Week 3: Customer Segmentation and Targeting I (Factor Analysis) Flashcards
Sources of Customer Heterogeneity
Individual differences
Life experiences
Functional needs
Self identity/image
Marketing activities
Advantages of Market Segmentation
- Identification of customer groups
- Target group specific product design
- Target group specific communication
- Price differentiation
- Identify opportunities and threats
- Differentiate from competitors
Data Sources for Segmentation
Primary data
Exisiting data
Third party data
The Segmentation Process: Data Analytics
Data preparation - Factor analysis
Cluster analysis
Review & Refinement
Factor Analysis
Factor Analysis is a class of analytical procedures primarily used for data reduction and summarization
Applications of Factor Analysis in Marketing
Pricing
Advertising
Product Development
Market Segmentation
Types of Factor Analyses
Exploratory Factor Analysis
Principal Component Analysis
Confirmatory Factor Analysis
Common Factor Analysis
Eigenvalue
Total variance of all variables accounted for by one factor
Factor loadings
Correlations between the variables and factors (ranges from -1 to +1)
Factor scores
Relation between observations and factors
DO NOT SEE FACTOR SCORE IN THE OUTPUT
Communality
Proportion of one variable’s variance explained by all factors extracted
Factor analysis process
- Check assumptions and check if it makes sense to conduct an EFA
- Determine the number of factors
- Interpret the factor solution
- Evaluate the goodness-of-fit
of the factor solution
Assumptions
- Measurement level: Interval or ratio scales
- Standardised data (mean = 0 and standard deviation = 1), especially if variables were measured on different measurement
units or ranges - Sample size: rule of thumb >100 observations
Does it make
sense to conduct
an EFA?
- There should be high correlations among sets of variables (>0.5: moderately high; >0.6: high; >0.7: very high)
- Sample adequacy: Kaiser-Meyer-Olkin (KMO) evaluates all
correlations among all variables - Bartlett’s test of sphericity:
Significance indicates sufficient
correlations.
A multivariate statistical technique for studying interrelationships among variables, usually for discovering underlying constructs or data reduction is known as:
a)Multiple regression
b)Factor analysis
c)Discriminant analysis
d)Canonical correlation analysis
b)Factor analysis
To determine which variables relate to which factors, a researcher would use:
a)Factor loadings
b)Communalities
c)Eigen values
d)Beta coefficients
a)Factor loadings
If a researchers wants to determine the amount of variance in the original variables that is associated with a factor, they would use:
a) factor loadings
b) communalities
c) eigen values
d) beta coefficients
e) none of the above
c) eigen values
Which of the following can be used to determine how many factors to extract from a factor analysis:
a) Eigen values and percentage of variance explained by each factor
b) Scree plots
c) Factor loadings
d) All of the above
e) None of the above
d) All of the above
In exploratory factor analysis, how much variance would a good model be likely to explain?
a) 0 to 25%
b) 25% to 50%
c) 50% to 75%
d) 75% to 100%
c) 50% to 75%
When a factor loading matrix is rotated, what will be the likely outcome:
a) The pattern of factor loadings changes and the total variance explained by the factors remains the same.
b) The pattern of factor loadings stays the same and the total variance explained by the factors remains the same.
c) The pattern of loadings changes and the total variance explained by the factors changes too.
d) The pattern of loadings stays the same and the total variance explained by the factors changes
a) The pattern of factor loadings changes and the total variance explained by the factors remains the same.
In SPSS, orthogonal rotation in factor analysis is called:
a) Oblimin
b) Oblimax
c) Oblique
d) Varimax
e) None of the above
d) Varimax
The total of all eigen values will equal:
A) 1
B) 50
C) 100
D) The number of variables in the analysis
E) Impossible to tell without further information
D) The number of variables in the analysis
- Which of the following is not the part of the exploratory factor analysis process?
a) Extracting factors
b) Determining the number of factors before the analysis
c) Rotating the factors
d) Refining and interpreting the factors
b) Determining the number of factors before the analysis
Which of the following criteria cannot be used to determine the number of factors in an EFA?
a) Asking a group of researchers before the analysis
b) Eigenvalue rule
c) Scree test
d) Parallel analysis
a) Asking a group of researchers before the analysis
Which of the following is not an oblique rotation technique?
a) Promax
b) Oblimax
c) Varimax
d) Quartimin
c) Varimax
(T/F) Factor analysis works best if each variable has a high correlation with every other variable.
False
Each new variable created in a factor analysis is called a _______.
a) factor
b) indicator
c) vector
a) factor
In which kind of factor analysis does the researcher have a priori knowledge of the number and nature of the factors?
a) Exploratory factor analysis
b) Confirmatory factor analysis
b) Confirmatory Factor Analysis
When checking to see if a study’s data are “suitable” for a factor analysis, most researchers check their sample size and then examine 3 things: multicollinearity, the determinant, and the _______.
a) sample means on each variable
b) KMO measure of sampling adequacy
c) factor structure
d) standardized interquartile range
b) KMO measure of sampling adequacy
The number of eigenvalues is equal to the _______ whereas the sum of the eigenvalues quantities is equal to the _______.
a) sample size; number of variables
b) sample size; number of factors
c) number of variables; number of factors
d) number of factors; number of variables
d) number of factors; number of variables
Communalities primarily help indicate the usefulness of _______ (variables/factors).
variables
Variables
A factor loading is:
a) a correlation coefficient between a variable and a factor (cluster of variables).
b) the correlation between a binomial variable and a variable which has a continuous distribution of scores.
c) empirically based hypothetical variable consisting of items which are strongly associated with each other.
d) the correlation of a variable with a whole score.
a correlation coefficient between a variable and a factor (cluster of variables).
A factor loading of 0.80 means, generally speaking, that:
a) there is no relationship between that variable and the factor.
b) the item correlates well with the factor, though not perfectly.
c) the variable is moderately related with the factor.
d) the item is poorly related to the factor
b) the item correlates well with the factor, though not perfectly
Rotation usually involves _____ high correlations and _____ low ones.
maximising; minimising
The amount of variance ‘explained’ by a factor is shown by:
a) the factor loadings.
b) the eigenvalue.
c) the name of the factor.
d) the correlation matrix.
b) the eigenvalue.
The scree plot is…..
the number of factors plotted agains varaince accounted for.