function is one of the most widely used lllultlvar~ate procedures 111
behavioural research. I t is a powerful tool which is regarded as a univariarc prohlem related to
multiple regression or a multivariate problem related to statistical test. 'T'he process of
discrimination that the statistical tool embodies is distinct from the prtress of classification tilt.
Factor Analysis embodies. The problem in discriminant analysis is finding a linear combination
of variables that produce the maximum difference between the groups (usually two) considered
for discrimination.
A simple linear discrimination function transforr~ic an or~giniil set of measurements of n
sample into a single discriminant score. That score, or transt'ormcd variable, represents the
sample's position along a line defined by linear discriminant function 'fl~erefore, tlie d~scrim~nan!
function is a way of collapsing a multivariate problem down into a problem which involves only
one variable.
The discriminant index is the point along the discriminant function line which is exactly
halfway between the centre of the group 1 and the centre of group 2. The groups themselves have
a mean each, which is in essence the centre of the groups. The two groups discriminated against
each other therefore have two means which define the centre of the two original groups along the
discriminant function.
It is possible to test the significance of the discriminatnt function. To test the significance
of the separation between the'two groups, five basic assumptions about the data are necessary to
test the significance of the discriminant function. They are :
the observations in each group are randomly chosen.
the probability of an unknown observation belonging to either group is equal
variance are normally distributed within each group.
the variance and covariance of the groups are equal in sire, and
none of the observations used to calculate the d~scriln~nant function was
misclassified.
However, there are two major ways in which discriminant functlon may he attempted.
One is from the angle of observations, so that each of the observatic~ns may k assigned a
discriminant score which would help place it along the linear deiscr~nllnilnt tinction line. In thls
way, it would be possible to work towards the misclassificat~ons in thc two sets of' observations
analysed. The alternative is to find the discriminant co-cfficlcnts fo r the variables used in the
discriminant analysis. In this study the latter method is used lo ohtalrl discrllnirlant co-efficients
by using the variable responses of the individuals.
When a discriminant analysis is carried out, the goal is to develop a model that will result
in a large proportion of the cases being correctly classified. The discrirr~inant equation can then
be used to predict to which class a new case will belong, or more importantly, to demonstrate
which variables are most important in distinguishing between the groups. T h ~ s provides the scope
to the manufacturers to concentrate on that particular vanable. There are various methods for
variable selection. In the present study, Mahalanobis Distance Measure 1s used to identify the
variable selection.
The test of significance may be daived from a 'distance' measure. This distance measure
is the DZ known as the Mahalanobis' d i i (or the generalised distance). This distance is
calculated from the two multivariate merns and is expressed in units of the pooled variance.
Further, the test may be transformed to an F-test. The null hypothesis tested by this stat~stic 1s
that the two multivariate means are equal or that the distance between them is zero. If the means
are well separated and the scatter about tk means is small, the discrimination will tx relat~vely
easy. Thus, the variables which have the minimum D2 are considered as llliportant variables for
discrimination.
The accuracy of the discriminant analysis can be known by a clahsification matrix (also
called a confusion matrix). The percentage of exact class~fication is a nlcasure of the accuracy
of the functions. However, only by testing a model on the data used to develop, a biased estlrnate
of its accuracy can be obtained. Therefore, it is desirable to keep a Iloldout samples when
conducting a discriminant analysis. That is as many as 50 percent of the original sample are not
used to develop the discriminant model. Instead they are held out and used to develop the
confusion matrix. This approach gives a more valid estimate of thc accuracy of discrirn~nation
function.
In order to arrive at correct conclusions the discrimination analys~s was carried out in two
parts. The first level analysis is to identify the variables which prov~des scope of discrimination
between urban and mi-urban consumers. The second level analysis 1s done by using the
classification matrix to know the effect of the discrimination on the selected variables.
