Factor Analysis

Data Analysis

Prof. Sudipta Mandal / Department of Marketing 2

 Data Analysis

Prof. Sudipta Mandal / Department of Marketing 3

 Concept
Factor Analysis
• Basic Concept in a Single Line à Factor analysis
allows us to look at groups of variables that tend
to be correlated to each other and identify
underlying dimensions that explain these
correlations

Prof. Sudipta Mandal / Department of Marketing 4

 Concept
Factor Analysis

• Primary use: Data Reduction and Summarization

• E.g. à Suppose you are interested in trying to

understand how customers evaluate banks?

Prof. Sudipta Mandal / Department of Marketing 5

 Concept
These variables could be –
ü Interest rates on loans ü Speed of service

ü Reputation in the community ü Banking hours

ü Low rates for checking ü Recommendations from friends and

relatives
ü Friendly and personalized service
ü Physical structure and ambience
ü Easy-to-read monthly statements
ü Community involvement
ü Obtainability of loans
ü Competence of employee
ü Branch location
ü Availability of auxiliary banking
ü ATM locations services
Prof. Sudipta Mandal / Department of Marketing 6
 Concept
Factor Analysis

• There could be a large number of variables (many

of which may be correlated) and these variables
need to be reduced to a manageable level

Prof. Sudipta Mandal / Department of Marketing 7

 Concept
Factor Analysis

• You would use Factor Analysis when such a large

number of explicit variables needs to be reduced to
a smaller set of latent, i.e., implicit underlying
“factors” that contain most of the variance of the
original variables

Prof. Sudipta Mandal / Department of Marketing 8

 Concept
Factor Analysis
• Technique which, by analyzing correlations
between variables, reduces their number into fewer
factors which explain much of the original data
more economically

Prof. Sudipta Mandal / Department of Marketing 9

 Concept
Factor Analysis
• Looking at a creative use of Factor Analysis
• Suppose you have 2 groups –
• Users vs. non-users of a brand, OR
• Active vs. Inactive customers at a retail store

Prof. Sudipta Mandal / Department of Marketing 10

 Concept
Factor Analysis
• Do a factor analysis separately for the 2 groups
• Check what differences exist in the factors
extracted
• This will be an indirect way of figuring out
differences in buying criteria for the same product
category among the 2 groups

Prof. Sudipta Mandal / Department of Marketing 11

 Concept
• Suppose if your objective is to classify a person by
predicting whether she will be a buyer or non-buyer
then how do you identify such a person?
• Use Discriminant analysis NOT Factor Analysis

Prof. Sudipta Mandal / Department of Marketing 12

 Concept
• Example of Factor Analysis…

• You want to determine the psychographic profiles of

consumers on the basis of a set of lifestyle statements

• You have to factor analyze these lifestyle statements to

identify the hidden or latent underlying psychographic
factors

Prof. Sudipta Mandal / Department of Marketing 13

 Example – 1
The set of lifestyle statements in your survey questionnaire (1 = strongly
disagree; 7 = strongly agree) are:

• I like watching football

• I like watching tennis

• I like watching movies

• I like watching plays

• I like spending my evenings at home

• Home is the best place for me

• I like to party with my friends

• I like hanging out with friends outside home

Prof. Sudipta Mandal / Department of Marketing 14
 Example – 1
Factor 2
Tennis
Football

Evenings at home
Partying

Factor 1

Hanging Out
Home best place

Plays
Movies
Prof. Sudipta Mandal / Department of Marketing 15
 Example – 1
• Each factor is a linear combination of some of the original variables

• Factor 1 ⇒ Home-centered vs. Socialite

• Or, Factor 1 ⇒ Introvert vs. Extrovert

• Factor 2 ⇒ Sports-oriented vs. Movies/Plays-oriented

• These underlying “factors” are not explicit variables that you started off with

• You identified them basis the factor analysis

• That is why these are called latent, i.e., hidden or implicit underlying dimensions

• Latent variables are not directly observed but are rather inferred (through a mathematical
model) from other variables that are observed, i.e., directly measured

• More importantly, the nomenclature of the identified factors is in your hands, i.e., it is your
subjective decision

Prof. Sudipta Mandal / Department of Marketing 16

 Concept
• Where else would you use Factor Analysis?
• Suppose you want to identify a new, smaller set of
uncorrelated variables to replace the original set of
correlated variables in subsequent multivariate analysis
(regression or discriminant analysis)
• For example, the psychographic factors identified may
be used as independent variables in explaining the
differences between loyal and non-loyal consumers.
Thus, instead of the 7 correlated psychographic
variables in the previous example, we can use the two
uncorrelated factors.
Prof. Sudipta Mandal / Department of Marketing 17
 Concept
• Look at the orientation of the 2 factors in the
previous example
• They are orthogonal
• Orthogonality ⇒ There is no correlation or (minimum
correlation)
• Can help circumvent the problem of multicollinearity
• Multicollinearity occurs when there are high correlations
between two or more predictor variables ⇒ one
predictor variable can be used to predict the other
• E.g. à Using ‘Education’ and ‘Income’ as IVs
Prof. Sudipta Mandal / Department of Marketing 18
 Concept
• So, are these uses of Factor Analysis exploratory or
conclusive in nature?
• Exploratory in nature
• Factor analysis is also called Exploratory Factor
Analysis (EFA)

