Experimental Research 2025
Lecture 1 2
Problem identification 5
Lecture 2 - Experimental Designs 6
Conceptual models 7
Experimental designs 9
Manipulation the IV 11
Reliability vs. validity 15
Measuring the DV (and the mediator) 16
Lecture 3 - Experimental Design: 18
Internal validity: 18
Threats to internal validity 19
Hypothesis testing 23
Intuition behind ANOVA 24
Lecture 4 - one and 2 way anova 39
One-way ANOVA 40
Two-way ANOVA 47
Two-way ANOVA: How it’s done 49
Lecture 5 - 3-way anova, ancova 57
3-Way ANOVA 61
ANCOVA 66
Lecture 6 - Repeated Measures Mixed Design 73
T-Test 75
1-way Repeated Measures Anova 75
Mixed Design 81
Lecture 7 - Mediation 89
Mediation Analysis 89
Mediation analysis SPSS using regression 94
Mediation analysis using process 95
Mediation by experimentation (process through moderation) 100
Lecture 8 - Power and effect size 102
Hypothesis testing 102
Power 104
Increase power through sample size (power analysis) 106
Increasing power through effect size (improving your design) 109
Lecture 9 - MODERATION WITH CONTINOUS VARIABLES SPOTLIGHT ANALYSIS 116
Regressions 118
Multiple regression with interaction 123
Spotlight Analysis 127
Example with two continuous variables 132
Lecture 10 - Replication crisis and flexible data analysis 138
, Flexible data analysis 139
Confound or covariate: revisited 144
Lecture 1
Types of marketing research
Experiments are very narrow and have ‘small’ questions so you can really dig into
your research. Surveys are very broad and have ‘large’ questions
Survey methods:
● Cross-sectional: This method captures data from a population at a
specific point in time to understand the current status or opinions of the
sample group. (Election votes)
● Longitudinal: This involves repeatedly collecting data from the same
subjects over an extended period to track changes and developments over
time. (job satisfaction of employees)
● Panel: Similar to longitudinal surveys, panel surveys also gather data from
the same group of people over time but typically focus on specific topics,
allowing for deeper insights into changes in attitudes or behaviors within
that subject area. (preference in smartphones as new models are
released)
Why experimenting?
In marketing (and other sciences) we seek to:
– Describe
– Predict
– Explain behavior of (market-parties, employees, buyers)
• => Break up phenomena in variables and relations between those variables
Behavioral Research
1. Descriptive research: Thoughts, feelings, ideas, behaviors
,2. Correlational research: Identifying relationships between different observed
variables: measuring thoughts, feelings, behavior
– Examples:
• “people save more during the economic crisis”
• “smoking mothers have more often problem children
= The more the mother smokes, the more negative behavior
= This means the more the mother smokes, the less the negative behavior
(negative correlation = one goes up the other goes down)
Problems with correlational research:
1. Direction of relationship
2. Spurious correlations: when two variables are statistically related but not
causally related
(are often interpreted as causal)
-
, - Is the number of murders committed caused by the number of ice
creams eaten?
- Do criminals celebrate their crimes with an ice cream?
- More people on the street
- • More open windows
- • More rapes
- Third explaining variable
- => spurious correlation
From description to prediction
1. Description: careful observation
2. Correlation: relationship between observed variables
3. Experimental: Testing causality, A > B?
Experimental research: settings
Field experimentation
– Real-life setting
– Measure of external validity: natural behaviour | setting | treatment
– Less control
Laboratory experimentation
– High control (strip other influences so you can isolate the variables ur testing)
– Better able to manipulate variables
– No natural setting
Practical example
“Why did milk sales drop 25%? Taste change or redesign?”
Experiment 1 (blind taste test): New taste better liking
Experiment 2 (package liking & eye tracking): Redesign better
So what was the problem?
- Field experiment (camera observation): search time increases >
consumers unable to find ‘their’ brand due to redesign.