Groups for Discriminant Annlyshr
Thus, as has been made clear in tbe earlier chapter, pertaining to consumer satisfaction
behaviour with respect to five variables chosen for the study with regard to two types of areas
: urban and semi urban. The pmcqd ida behind the selection of areas - urban and semi urban -
being that behaviour would be quite different because of the milieu differences. In discriminating
the two groups of samples, the urban and semi-urban, five groups of variables have been taken;
the first group of variables are the same as that used in factor analysis: six in Company Image,
twelve in the Product Feature variable, nine in the Customer Support Service variable, five in
the Delivery Terms variable and six in the Product Price variable. These variables are
discriminated against urban and semi-urban areas. The results of the analyses are such that each
of the five groups of variables have a mean, standard deviation, discriminant ctxfficient. and the
differences are deduced from differences in the five measures. Also, the group d~fference are
measured by D'.
The computations were made individually for the five variables uung the urban and seml
urban categories as predicator variables. In all, five runs were made and the results and
inferences derived are summarised.
Discrirnina~~t Analysis for Company Image Variable
The results derived from the discriminant analysis ol the consumers' responses from the
urban and semi urban samples for the Company Image var~ables have glven surprisingly very
small differences in all cases which is contrary to the perception of d~ffering milieus. fiowever.
in all cases the null hypothesis is that there is no difference between the groups (urban and semi
urban) discriminated against for the Mahalanobis distance function gives the following results.
Table 4.17.
Mahalanobis DZ Discriminant Distance Score for the Company lmage Variable
The distance values, although very small, indicate a difference alllong the satnples from
urban and semi-urban areas. In sum, there is a distinctly percepttble difference, between Urban
and Semi Urban groups as regards Brand Image, Service Image and R & I) Image. To conclude.
the urban and semi urban groups discriminate very much in Brand Image. Service lmage and to
a considerable extent in the. R & D Image. The difference is very low with regard to the
Reference Group Image and Media Image.
Attributes
Brand Image
Quality Image
Media Image
Reference Group Image
Service Image
R & D Image i
It is apparent that there are several misclassiticat~ons. 7'l11s means that some consutners
who belong to urban areas in fact belong to semi urban areas and vlcc vrrsa in their preferences
Some people, despite living in one kind of milieu, express in effect the taste of another kind ot
milieu. As seen from the classification table, a majority of the consumers cluster around the
means, which is as should be in a normal distribution. It can be understad from the classification
Table 4.18.
Mahalanobis D2
0.10369
0.09347
0.05813
0.07898
0.10293
0.09964
Variable Selection
Ranking Order
6
3
I
2
5
4
Table 4.18
Discriminant score Clas.sGcation Matrix on two groups for Company Image
Percentage of "Grouped" Cases Correctly Classified : 56.85 pcrcerlt
Actual Group
Urban
Semi Urban
Judging from the above classification it is clearly seen that the two predicator variables
versus the six Company Image variables the cases correctly classltied percentage works out at
56.85. In the remaining 43.15 percent of cases there 1s an overlapping of att1tudt.s in thelr
preferences. Thus, a high degree of discriminationexists hctween these two groups of'consumers.
In short, 284 cases out of 499 cases are in their respectl\c croups 'I'he rr~~laining 215 cases out
of 499 cases overlap in their preferences.
Discriminant Analysis for Product Feature Variable
The development of new products and greater accessibility of older products and services
can rapidly change the demographic profile of the market. There is an interaction between the
product and service being offend to the consumer and thelr respective demography. This
interactive nature of demographic variables can lead to some major problems for the marketer
in the decision making process. The results of discriminant a~alysis for consumers from the two
groups of the sample, namly urban and semi urban for the Product Feature variables indicate
only a very small difference in Pll cases. This is contrary to the general perception of that
differing attitudes of Product k t u r w between urban and semi urban may exist due to their need
value. However, in all cases thm is a small difference between the groups ( urban and semi
Total
154
245
Predicted Group Membership
Urban
I55 (61.0%)
116 (47.3%)
Semi Urban
99 (39.0%)
129 (52.74)
urban) discriminated against the Mahalanobis distance function which gives the following results.
(Table 4.19)
Table 4.19
Mahalanobis D? D i i n a n t Distance Score for the Product Feature Variable
The distance values, although very small for Product Feature attributes, are uniform
among all the attributes of Product Feature between the samples of urban arid semi-urban areas.
