FACTOR ANALYSIS

•PREPARED BY DR. SNEHA AMRE •1 •28 NOVEMBER 2022

•2 I NTRODUCTION

 Suppose we ask a respondent who is likely to buy

a 4-wheeler to rate the importance that he
would give to various aspects of a 4-wheeler like:
 Mileage
 Price
 smooth ride
 cost of spare parts
 servicing locations
 Cooling effect
 Leg space and so on
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•3 I NTRODUCTION

 Suppose he gives higher rating to following

attributes:
 Leg space
 Smooth ride
 Cooling effect
 Interiors
What is he looking for?
He is looking for comfort
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•4 I NTRODUCTION

Similarly, if the respondent gives high

scores on following attributes:
 Price of car
 Mileage
 Price of spare parts
 Interest on loan
What does he want?
He wants economy
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•5 I NTRODUCTION

 Thus ‘comfort’, ‘economy’ are not single

measurable entities but complex constructs
that are derived from many variables
 Such complex constructs are called Factors.
Identifying such factors greatly simplifies
understanding of complex phenomenon
 Factor analysis is a tool that is used to
identify factors from many inter-related
variables. Therefore, it is also called as data
reduction technique

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

FACTOR ANALYSIS MODEL

•PREPARED BY DR. SNEHA AMRE •6 •28 NOVEMBER 2022

•7

 In multiple regression and correlation one

variable is considered as dependent & another as
independent but in factor analysis
interdependence technique is used in that entire
set of interdependent relationship is examined.

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•8 FACTOR ANALYSIS MODEL

Leg space (X1) = w1 (comfort) + w2 (economy) + U1

 w1 is the index of relation between leg space

& comfort and w2 between leg space &
economy. Since leg space is more related to
comfort than economy, w1 will have higher
value than w2.
 U1 is that part of leg space variable not
expressed by factors ‘comfort’ & ‘economy’.
It is uniqueness of the variable

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•9 FACTOR ANALYSIS MODEL

In general, original variable can be expressed as

a linear combination of factors as below:
Xi = wi1 (F1) + wi2 (F2)+………+ wip (Fp)+Ui
wi1, wi2, … are called factor loadings (correlation
between the variables & factors) & F1, F2,…
are called common factors or simply factors.
Ui is the part of Xi not explained by factors.

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•10 S TEPS IN FACTOR ANALYSIS

 Step 1: Prepare statements:

 Respondents are given statements
related to the subject & response is
obtained on Graphic rating scale. Often
‘Likert scale’ is used.
 The responses are converted into scores
 Factor analysis is run on scores

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•11 S TEPS IN FACTOR ANALYSIS

 Step 2: To check goodness of fit

 Following are three outputs that

indicate appropriateness of model:
 Correlation matrix
 Bartlett’s test of sphericity
 Kaiser-Meyre-Olkin(KMO) statistic

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•12 S TEPS IN FACTOR ANALYSIS

 Correlation matrix:
 This matrix will show the simple
correlation between all possible pairs of
variables included in the analysis.
 Non-diagonal values of correlation matrix
close to zero show variables in row &
column are not related. If such values are
more, factor analysis will not be
appropriate for the data.
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
 •13

Correlation Matrix

afford freedom eco man power jeolous feel_good comfort safe legal
Correlation afford
1.000 -.183 .548 .074 .171 .167 .196 .309 .283 .475
freedom
-.183 1.000 -.548 -.089 -.330 .061 -.207 -.322 -.259 .024
eco
.548 -.548 1.000 -.144 .000 -.183 -.049 .439 .406 .320
man
.074 -.089 -.144 1.000 .898 .944 .915 -.277 -.027 .087
power
.171 -.330 .000 .898 1.000 .852 .955 -.143 .088 .123
jeolous
.167 .061 -.183 .944 .852 1.000 .886 -.328 -.048 .071
feel_good
.196 -.207 -.049 .915 .955 .886 1.000 -.222 .029 .167
comfort
.309 -.322 .439 -.277 -.143 -.328 -.222 1.000 .821 .060
safe
.283 -.259 .406 -.027 .088 -.048 .029 .821 1.000 -.100
legal
.475 .024 .320 .087 .123 .071 .167 .060 -.100 1.000
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•14 S TEPS IN FACTOR ANALYSIS

 Bartlett’s test of sphericity can be used to test the null

hypothesis that the variables are uncorrelated in the
population. If null hypothesis is accepted, factor analysis is
not suitable to data

 Kaiser-Meyer-Olkin (KMO) measures sampling adequacy.

Its value also tells us whether factor analysis is appropriate
for data.

 If KMO >= 0.50 factor analysis is appropriate for the data.

 Small value of the KMO statistic indicate that the

correlation between pairs of variables can not be explained
by other variables & that factor analysis may not be
appropriate.
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•15

KMO and Bartlett's Test

Kaiser-Meyer-Olkin Measure of Sampling Adequacy.

.618

Bartlett's Test of Sphericity Approx. Chi-Square

164.098
df
45
Sig.
.000

• approx chi square statistic is 164.098 with 45 degrees of freedom which is

significant at 0.05 level
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•16 S TEPS IN FACTOR ANALYSIS

Let us consider following factor loading matrix

(matrix of w’s):

Variables Factor 1 Factor 2

X1 0.8045 -0.2578
X2 0.7245 0.2354
X3 -0.2585 0.9541

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•17 COMMUNALITY

 Communalities indicate the amount of variance in

each variable that is accounted for.