Experimental Research: Crux (point of difficulty)
● Experimentation: the only type of research that (potentially) can
demonstrate that a change in one variable causes a predictable change in
another variable
● Most difficult: making sure that a change in Y was not caused by
something else than X
Experimentation & Causation
1. Needs to be correlation between 2 variables
2. Asymmetrical direction: goes from X to Y and not also another way
4. Change in A is accompanied by a change in B
Lecture 1 2
Problem identification 5
Lecture 2 - Experimental Designs 6
Conceptual models 7
Experimental designs 9
Manipulation the IV 11
Reliability vs. validity 15
Measuring the DV (and the mediator) 16
Lecture 3 - Experimental Design: 18
Internal validity: 18
Threats to internal validity 19
Hypothesis testing 23
Intuition behind ANOVA 24
Lecture 4 - one and 2 way anova 39
One-way ANOVA 40
Two-way ANOVA 47
Two-way ANOVA: How it’s done 49
Lecture 5 - 3-way anova, ancova 57
3-Way ANOVA 61
ANCOVA 66
Lecture 6 - Repeated Measures Mixed Design 73
T-Test 75
1-way Repeated Measures Anova 75
Mixed Design 81
Lecture 7 - Mediation 89
Mediation Analysis 89
Mediation analysis SPSS using regression 94
Mediation analysis using process 95
Mediation by experimentation (process through moderation) 100
Lecture 8 - Power and effect size 102
Hypothesis testing 102
Power 104
Increase power through sample size (power analysis) 106
Increasing power through effect size (improving your design) 109
Lecture 9 - MODERATION WITH CONTINOUS VARIABLES SPOTLIGHT ANALYSIS 116
Regressions 118
Multiple regression with interaction 123
Spotlight Analysis 127
Example with two continuous variables 132
Lecture 10 - Replication crisis and flexible data analysis 138
, Flexible data analysis 139
Confound or covariate: revisited 144
Lecture 1
Types of marketing research
Experiments are very narrow and have ‘small’ questions so you can really dig into
your research. Surveys are very broad and have ‘large’ questions
Survey methods:
● Cross-sectional: This method captures data from a population at a
specific point in time to understand the current status or opinions of the
sample group. (Election votes)
● Longitudinal: This involves repeatedly collecting data from the same
subjects over an extended period to track changes and developments over
time. (job satisfaction of employees)
● Panel: Similar to longitudinal surveys, panel surveys also gather data from
the same group of people over time but typically focus on specific topics,
allowing for deeper insights into changes in attitudes or behaviors within
that subject area. (preference in smartphones as new models are
released)
Why experimenting?
In marketing (and other sciences) we seek to:
– Describe
– Predict
– Explain behavior of (market-parties, employees, buyers)
• => Break up phenomena in variables and relations between those variables
Behavioral Research
1. Descriptive research: Thoughts, feelings, ideas, behaviors
,2. Correlational research: Identifying relationships between different observed
variables: measuring thoughts, feelings, behavior
– Examples:
• “people save more during the economic crisis”
• “smoking mothers have more often problem children
= The more the mother smokes, the more negative behavior
= This means the more the mother smokes, the less the negative behavior
(negative correlation = one goes up the other goes down)
Problems with correlational research:
1. Direction of relationship
2. Spurious correlations: when two variables are statistically related but not
causally related
(are often interpreted as causal)
-
, - Is the number of murders committed caused by the number of ice
creams eaten?
- Do criminals celebrate their crimes with an ice cream?
- More people on the street
- • More open windows
- • More rapes
- Third explaining variable
- => spurious correlation
From description to prediction
1. Description: careful observation
2. Correlation: relationship between observed variables
3. Experimental: Testing causality, A > B?
Experimental research: settings
Field experimentation
– Real-life setting
– Measure of external validity: natural behaviour | setting | treatment
– Less control
Laboratory experimentation
– High control (strip other influences so you can isolate the variables ur testing)
– Better able to manipulate variables
– No natural setting
Practical example
“Why did milk sales drop 25%? Taste change or redesign?”
Experiment 1 (blind taste test): New taste better liking
Experiment 2 (package liking & eye tracking): Redesign better
So what was the problem?
- Field experiment (camera observation): search time increases >
consumers unable to find ‘their’ brand due to redesign.
Experimental Research: Crux (point of difficulty)
● Experimentation: the only type of research that (potentially) can
demonstrate that a change in one variable causes a predictable change in
another variable
● Most difficult: making sure that a change in Y was not caused by
something else than X
Experimentation & Causation
1. Needs to be correlation between 2 variables
2. Asymmetrical direction: goes from X to Y and not also another way
4. Change in A is accompanied by a change in B