Prof. Sudipta Mandal / Department of Marketing 19

 Applications
• Applications of EFA in MR –
• Market segmentation à for identifying the underlying
variables on which to group the customers
• New car buyers might be grouped based on the relative
emphasis they place on economy, convenience,
performance, comfort, and luxury.
• This might result in five segments: economy seekers,
convenience seekers, performance seekers, comfort
seekers, and luxury seekers

Prof. Sudipta Mandal / Department of Marketing 20

 Applications
• Applications of EFA in MR –
• Product research à to determine the brand attributes
that influence consumer choice
• Toothpaste brands might be evaluated in terms of
protection against cavities, whiteness of teeth, taste,
fresh breath, and price

Prof. Sudipta Mandal / Department of Marketing 21

 Applications
• Applications of EFA in MR –
• Advertising studies à to understand media
consumption habits of the target market
• Users of frozen foods may be heavy viewers of cable
TV, see a lot of movies, and listen to country music

Prof. Sudipta Mandal / Department of Marketing 22

 Applications
• Applications of EFA in MR –
• Pricing studies à to identify characteristics of price-
sensitive consumers
• For example, these consumers might be methodical,
economy minded, and home centered

Prof. Sudipta Mandal / Department of Marketing 23

 EFA – Recommended Usage
• Unless some of the variables are correlated with
each other EFA is not recommended
• Why?
• Because it may result in forced extraction of non-
existing factors
• Therefore, there are some means to ensure that there
are indeed significant correlations among the variables
in the input data –
• Correlation Matrix

• Bartlett’s test of sphericity

Prof. Sudipta Mandal / Department of Marketing 24
 EFA Related Statistical Measures
Bartlett’s test of sphericity
• Test statistic or measure used to examine whether each
variable has a correlation with the other variables
• What is the corresponding null hypothesis for this test?

Prof. Sudipta Mandal / Department of Marketing 25

 EFA Related Statistical Measures
Correlation Matrix
• Lower triangle matrix showing the simple correlations, r,
between all possible pairs of variables included in the
analysis.

Prof. Sudipta Mandal / Department of Marketing 26

 EFA Related Statistical Measures
Kaiser-Meyer-Olkin (KMO) measure of sampling
adequacy
• This is an index used to examine the appropriateness
of factor analysis.
• High values (between 0.5 and 1.0) indicate factor
analysis is appropriate.
• Values below 0.5 imply that factor analysis may not be
appropriate.

Prof. Sudipta Mandal / Department of Marketing 27

 EFA Related Statistical Measures
Communality
• The proportion of variance in any one of the
original variables which is captured or explained
by the extracted/common factors is known as
Communality.
• In other words, communality is the amount of variance a
variable shares with all the other variables included in
the analysis.

Prof. Sudipta Mandal / Department of Marketing 28

 EFA Related Statistical Measures
Eigenvalue

• The eigenvalue represents the total variance explained by each factor.

• Approximately analogous to the “variance explained” concept of regression analysis

• Higher the eigenvalue of a factor, higher is the amount of variance explained by that factor

• Our attempt in EFA is to extract the least number of factors possible which will
maximize the explained variance

• Therefore, while interpreting the output, we are interested in Eigenvalue > 1

• Why?

• Because, it is assumed that each of the original variables has an Eigenvalue = 1 and given
that each factor is a linear combination of some of the original variables, therefore a factor
must explain at least as much of the variance if not more than a single original variable

Prof. Sudipta Mandal / Department of Marketing 29

 EFA Related Statistical Measures
Factor matrix
• A factor matrix contains the factor loadings of all the
variables on all the factors extracted.

Prof. Sudipta Mandal / Department of Marketing 30

 EFA Related Statistical Measures
Factor loadings
• Factor loadings are simple correlations between the
variables and the factors.

Factor Loading

Prof. Sudipta Mandal / Department of Marketing 31

 EFA Related Statistical Measures
Percentage of variance
• This is the percentage of the total variance attributed to
each factor.

Prof. Sudipta Mandal / Department of Marketing 32

 Stages in EFA
Generally, EFA is done in two stages:
Stage 1 –
• Extraction of Factors
• Identify how many factors will be extracted from the
data
• Method used à Principal Component Analysis (PCA)

Prof. Sudipta Mandal / Department of Marketing 33

 Stages in EFA
Suppose after Stage 1, you get a Factor Matrix that looks like:

If X ⇒ High loading, What does this imply?

