MR6A - Experimental Surveys

Marketing Research
6A – Explaining causality: the principles of experimental surveys

2022-2023
Fabienne Berger-Remy
QUAL BIG DATA & AI DESCRIPTIVE CAUSAL
Objectives EXPLORE - EXPLORE – DESCRIBE - EXPLAIN -
WHY? HOW? WHAT? HOW MUCH / WHAT’S THE
Get an in-depth Retrieve HOW MANY? IMPACT?
understanding, behavioral and Quantify, Determine
find new and consumption describe, make causes
great ideas patterns clusters
Characteristics Often at the Evolutive, based Based on a set of Based on a set of

beginning on machine hypothesis hypothesis
Flexible, learning Structured Manipulation of
evolutive design variables
Tools, Document Self-learning Survey on Quantitative

methods retrieval algorithm secondary data surveys
Experts analysis running through such as panels Experiments
Interviews massive amount Quantitative
(individual and of data surveys, tracking
collective) Social media Statistical or
Observation listening mechanical
Ethnographic Predictive observations
surveys analysis
The problem with descriptive
research
• A link between 2 variables does not
necessarily imply a causality…
3
Why?
Random effect? The set of variables is not

2 variables can be comprehensive
accidentaly correlated => latent variables?
https://www.tylervigen.com/spurious- Ex: sleeping with shoes on correlates
correlations with the headache prevalence
Latent variable (missing): alcohol
consumption at a party before going
to bed….
4
Causal research
REAL LIFE:
MARKET TEST
SIMULATED:
EXPERIMENTS
5
Causal research
• 2 sample groups are constituted randomly (so that

they are perfectly comparable)
• One group is the control group (no treatment), the
other is the intervention group (treated)
• Then, the variable (« treatment ») is manipulated
outside the judgement of the interviewee
• The difference of outcome between the 2 groups is
only due to the manipulated variable (the
« treatment ») => cause-to-effect relationship
7
8
Experimental design
• Outcome variable: A variable observed in an experiment that
is not under the direct control of the researcher, i.e. the
resulting variable.
– Ex in marketing research: (A) sales level, (B) brand awareness
• Manipulated variable(s): controlled by the researcher; serve

as inputs, i.e. the cause.
– (A) Price can be a manipulated variable so as to determine the effect
on sales
– (B) Advertising campaign can be manipulated (with / without) so as to
determine the effect on brand awareness
Example of experiment
Manipulated variable
Perfume likeability
Neutral Nice
perfume perfume Outcome
variables:
Manipulated variable
- purchase
Perceived
Care perception
80 women 80 women intention

as caring
- would
recommend to a
Not friend
perceived 80 women 80 women
as caring
2 control groups
11
Analysis by comparisons
• Inference and hypothesis testing

– Ex: describe behavioral factors by comparing a « control
group » with several « experimental groups »
• Most common analysis:
– mean comparison
– two sample comparison => student t test
– multiple samples comparison: ex / Newman&Keuls
– ANOVA (variance analysis) => Fisher test
Dumex case
Dumex is an infants' and children's nutrition company. Its products span
the maternal and growing-up milk categories, and are sold mainly in South-
East Asia and China. Dumex carry milk for babies and infants through toddlers
6 years old, nutritional supplements for expecting and lactating mothers and a
full cream milk with added nutrition for adults.
Nestlé, the world's largest food company, plans to raise the price of some
milk powders by 5 to 10 percent in China, a move that may be copied by other
dairy companies.
Given the Nestlé announcement on the Chinese market, the management
team of Dumex Vietnam is expecting the same rise on the Vietnamese market.
They plan to do an experiment to be able to take the right decision regarding
their pricing strategy.
13
Experimental design
Manipulated variable Outcome variable

• Price of Dumex • Intention to buy Dumex
• Price of Nestlé • Intention to buy Nestlé
Sampling criteria
• Mother of at list 1 child aged from 6 month to 1 year

– Quota 50% primipare, 50% multipare
• Buyer of infant milk powder at least occasionnally
– Quota on product / brand usage: Has bought at least 1 box
in the last 3 month, 1/3 Dumex exclusive, 1/3 Nestlé
exclusive, 1/3 mixing brands
• Data collection device: internet website, or e-mail on
maternity list
Current price of Price increase 5% Price increase 10%
Nestlé
Current price of Nestlé: 450 000 VND Nestlé: 472 500 VND Nestlé: 495 000 VND
Dumex: 500 000 VND Dumex: 500 000 VND Dumex: 500 000 VND
Dumex
Price increase 5% Nestlé: 450 000 VND Nestlé: 472 500 VND Nestlé: 495 000 VND
Sample size: 1080 persons
Current price of Price increase Price increase
Nestlé 5% 10%
Current price of 90 90 90
Dumex
Price increase 90 90 90
5%
7%
10%
Nestlé 5% 10%
Current price of
Dumex
Price increase
5%
Price increase
7%
Price increase
10%
450 000 VND
• Would you consider to buy that product?
• Answer on a 5 point scale
– (5) I definitely would buy that product
– (4) I probably would buy that product
– (3) I might buy that product
– (2) I probably wouldn't buy that product
– (1) I definitely wouldn't buy that product
500 000 VND
Nestlé 5% 10%
Current price of
Dumex
Price increase
5%
Price increase
7%
Price increase
10%
472 500 VND
535 000 VND

MR6A - Experimental Surveys

Uploaded by

Copyright:

Available Formats

MR6A - Experimental Surveys

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

MR6A - Experimental Surveys

Uploaded by

Copyright:

Available Formats

Marketing Research

6A – Explaining causality: the principles of experimental surveys

Characteristics Often at the Evolutive, based Based on a set of Based on a set of

Tools, Document Self-learning Survey on Quantitative

Random effect? The set of variables is not

• 2 sample groups are constituted randomly (so that

• Manipulated variable(s): controlled by the researcher; serve

80 women 80 women intention

• Inference and hypothesis testing

Manipulated variable Outcome variable

• Mother of at list 1 child aged from 6 month to 1 year

You might also like