Lecture 1 – Collecting and treating data
Academic research
What is marketing research?
Research = studying a topic in detail to discover new information or gain a better
understanding.
In marketing practice, research is important because manager base decisions on research
results. You will read many reports from consultants. To judge whether research is good, you
must understand how it was done.
Used for segmentation, targeting, positioning, A/B testing, marketing mix analysis
Hourglass model
Academic research usually follows the hourglass model:
1) Introduction: problem introduction, problem statement, research question(s)
2) Literature review: conceptual model, hypothesis
3) Research methodology
4) Analysis & results
5) Conclusion & discussion: summary, scientific and managerial implications, limitations
and suggestions for future research
Quality drivers of academic research
Good academic research requires:
• Academic thinking = critical choices, use of scientific literature, clear
conceptualization and operationalization, correct distinction between results vs.
conclusions, and conclusions vs. implications
• Academic writing = precise and objective, clear structure and argumentation, correct
referencing (avoid plagiarism)
AI can help polish text, but can’t replace academic thinking.
General rule: everything relevant should be included, and everything included should be
relevant.
Conceptualization
Conceptualization = drawing clear boundaries around concepts to remove vagueness and
ambiguity.
What exactly does “tall” mean? Does “mouse” mean an animal or a computer device?
,The outcome of conceptualization is a conceptual model = shows
which concepts (variables) you study and how they are related.
Independent vs. dependent, antecedents and outcomes,
mediators and moderators.
This is shown in a conceptual figure
Operationalization
Operationalization = translating abstract concepts into measurable variables.
This involves choosing indicators, writing survey questions, and using standardized
measures when possible.
Conceptualization = what does X mean?
Operationalization = how do we measure X?
Collecting data
Before collecting data, you must decide:
• Who is in the population?
• How do you sample?
• Is the research exploratory, descriptive, or causal?
• Do you use qualitative or quantitative data? And survey or experiments?
• Measurement level
• Plan of analysis
Measurement levels
1) Nominal = categories → mode (transport type)
2) Ordinal = order matters → median (satisfaction scale)
3) Interval = equal distances → mean (temperature)
4) Ratio = true zero → all calculations (age, income)
Important note: the answering scale determines the measurement level, not the concept
itself (number of years, age categories, or “are you over 18?” is all the same concept).
Biased scales
Scales can push respondents toward certain answers.
Example: participants are asked to rate the quality of two types of coffee. Rating scale: 0 =
impossible to drink, 5 = ok, 7 = tasty, 10 = excellent. Now, people avoid extreme negative
options. Results may look more possible than they really are.
Experimental research design
There are three main research designs:
• Exploratory = explore ideas and processes → often qualitative
• Descriptive = describe markets or characteristics → often quantitative
• Causal = test cause-effect relationships → best method is experiments
, Causality
“X causes Y” means that X makes Y more likely, but X is not the only possible cause. Causality
can never be proven with 100% certainty → correlation ≠ causation
3 conditions for causality
1) Concomitant variation: X and Y change together
2) Time order: X happens before Y
3) No other causes: other explanations are ruled out of controlled
Random assignment
Experiments
An experiment involves:
• Manipulating one or more independent variables
• Measuring the effect on dependent variables
Advantages
• Strong control
• Allows causal inference
Within-subjects design / repeated measures design = one participant provides multiple
measurements.
• Advantages:
o Constant subject variables
o Higher statistical power
o Fewer participants needed
• Disadvantages: bias because participants may change over time:
o Maturation
o Instrumentation
o Testing
Between-subjects design = each participant is exposed to only one level of the variable.
• Advantage:
o No order effects
o Participants don’t compare treatments directly
• Disadvantages:
o Requires many participants (rule of thumb: 75 participants per cell)
o Less statistical power
o Selection bias
Academic research
What is marketing research?
Research = studying a topic in detail to discover new information or gain a better
understanding.
In marketing practice, research is important because manager base decisions on research
results. You will read many reports from consultants. To judge whether research is good, you
must understand how it was done.
Used for segmentation, targeting, positioning, A/B testing, marketing mix analysis
Hourglass model
Academic research usually follows the hourglass model:
1) Introduction: problem introduction, problem statement, research question(s)
2) Literature review: conceptual model, hypothesis
3) Research methodology
4) Analysis & results
5) Conclusion & discussion: summary, scientific and managerial implications, limitations
and suggestions for future research
Quality drivers of academic research
Good academic research requires:
• Academic thinking = critical choices, use of scientific literature, clear
conceptualization and operationalization, correct distinction between results vs.
conclusions, and conclusions vs. implications
• Academic writing = precise and objective, clear structure and argumentation, correct
referencing (avoid plagiarism)
AI can help polish text, but can’t replace academic thinking.
General rule: everything relevant should be included, and everything included should be
relevant.
Conceptualization
Conceptualization = drawing clear boundaries around concepts to remove vagueness and
ambiguity.
What exactly does “tall” mean? Does “mouse” mean an animal or a computer device?
,The outcome of conceptualization is a conceptual model = shows
which concepts (variables) you study and how they are related.
Independent vs. dependent, antecedents and outcomes,
mediators and moderators.
This is shown in a conceptual figure
Operationalization
Operationalization = translating abstract concepts into measurable variables.
This involves choosing indicators, writing survey questions, and using standardized
measures when possible.
Conceptualization = what does X mean?
Operationalization = how do we measure X?
Collecting data
Before collecting data, you must decide:
• Who is in the population?
• How do you sample?
• Is the research exploratory, descriptive, or causal?
• Do you use qualitative or quantitative data? And survey or experiments?
• Measurement level
• Plan of analysis
Measurement levels
1) Nominal = categories → mode (transport type)
2) Ordinal = order matters → median (satisfaction scale)
3) Interval = equal distances → mean (temperature)
4) Ratio = true zero → all calculations (age, income)
Important note: the answering scale determines the measurement level, not the concept
itself (number of years, age categories, or “are you over 18?” is all the same concept).
Biased scales
Scales can push respondents toward certain answers.
Example: participants are asked to rate the quality of two types of coffee. Rating scale: 0 =
impossible to drink, 5 = ok, 7 = tasty, 10 = excellent. Now, people avoid extreme negative
options. Results may look more possible than they really are.
Experimental research design
There are three main research designs:
• Exploratory = explore ideas and processes → often qualitative
• Descriptive = describe markets or characteristics → often quantitative
• Causal = test cause-effect relationships → best method is experiments
, Causality
“X causes Y” means that X makes Y more likely, but X is not the only possible cause. Causality
can never be proven with 100% certainty → correlation ≠ causation
3 conditions for causality
1) Concomitant variation: X and Y change together
2) Time order: X happens before Y
3) No other causes: other explanations are ruled out of controlled
Random assignment
Experiments
An experiment involves:
• Manipulating one or more independent variables
• Measuring the effect on dependent variables
Advantages
• Strong control
• Allows causal inference
Within-subjects design / repeated measures design = one participant provides multiple
measurements.
• Advantages:
o Constant subject variables
o Higher statistical power
o Fewer participants needed
• Disadvantages: bias because participants may change over time:
o Maturation
o Instrumentation
o Testing
Between-subjects design = each participant is exposed to only one level of the variable.
• Advantage:
o No order effects
o Participants don’t compare treatments directly
• Disadvantages:
o Requires many participants (rule of thumb: 75 participants per cell)
o Less statistical power
o Selection bias