To sum up, there is a considerable difference, distinctly noticeable, between the groups compared
with respect to the Prodw Feature variables 1, 4, 5, 7, 9, 11 and 12. Thus, a moderate
discrimination exists between the urban and semi urban consumers as regards Appearance,
Technical Specification, Pnurimity of Supply, Problem Freeness, Brake and Control Efficiency,
Seating Comfort and Riding Comfort. In a nutshell, it may be concluded that the urban and semi
Attribute
Product Appearance
Technology
Emergency Needs
Technical Specifications
Proximity of Outlet
Range of Models
Problem Freeness
Fuel Efficiency
Pick Up / Brake Control
Environment Safety
Seating Comfort
Citylh'ighway Riding
Mahalanobis D2
0.00053
0.03818
0.09037
0.09452
0.09273
0.083 17
0.09297
0.05572
0 04438
0 083 17
0.09 105
0.09588
Variable Selection
Ranking Order
9
1
5
12
8
3
7
2
10
4
6
1 1
urban groups discriminate very much in Efficiency and Comfort Factor variables of the product
rather than the other factors identified earlier.
In general there are several misclassifications of consumers In the~r purchase preferences
among the urban and semi urban groups. As can be seen from the class~ficatlon table 4.20 a
majority of the consumers cluster around the means, which is as should be 111 a nornlal
distribution.
Table 4.20
Discriminant score Classification Matrix on two groups for Product Feature
Percentage of "Grouped" Cases Correctly Class~fied 5(1 5 percell1
Actual Group
Urban
Semi Urban
The classification table proves that the two predicator var~ahles of Ilrhan and Sern~ Urban
versus the twelve Product Feature variables. The correct classification ratlo works out to be 56.5
percent. In all, only 147 cases out of 254 cases were grouped correctly in the urban category and
135 cases out of 245 in the semi urban group are classified correctly In all 389 cases out of total
sample of 499 were classified correctly. This shows that there 1s a moderate variation of
responses among the two groups for the balance of 43.5 percent of cases. Though the distance
between the centroid mean and mean of the respective mean is less for the total group with the
individual group of urban and semi urban. Little discrimination is effected by the consumer
responses on the Product Features through the b9 but misclassification provides that majority of
cases differ in their preferences.
254
245
Predicted Group Membership
Urban
147 (57.9%)
110 (44.9%)
Semi Urban
107 (42.1 1 )
135 (55.IY)
Discriminant Analysis for Customer ! kmb Variable
Corporate decision makers seem to place as much, if not more. confidence in
demographics than either academic marketers or research practitioners. To arrive at the proper
information in this direction the discriminant analysis for the Customer Support Service attributes
are carried out. The results derived from the analysis of consumers of two groups of the sa~nple
urban and seml urban - for the customer support service var~ahles give a li~gh range of
differences among the two groups for almost all the attrihutes. This is contrary 1 0 the perception
of that differing perception of Product Features and Co~npany lrnage variables, However, In all
cases a high degree of difference is noticed between the groups ( urban and semr urbaa)
discriminated against for the Mahalanobis distance funct~on
Table 4. 21
Mahalanobii W Discriminant Distance Score for the
Customer Support Service Variable
Variahle Selection
Ranking Order
9
3
7
8
4
5
6
2
1
Attributes
Dependable Service
Employee Behaviour
Service Centre Appearance
Record Maintenance
Problem Appraisal
Complaint Register
Working Hours
Delivery Schedule
Maintenance Awareness
Mahalanohis D2
0.077 10
0.15728
0.13172
0.10153
0.15717
0.15683
0.15256
0.15739
0.15885
Table 4.22
Discriminant Score CLassiTkation Matrix on Two Groups for
Customer Support Service
Percentage of "Grouped" Cases Correctly Classified : 58 1 pt'rct'lll
Actual Group
[I rban
Serni Urban
The classification table clearly shows that the correctly classified cases appear on the
diagnol of the table since the predicted and actual groups are the same. [:or t'xa~nple, of the 254
cases of urban category, 152 cases are predicted correctly to the l~letrlhcrs of Group 1 (59.8
percent). While 102 cases are assigned to semi urban group . Similarly. 138 cases out of 245 of
semi urban category predicted correctly whereas the remalnlng I07 cases prrd~cted incorrectly
In total, 290 cases out of 499 sample are classified correclly wh~cli works out at 58.1 percent,
but the remaining 41.9 percent of cases there is a overlapping of the obscrvarlons In thc~r
preferences.