Variables 2, 4, and 5 have high loadings on both Factors 1 and 2

Therefore you are not sure whether variables 2, 4, and 5 are part of Factor 1
or Factor 2 making interpretation of the underlying factors extremely difficult

Prof. Sudipta Mandal / Department of Marketing 34

 Stages in EFA
This warrants the 2nd stage of EFA

Stage 2 –
• Rotation of the principal components, i.e., rotation of the
solution obtained in Stage 1

• The objective is to find variables which have a high loading

on one factor, but low loading on the other factors

• Typically the rotation methods used are one of – Varimax

Rotation or Quartimax Rotation

• Most popularly used is the Varimax Rotation

Prof. Sudipta Mandal / Department of Marketing 35
 Stages in EFA
Pictorially, the rotation in Stage 2 achieves

Rotation

Prof. Sudipta Mandal / Department of Marketing 36

 Example
Imagine that a two wheeler manufacturer is interested in
determining which variables his potential customers think about
when they consider his product.
Assume that twenty two-wheeler owners were surveyed by this
manufacturer (or by a marketing research company on his
behalf).
They were asked to indicate on a seven point scale
(1=Completely Agree, 7=Completely Disagree), their agreement
or disagreement with a set of ten statements relating to their
perceptions and some attributes of the two-wheelers.
Essentially, these 10 statements represent the 10 variables of
interest
Prof. Sudipta Mandal / Department of Marketing 37
 Example
1. I use a two-wheeler because it is affordable.
2. It gives me a sense of freedom to own a two-wheeler.
3. Low maintenance cost makes a two-wheeler very economical in the
long-run.
4. A two-wheeler is essentially a man’s vehicle.
5. I feel very powerful when I am on my two-wheeler.
6. Some of my friends who don’t have their own vehicle are jealous of me.
7. I feel good whenever I see the ad for my two-wheeler on TV, in a
magazine or on a hoarding.
8. My vehicle gives me a comfortable ride.
9. I think two-wheelers are a safe way to travel.
10. Three people should be legally allowed to travel on a two-wheeler.
Prof. Sudipta Mandal / Department of Marketing 38
 Example
Why do I need to do an EFA in this scenario?
Because –
The objective of doing EFA is to find underlying "factors"
which would be fewer than 10 in number, but would be
linear combinations of some of the original 10
variables

Prof. Sudipta Mandal / Department of Marketing 39

 Example
In the data file, columns ‘St1’ to ‘St10’ contain
responses of twenty respondents to the 10 statements
Steps –
1st à Select ANALYZE from the SPSS menu bar
2nd à Click DIMENSION REDUCTION and then
FACTOR.
3rd à Move all the variables into the VARIABLES box

Prof. Sudipta Mandal / Department of Marketing 40

 Example
4th à Click on DESCRIPTIVES. In the pop-up window, in
the STATISTICS box check INITIAL SOLUTION. In the
CORRELATION MATRIX box check KMO AND
BARTLETT’S TEST OF SPHERICITY and also check
REPRODUCED. Click CONTINUE.
5th à Click on EXTRACTION. In the pop-up window, for
METHOD select PRINCIPAL COMPONENTS (default). In the
ANALYZE box, check CORRELATION MATRIX.
In the EXTRACT box, select BASED ON EIGENVALUE and
enter 1 for EIGENVALUES GREATER THAN box. In the
DISPLAY box check UNROTATED FACTOR SOLUTION. Click
CONTINUE.

Prof. Sudipta Mandal / Department of Marketing 41

 Example
6th à Click on ROTATION. In the METHOD box check
VARIMAX. In the DISPLAY box check ROTATED
SOLUTION. Click CONTINUE.
7th à Click on SCORES. In the pop-up window, check
DISPLAY FACTOR SCORE COEFFICIENT MATRIX. Click
CONTINUE.
8th à Click OK.

Prof. Sudipta Mandal / Department of Marketing 42

 Example – Interpreting the output
KMO = 0.618 > 0.5
Bartlett’s Test of Sphericity à 164.09, p < 0.05
Conclusion: It is appropriate to do an EFA

Prof. Sudipta Mandal / Department of Marketing 43

 Example – Interpreting the output
Then we look at the Factor Matrix (Unrotated) also called
the ‘Component Matrix’ ⇒ 3 factors extracted

Prof. Sudipta Mandal / Department of Marketing 44

 Example – Interpreting the output
Then focus on the ‘Total Variance Explained’ à This gives
the Eigenvalues and the cumulative percentage of
variance.