 For principal components analysis, this is always equal

to 1.0 (for correlation analyses) or the variance of the
variable (for covariance analyses).

 Extraction communalities are estimates of the

variance in each variable accounted for by the factors
(or components) in the factor solution

 Small values indicate variables that do not fit well

with the factor solution, and should possibly be
dropped from the analysis.
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•18

V1

V2
V3
V4

V5

V6

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•19 S TEPS IN FACTOR ANALYSIS

 Step 3: To decide no. of factors:

 By default, number of factors = number of
variables

 Factors with eigen value more than 1 are

considered

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•20 S TEPS IN FACTOR ANALYSIS

 An Eigenvalue represents the total variance

explained by each factor.

 Hence only factors with variance greater than 1

are included.

 It is computed as sum of squares of factor

loadings on the factor.

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•21
FACTOR L OADING IN
C OMPONENT M ATRIX

Variables Factor 1 Factor 2

X1 0.8045 -0.2578
X2 0.7245 0.2354
X3 -0.2585 0.9541
Here the model will be:
X1 = 0.8045 F1 – 0.2578 F2+U1
X2 = 0.7245 F1 + 0.2354 F2+U2
X3 = -0.2585 F1 + 0.9541 F2+U3
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•22 S TEPS IN FACTOR ANALYSIS

Variable Factor 1 Factor 2

X1 0.8045 -0.2578
X2 0.7245 0.2354
X3 -0.2585 0.9541
Eigenvalue (0.8045)2+(0.7245)2+(-0.2585)2 (-0.2578)2 +
= 1.2389 (0.2354)2 + 0.9541)2
(Sum of = 1.0322
squares)

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•23 S TEPS IN FACTOR ANALYSIS

 Once number of factors are decided, one can find

information from each variable captured by
selected factors

 This is called communality of each variable

 It is sum of squares of factor loadings of the

variable

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•24

Variables Factor 1 Factor 2 Communality

X1 0.8045 -0.2578 0.7137
X2 0.7245 0.2354 0.5803

X3 -0.2585 0.9541 0.9771

 Communality of X1 = (0.8045)2+(-0.2578)2=0.7137.
 It means out of total variance (information) of 1 unit of
X1, 0.7137 is captured by common factors F1 and F2.
Remaining 0.2863 is unique to X1, which could not be
captured by common factors. It is U1in the model
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•25 S TEPS IN FACTOR ANALYSIS

Variable Factor 1 Factor 2 Communality

X1 0.8045 -0.2578 0.7137

X2 0.7245 0.2354 0.5803
X3 -0.2585 0.9541 0.9771
Eigenvalue 1.2389 1.0322

From above , X1 & X2 can be combined with Factor 1 and X3 with

Factor 2. For such allocation we would prefer factor loadings of
a variable high on only one factor. This can be done with
rotation of axis
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•26
Component Matrix(a)
Extraction Method: Principal Component Analysis.
a 3 components extracted.

Component
1 2 3
feel_good .971
man .966
jeolous .952
power .951
eco .820
comfort .775
safe .735
afford .670
freedom -.608
legal .814

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•27 S TEPS IN FACTOR ANALYSIS

 Steps 4: Rotate axis:

 To understand rotation of axis, let us see the
graph in next slide

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•28 S TEPS IN FACTOR ANALYSIS

0.6

0.5

0.4

0.3

0.2

0.1

0
-0.6 -0.4 -0.2 0 0.2 0.4 0.6

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•29 S TEPS IN FACTOR ANALYSIS

0. 6

0. 5

0. 4

0. 3

0. 2

0. 1

-0. 6 -0. 4 -0. 2 0 0. 2 0. 4 0. 6

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•30 S TEPS IN FACTOR ANALYSIS

 By rotating axis, points have come closer to new X

& Y axis. Co-ordinates of points w.r.t. rotated axis
will be such that either X co-ordinate will be high
or Y co-ordinate will be high but not both.

 By rotating axis, we may get factor loadings high

only on one factor and low on other factors. Such
factor loadings are easy to interpret

 ‘Varimax’ rotation of axis is preferred method of

rotation
•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022
•31 S TEPS ON FACTOR ANALYSIS

 Step 5: Name factors :

 By combining variables with high factor loadings,
irrespective of sign, will give us factors

 Based on variables that get combined, we name

the factor. Naming the factor is everybody’s
judgment

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

 FACTOR ANALYSIS APPLICATIONS
•32
IN MARKETING RESEARCH

 It can be used in market segmentation for

identifying the underlying variables on which to
group the customers.

 In product research it can be employed to

determine the brand attributes that influence
consumer choice. Toothpaste brand might be
evaluated in terms of protection against cavities,
whiteness of teeth, taste, fresh breath & price.

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•33

 In pricing studies, it can be used to identify the

characteristics of price sensitive consumers. Eg.
These consumer might be methodical, economy
minded, & home centered.

 In advertising studies it is used to understand the

media consumption habits of the target market.
Eg. The users of frozen food may be heavy
viewers of cable TV, see lot of movies etc.

PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022

•34

END OF
FACTOR ANALYSIS

•PREPARED BY DR. SNEHA AMRE •28 NOVEMBER 2022