COSTUMER MARKETING ANALYTICS
Table of Contents
Lecture 1: Introduction .................................................................................................................. 2
Type of Marketing Research ................................................................................................................... 2
Lecture 3: Measurement and Scaling: Reliability, Validity, Dimensionality (Factor Analysis) ...... 3
Measurement Scales............................................................................................................................... 3
Popular Scaling Techniques .................................................................................................................... 4
Reliability and Validity ............................................................................................................................. 4
Factor Analysis ........................................................................................................................................ 5
Lecture 4: Creating perceptual Maps using Factor Analysis ......................................................... 8
Measuring Consumer perception ........................................................................................................... 8
Positioning Brands on the map............................................................................................................... 9
Lecture 5: Market Response Model and Multiple Regression Analysis ........................................ 9
Simple Regression ................................................................................................................................. 10
Multiple Regression .............................................................................................................................. 11
Data Transformation ............................................................................................................................. 12
Lecture 6: Moderation and Mediation Analysis .......................................................................... 12
Mediation .............................................................................................................................................. 12
Simple Mediation ..............................................................................................................................................13
Serial mediation + Parallel mediation...............................................................................................................15
Moderation ........................................................................................................................................... 17
Moderated Regression .....................................................................................................................................17
Simple Moderation Analysis with PROCESS TOOL ...........................................................................................19
Confidence Intervals .........................................................................................................................................20
Moderated Mediation .......................................................................................................................... 22
Lecture 7 & 8: Predicting Customer Response using Logistic Regression ................................... 22
RMF Model ............................................................................................................................................ 22
Logistic Regression ................................................................................................................................ 23
Interpretation of Outcomes ................................................................................................................. 23
Lecture 9 & 10: Understanding Individual Customer Preferences using Conjoint Analysis ........ 25
Performing a Conjoint Analysis............................................................................................................. 26
Market Share Analysis .......................................................................................................................... 28
Dummy Coding...................................................................................................................................... 28
Re-scaling the Regression coefficients ................................................................................................. 30
1
,Lecture 1: Introduction
Value of marketing research is decreasing uncertainty and improving the marketing
performance on micro (individual) and macro (market) level
Marketing Decision Problem: What actions should be taken?
Marketing Research Problem: What information is needed?
Type of Marketing Research
1. Qualitative Research = Feelings and opinions (Interviews & Focus Groups).
- Focuses on attitudes, behaviors, and motivations.
- Methods: Interviews, Focus Groups
- Advantages: Gaining deeper understanding of customers, leaves room for
creativity
2. Quantitative Research = Numbers and statistics (Surveys & Experiments).
- Emphasizes numerical data and statistical analysis.
- Methods: Surveys, Experiments
- Advantages: Profiling data usage and behavior, highlighting differences
between groups, measuring preferences
3. Exploratory Research = Discovering new ideas (Literature & Case Studies).
- Aims to gain insights and clarify concepts.
- Approaches: Literature surveys, focus groups.
4. Descriptive Research = Describing what exists (Data Analysis).
- Describes characteristics of groups and estimates proportions.
- Involves analyzing data to find patterns.
5. Causal Research = Understanding causes and effects (Experiments).
- Focuses on determining cause-and-effect relationships through experiments
- Examines associations between changes in variables
Data Sources
- Primary Data: Collected specifically for current research (e.g., surveys, experiments).
- Secondary Data: Previously collected for other purposes (e.g., market reports,
internal records).
- Syndicated Research: Large-scale data available for subscription from a research firm
(e.g., Nielsen, Kantar).
Types of variables
- Dependent Variable is what is tried to be explained (e.g. overall liking)
- Independent Variables need to have several variations (e.g. gender, age, income)
Item vs. Variable vs. Observation
- Item: A specific question or measurement used to collect data (e.g., "How satisfied are
you with our service?").
2
, - Variable: A characteristic or attribute that can vary among subjects, represented by
one or more items (e.g., "customer satisfaction" or "age").
- Observation: A single data point recorded for a specific subject or unit, representing a
value for a variable (e.g., John's satisfaction score of "4")
Lecture 3: Measurement and Scaling: Reliability, Validity,
Dimensionality (Factor Analysis)
Measurement = Assigning numbers to characteristics of objects or people based on
predefined rules (1= women, 0 = men)
Scaling = Creating a continuum where measured objects are placed based on their attributes.
® Example: Attitude toward a brand rated from 1 (unfavorable) to 3 (favorable).
® Meaning
Measurement Scales
The identification of the right measurement scale depends on the basic nature of the
attribute (the data)
1. Nonmetric Measurement Scales
- Nominal Scale: Categories without a specific
order (e.g., gender).
- Ordinal Scale: Categories with a specific
order but no precise difference (e.g., rank
preferences or race finishers). You can
determine the order (e.g., rank-order
correlations), find the mode or median, but
you can’t perform arithmetic operations like
addition or multiplication.