Prof. Sudipta Mandal / Department of Marketing 45

 Example – Interpreting the output
Cumulative percentage of variance ⇒ the 3 factors or
components extracted together account for 80.3% of the total
variance, i.e., information contained in the original 10 variables

Prof. Sudipta Mandal / Department of Marketing 46

 Example – Interpreting the output
This is a pretty good bargain, because we are able to economize on the
number of variables (from 10 we have reduced them to 3 underlying factors),
while we lost only about 20 percent of the information content (80 percent is
retained by the 3 factors extracted out of the 10 original variables).

Prof. Sudipta Mandal / Department of Marketing 47

 Example – Interpreting the output
Our next task is to try and interpret what these 3 extracted
factors represent by focusing on the rotated and unrotated
factor matrices.

Prof. Sudipta Mandal / Department of Marketing 48

 Example – Interpreting the output
Factor 1 is a combination of 4 original variables ⇒ “a man’s
vehicle” (St4), “feeling of power” (St5), “others are jealous of
me” (St6) and “feel good when I see my 2-wheeler ads” (St7).

Prof. Sudipta Mandal / Department of Marketing 49

 Example – Interpreting the output
Factor 2 is a combination of 2 or 3 original variables
(researcher’s choice) ⇒ “low maintenance” (St3), “comfort”
(St8), and “safety” (St9).

Prof. Sudipta Mandal / Department of Marketing 50

 Example – Interpreting the output
Factor 3 is a combination of 1 or 2 original variables
(researcher’s choice) ⇒ “affordability” (St1) and “cost saving”
(St10).

Prof. Sudipta Mandal / Department of Marketing 51

 Example – Interpreting the output
Factor 1 à “Machismo” or “Pride of ownership” (whatever
you want to call) because it is a combination of the 4
original variables ⇒ “a man’s vehicle” (St4), “feeling of
power” (St5), “others are jealous of me” (St6) and “feel
good when I see my 2-wheeler ads” (St7).
Factor 2 à “Utility” or “Functional Features” because it is a
combination of 2 or 3 original variables ⇒ “low
maintenance” (St3), “comfort” (St8), and “safety” (St9).
Factor 3 à “Economy” or “Low Cost” because it is a
combination of 1 or 2 original variables (researcher’s
choice) ⇒ “affordability” (St1) and “cost saving” (St10).
Prof. Sudipta Mandal / Department of Marketing 52
 Points to Pay Attention To
We must guard against the possibility that a variable
may load highly on more than one factors.
Strictly speaking, a variable should load close to 1.00
on one and only one factor, and load close to 0 on the
other factors.
If this is not the case, it indicates that either the sample
of respondents have more than one opinion about the
variable, or that the question/ variable may be unclear
in its phrasing.

Prof. Sudipta Mandal / Department of Marketing 53

 Points to Pay Attention To
Another issue important in practical use of factor
analysis is the answer to the question “what should
be considered a high loading and what is not a
high loading?”
Unfortunately, there is no clear-cut guideline, and
many a time, we must look at relative values in the
factor matrix.
Sometimes, 0.7 may be treated as a high value, while
sometimes 0.9 could be the cutoff for high values.

Prof. Sudipta Mandal / Department of Marketing 54

 EFA Related Statistical Measures
Communality
• The proportion of variance
in any one of the original
variables which is captured
by the extracted factors is
known as Communality.

• This means that 0.72243

or 72.24 percent of the
variance (information
content) of variable 1 is
being captured by our 3
extracted factors
together.

Prof. Sudipta Mandal / Department of Marketing 55

 Points to Pay Attention To
Variable 2 exhibits a low communality value of 0.45214.
This implies that only 45.214 percent of the variance in
variable 2 is captured by our extracted factors.

This may also partially explain why variable 2 is not

appearing in our final interpretation of the factors.

It is possible that variable 2 is an independent variable

which is not combining well with any other variable, and
therefore should be further investigated separately.
“Freedom” could be a different concept in the minds of our
target audience.
Prof. Sudipta Mandal / Department of Marketing 56
 Points to Pay Attention To

Prof. Sudipta Mandal / Department of Marketing 57

 EFA Related Statistical Measures
Scree plot
• A scree plot is a plot of the eigenvalues against the
number of factors in order of extraction.

Prof. Sudipta Mandal / Department of Marketing 58

 EFA Related Statistical Measures

Prof. Sudipta Mandal / Department of Marketing 59

 Points to Pay Attention To
It is recommended that we use the rotated factor
matrix (rather than unrotated factor matrix) for
interpreting factors, particularly when we use the
principal components method for extraction of factors
in Stage 1.

Prof. Sudipta Mandal / Department of Marketing 60

 Question
How do you relate all of what you have learnt to
Cronbach’s Alpha?

Prof. Sudipta Mandal / Department of Marketing 61