Discriminant Analysis for Delivery Terms Variable
Buyer behaviour for durable goods and particularly post purchase sat~sfaction in purchase
of durable goods can be explained and predicted on the basis of life style and changes in it. The
interaction between lifestyles and life cycle is of special importance in determining the buyer
behaviour. The search behaviour and availability of alternative can serve as useful bases for
marketing segmentation. In this di i t ion the data relate to the consumer preference on Delivery
Terms are analysed by using the discriminant analysis. The results confirm that the consumer
Total
254
245
Predicted Group Membership
Urban
152 (59.8%)
107 (43.7%)
Semi llrhan
102 (40.2%)
138 (56.3%)
responses on Delivery Terms among the urban and semi urban samples provlde the high degree
of differences in almost all attributesexcept the Dependable Service. Thls confirms that the urban
and semi urban groups differ in their perception on Delivery Terms. This difftrence exist due
to longer delivery time taken by the semi urban retail outlets. However. in all cases only
moderate differences are noticed through Mahalnobis D
Table 4.23
Mahalanobis D2 Discriminant Distance Score for the Delivery I'ern~s Variable
The D2distance between the mean of the centrod w ~ t h tire groups respecrive mean values
are smaller when compared to the other preference variables Higher difference IS noticed for the
Agency Switch Over Option when compared to all other attributes of Dellvery Terms. Apparently
such higher difference exists due to the fact that there is less number of agencies in semi urban
areas, whereas the number of outlets are more in the urban areas. This provides a better scope
for longer search and selection process for their vehicle. The second discriminant function goes
to the Free Availability of the vehicle. If the vehicle is not available in one agency in the urban
area, there is a chance for the consumers to identify another proper outlet where the vehicle is
readily available. But in semi urban areas this particular situation is not feasible due to the less
number of dealer outlets. This particular phenomenon contributes high differences in their
preferences. The third, milieu difference. is noticed for the Change of Colour Option. The fourth
Attributes
Prompt Delivery
Colour Choice
Choice of Outlet
User Guidance
Easy Availability
Mahalanobis P
0.03386
0.02563
0.14082
0 02436
0.08709
Variable Selection
Ranking Order
5
3
1
4
7
difference is noticed for &ision of User Guidance at the time of delivery. The last and least
difference is noticed for the Prompt Delivery as per the delivery schedule In short, the difference
between the urban and semi urban customer the preference discrimination is due to non
availability of sufficient number of dealer outlets.
T o see how much the two groups overlap in their attitudes can be clearly understood with
classification table.4.24
Table 4.24
Discriminant score Classification Matrix on two groups for 1)elivery Terms
Percentage of "Grouped" Cases Correctly Class~lied SY I prcrllr
Actual Group
Urban
Semi Urban
+
When the actual group membership is known, Illis can be compared to the pred~cted
group using the discriminant function thereby preparing the classification table. Judg~ng from
the classification table group differences in their preferences can be clearly understcxxi. As
regards the Delivery Terms attributes 147 urban consumers out of 254 are classified in their
respective group. Whereas, 148 semi urban consumers out of 245 are grouped correctly. In the
remaining 42.1 percent of cases of urban and 39.6 percent cases of semi urban consumers there
is a overlapping of the observations. It indicates that the discrimination between the two groups
exists when compared to overall sample population.
Total
254
245
Predicted Group Membership
Urban
147 (57.9%)
97 (39.6%)
Semi llrban
107 (42.1 %()
148 (h0.4'%)
Discriminant Analysis for qroduct Rice Variable
The results of the discriminant analysis of consumer responses on Product Price Terms
among the urban and semi urban samples gives high degree of differences In almost all attributes
except the Loan Facility. This confirms that the urban and semi urban purchase preference
perceptions on Product Price are differing in nature.