2. Metric Measurement Scales
- Interval Scale: Ordered categories with equidistant differences (e.g., temperature or
IQ). You can add and subtract values and calculate averages, but you cannot calculate
meaningful ratios (e.g., you can't say that 30°C is "twice as hot" as 15°C).
- Ratio Scale: Like interval, but has a true zero point (e.g., weight, height). All
mathematical operations are valid. You can calculate means, medians, modes, and you
can make meaningful comparisons between values (e.g., "twice as much" or "half as
much").
3
, Why is it important?
Critical in determining which data analysis techniques are the most applicable to the data.
- Chi-Square determines whether there is a significant correlation between two
categorical variables à nominal
- Pearson correlation coefficient examines the strength and direction of the linear
relationship between two continuous variables à Interval & Ratio
- Linear regression à DV is interval & ratio
- Logistic regression à DV is nominal
Popular Scaling Techniques
Comparative Scaling:
- Paired Comparison: Choose one option out of two (either Heineken or Grolsch)
- Rank Order: Rank several options (1. Grolsch, 2. Heineken, 3. Bavaria)
Disadvantage: you must have a preference
- Constant Sum: Allocate points among options (Grolsch 50, Heineken 25, Bavaria 25)
Disadvantage: the more items, the more complex
Non-Comparative Scaling:
Advantage: The more people know, the more complex they will answer
Important: Always uneven scores to enable neutral response and label with numbers
- Likert Scale: Indicate degree of agreement with a series of statements
- Semantic Differential Scale: Rate perceptions on a bipolar scale with endpoints (e.g.
pleasant vs. unpleasant)
- Continuous Rating Scale (Slider Scale): Rate objects by placing a mark at the
appropriate position on a continuous line
More fine grades possible than by choosing
Multiple Item Measurement
Importance: Multiple Items better capture complex characteristics through several
questions to improve reliability and validity.
Example: Measuring Impulsive Buying Tendency with multiple related statements.
- Latent variable (construct) e.g. Impulsive Buying Tendency
- Indicator (items/questions asked)
IBT Score: Mean Score of items = Question A+B+C+D+E / 5
Reliability and Validity
Reliability
Consistency in measurement results over time (when repeated)
Testing Reliability:
- Items need Variance and should correlate with each other
- Internal consistency of questions must be given (no double negation in items)
- Methods: Test/re-test, internal consistency (Cronbach’s alpha)
4
Table of Contents
Lecture 1: Introduction .................................................................................................................. 2
Type of Marketing Research ................................................................................................................... 2
Lecture 3: Measurement and Scaling: Reliability, Validity, Dimensionality (Factor Analysis) ...... 3
Measurement Scales............................................................................................................................... 3
Popular Scaling Techniques .................................................................................................................... 4
Reliability and Validity ............................................................................................................................. 4
Factor Analysis ........................................................................................................................................ 5
Lecture 4: Creating perceptual Maps using Factor Analysis ......................................................... 8
Measuring Consumer perception ........................................................................................................... 8
Positioning Brands on the map............................................................................................................... 9
Lecture 5: Market Response Model and Multiple Regression Analysis ........................................ 9
Simple Regression ................................................................................................................................. 10
Multiple Regression .............................................................................................................................. 11
Data Transformation ............................................................................................................................. 12
Lecture 6: Moderation and Mediation Analysis .......................................................................... 12
Mediation .............................................................................................................................................. 12
Simple Mediation ..............................................................................................................................................13
Serial mediation + Parallel mediation...............................................................................................................15
Moderation ........................................................................................................................................... 17
Moderated Regression .....................................................................................................................................17
Simple Moderation Analysis with PROCESS TOOL ...........................................................................................19
Confidence Intervals .........................................................................................................................................20
Moderated Mediation .......................................................................................................................... 22
Lecture 7 & 8: Predicting Customer Response using Logistic Regression ................................... 22
RMF Model ............................................................................................................................................ 22
Logistic Regression ................................................................................................................................ 23
Interpretation of Outcomes ................................................................................................................. 23
Lecture 9 & 10: Understanding Individual Customer Preferences using Conjoint Analysis ........ 25
Performing a Conjoint Analysis............................................................................................................. 26
Market Share Analysis .......................................................................................................................... 28
Dummy Coding...................................................................................................................................... 28
Re-scaling the Regression coefficients ................................................................................................. 30
1
,Lecture 1: Introduction
Value of marketing research is decreasing uncertainty and improving the marketing
performance on micro (individual) and macro (market) level
Marketing Decision Problem: What actions should be taken?
Marketing Research Problem: What information is needed?
Type of Marketing Research
1. Qualitative Research = Feelings and opinions (Interviews & Focus Groups).
- Focuses on attitudes, behaviors, and motivations.
- Methods: Interviews, Focus Groups
- Advantages: Gaining deeper understanding of customers, leaves room for
creativity
2. Quantitative Research = Numbers and statistics (Surveys & Experiments).
- Emphasizes numerical data and statistical analysis.
- Methods: Surveys, Experiments
- Advantages: Profiling data usage and behavior, highlighting differences
between groups, measuring preferences
3. Exploratory Research = Discovering new ideas (Literature & Case Studies).
- Aims to gain insights and clarify concepts.
- Approaches: Literature surveys, focus groups.
4. Descriptive Research = Describing what exists (Data Analysis).
- Describes characteristics of groups and estimates proportions.
- Involves analyzing data to find patterns.
5. Causal Research = Understanding causes and effects (Experiments).
- Focuses on determining cause-and-effect relationships through experiments
- Examines associations between changes in variables
Data Sources
- Primary Data: Collected specifically for current research (e.g., surveys, experiments).
- Secondary Data: Previously collected for other purposes (e.g., market reports,
internal records).
- Syndicated Research: Large-scale data available for subscription from a research firm
(e.g., Nielsen, Kantar).
Types of variables
- Dependent Variable is what is tried to be explained (e.g. overall liking)
- Independent Variables need to have several variations (e.g. gender, age, income)
Item vs. Variable vs. Observation
- Item: A specific question or measurement used to collect data (e.g., "How satisfied are
you with our service?").
2
, - Variable: A characteristic or attribute that can vary among subjects, represented by
one or more items (e.g., "customer satisfaction" or "age").
- Observation: A single data point recorded for a specific subject or unit, representing a
value for a variable (e.g., John's satisfaction score of "4")
Lecture 3: Measurement and Scaling: Reliability, Validity,
Dimensionality (Factor Analysis)
Measurement = Assigning numbers to characteristics of objects or people based on
predefined rules (1= women, 0 = men)
Scaling = Creating a continuum where measured objects are placed based on their attributes.
® Example: Attitude toward a brand rated from 1 (unfavorable) to 3 (favorable).
® Meaning
Measurement Scales
The identification of the right measurement scale depends on the basic nature of the
attribute (the data)
1. Nonmetric Measurement Scales
- Nominal Scale: Categories without a specific
order (e.g., gender).
- Ordinal Scale: Categories with a specific
order but no precise difference (e.g., rank
preferences or race finishers). You can
determine the order (e.g., rank-order
correlations), find the mode or median, but
you can’t perform arithmetic operations like
addition or multiplication.
2. Metric Measurement Scales
- Interval Scale: Ordered categories with equidistant differences (e.g., temperature or
IQ). You can add and subtract values and calculate averages, but you cannot calculate
meaningful ratios (e.g., you can't say that 30°C is "twice as hot" as 15°C).
- Ratio Scale: Like interval, but has a true zero point (e.g., weight, height). All
mathematical operations are valid. You can calculate means, medians, modes, and you
can make meaningful comparisons between values (e.g., "twice as much" or "half as
much").
3
, Why is it important?
Critical in determining which data analysis techniques are the most applicable to the data.
- Chi-Square determines whether there is a significant correlation between two
categorical variables à nominal
- Pearson correlation coefficient examines the strength and direction of the linear
relationship between two continuous variables à Interval & Ratio
- Linear regression à DV is interval & ratio
- Logistic regression à DV is nominal
Popular Scaling Techniques
Comparative Scaling:
- Paired Comparison: Choose one option out of two (either Heineken or Grolsch)
- Rank Order: Rank several options (1. Grolsch, 2. Heineken, 3. Bavaria)
Disadvantage: you must have a preference
- Constant Sum: Allocate points among options (Grolsch 50, Heineken 25, Bavaria 25)
Disadvantage: the more items, the more complex
Non-Comparative Scaling:
Advantage: The more people know, the more complex they will answer
Important: Always uneven scores to enable neutral response and label with numbers
- Likert Scale: Indicate degree of agreement with a series of statements
- Semantic Differential Scale: Rate perceptions on a bipolar scale with endpoints (e.g.
pleasant vs. unpleasant)
- Continuous Rating Scale (Slider Scale): Rate objects by placing a mark at the
appropriate position on a continuous line
More fine grades possible than by choosing
Multiple Item Measurement
Importance: Multiple Items better capture complex characteristics through several
questions to improve reliability and validity.
Example: Measuring Impulsive Buying Tendency with multiple related statements.
- Latent variable (construct) e.g. Impulsive Buying Tendency
- Indicator (items/questions asked)
IBT Score: Mean Score of items = Question A+B+C+D+E / 5
Reliability and Validity
Reliability
Consistency in measurement results over time (when repeated)
Testing Reliability:
- Items need Variance and should correlate with each other
- Internal consistency of questions must be given (no double negation in items)
- Methods: Test/re-test, internal consistency (Cronbach’s alpha)
4