Table 4.25
Mahalanobis DL Discrimiint Distance Score for the Product Price Variable
Among all the six price preferences the acceptance of Higher prictng In case of financing
acilities the Mahalanobis @ distance is higher. Scope for financial assistance frorn the authorised
inancial institutions are more for the Urban consumers whereas the sttuatton is contrary for the
emi urban consumers. Thus, the semi urban consumers are ready to pay some extra cost for easy
.nancial assistance. This provides higher discrimination for this attribute. The second
iscriminant function goes to the Competitive Pricing. The third, milieu difference is noticed for
le R & D Cost . The fourth difference is noticed in Resale Value of the vehicle. The fifth
fference is noticed in the Notification of Change in Price. The last and least difference is
Attributes
Competitive Pricing
Notification of Price alteration
Pollution Awareness
Loan Facility
Cost of R & D.
Resale Value
Mahalanobis D
0.03499
0.00220
0.00566
0. 10332
O.00105
0.0()349
Variable Selection
Ranking Order
2
5
h
1
3
4
noticed for the Prompt Delivery as per the delivery schedule. In all the four attributes the D'
distance in very much low, so there is not much difference in their preferences.
In order to arrive at a proper conclusion, the classification matns is framed to Identify
the misclassifications. The table 4.26 lists the consumers on actual group and also the~r predicted
groups through the discriminant scores.
Table 4.26
Discriminant Score Classification Matrix on Two Groups for Product Price
Percentage of "Grouped" Cases Correctly Class~tied 63.3 percent
Actual Group
Urban
Semi Urban
The above table presents the results of the cross valldat~on class~ticat~on. I t 15 clear that
the there is a high degree of correct classification exists in the cross val~dat~on of samples on the
basis of the discriminant function. It is possible to classify 168 out of 254 of urban consumer
correctly. But only 148 out of 245 can be classified correctly in sernl urban category Judging
from the classification table it is obivious that the two predicator variables versus the S I X Product
Price attributes 63.3 percent are grouped correctly but the remaining 36.7 percent of cases there
is a overlapping of the observations. It indicates that the discrimination between the two groups
exists but with less diffkrence.
The following conclusions are drawn from the above discriminant analysis between the
urban and semi urban categories.
Total
254
245
Predicted Group Membership
Urban
168 (66.1%)
97 (39.6%)
Semi Urban
86 (33.9%)
1 48 (60.4 'X )
Mahalanobis DZ l l id&unt Distance Scorn based on the Factor Scores
Factor
Brand & Quality Image Factor
Reference Group & Media Image
Facility Image Factor
Product Performance Factor
Product Comfort Factor
Additional Necessity Factor
Product Availability Factor
Quality of Service Factor
Promptness in Service Factor
Prompt Deliver Fador
Chage Over Option Factor
Product Price Value Factor
Pricing Terms Factor
-
Mahdanobis DZ
0.45002
0.20532
0.291 15
0.24959
0.25000
0.00795
0.31 993
0.57.1 00
0.331 0 1
0.15002
0.12105
0.00393
0.27471
, Variable Selection Ranking Order
12
5
9
6
7
2
10
'1 3
11
4
3
-1
8
The two groups discriminate much in Brand Image, Service Image and to a considerable
:nt in the R & D Image. The difference is very low with regard to the Reference Group
ige.
The discrimination is very low in the Product Feature variables arnong the urhan and seml
)an categories. The two groups discrimate to a sizable extenl on Product Efficlericy and
3duct Comfort Factors.
Higher discrimination is noticed in the Custorner Support Service var~ahle. The
scrimination is more for the Customer Support Service Facilities and Stwlce 'Tlrnlngs factors
nong the urban and semi urban categories.
The Delivery Terms variable does not contribute any s~gri~ticant cl~scr~m~nat~on among
le two groups of consumers. Even for the Product Price va r~ah l r . a low amount of
iscrimination has been observed through the Mahalanohis distance