Transcript Chapter 8

Chapter Eight
Experiments and Test Markets
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-1
Learning Objectives
1.
2.
3.
4.
Describe and explain the importance of and
differences between the variables used in
experimentation
Explain the theoretical importance and
impact of internal, external and construct
validity in experiments.
Discuss the major types of experimental
designs used in marketing research
Explain what test markets are and how
researchers and marketing practitioners use
the resulting data
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-2
Introduction
 If the researcher wishes to develop clear insights
into why certain events occur and why they
happen under some conditions and not others,
such as:
 Predicting sales, uncovering valuable market
information, or anticipating the consequences
of a marketing program.
 Determining customer attitudes.
 Investigating cause–effect relationships.
 Then the researcher should consider using a
causal research design, such as experimental
procedures or test marketing.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-3
Select the Research Design- Causal
Research Designs
 The need for causal research
 Need substantial amounts of information from
enough members of the target population so that
inductive logic and probabilistic inferences can be
drawn.
 Causal research designs include a range of
methods:
 Observational techniques, experimentation
techniques such as pre-experimental designs, true
experimental designs, quasi-experimental designs,
field experiments and test marketing.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-4
The Nature of Experimentation
 All marketing research practices
require either manipulation or the
measurement of variables.
 Variables
 Any observable and measurable element
(or attribute) of an item or event.
 Functional relationship
 An observable and measurable systematic
change in one variable as another variable
changes.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-5
Behaviourscan
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-6
Types of Variables Used in
Experimentation
Independent variables
Dependent variables
Control variables
Extraneous variables
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-7
Variables Used in
Experimental Designs
Type of Variable
Comments
Independent
variable
Also called predictor or treatment variable (X).
An attribute or element of an object, idea, or
event whose measurement values are directly
manipulated by the researcher. The independent
variable is assumed to be the causal factor of a
functional relationship with a dependent
variable.
Dependent
variable
Also called criterion variable (Y). A singular
observable attribute or element that is the
measured outcome or effect change on specified
test subjects and is derived from manipulating
the independent variable(s).
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-8
An Example
You need to determine which is the
independent (IV) and which is the
dependent variable (DV) in the
following situation:
To determine the effect of gender
on sales.
IV………………..
DV………………
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-9
Variables Used in Experimental
Designs
Type of Variable
Comments
Control
variables
Variables that the researcher does not allow to
vary freely or systematically with independent
variables. These control variables should not
change as the independent variable is
manipulated.
Extraneous
variables
Variables that the researcher cannot control
but should average out over a series of
experiments. If not accounted for, they can
have a confounding impact on the dependent
variable measures that could weaken or
invalidate the results of an experiment.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-10
Randomisation, and the Role of Theory
Concept
Comments
Randomisation
Using randomisation, researchers can assign
subjects to different treatment conditions,
resulting in each group averaging out any
systematic effect on the functional
relationship between the IV and the DV.
Theory
A large body of interconnected propositions
about how some portion of a certain
phenomenon operates.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-11
Validity Concerns in Experimentation
Validity
The extent to which the
conclusions drawn from the
experiment are true.
Internal
Validity
The extent to which the research
design accurately identifies
causal relationships.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-12
External and Construct Validity
Concept
Comments
External validity
The extent to which a causal relationship
found in a study can be expected to be true
for the entire target population.
Construct validity The extent to which the variables under
investigation are completely and accurately
identified before hypothesising any functional
relationship.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-13
Threats to Internal Validity
History
When extraneous factors that enter the
experiment process between the first and
later manipulations affect measures of the
dependent variable.
Maturation
Changes in the dependent variable based on
the natural function of time and not
attributed to any specific event.
Testing
When learned understanding gained from
the first treatment and measures of the
dependent variable distort future treatments
and measurement activities.
Instrumentation
Contamination from changes in
measurement processes, observation
techniques and/or measuring instruments.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-14
Threats to Internal Validity
Selection bias
Contamination created by inappropriate
selection and/or assignment processes of test
subjects to experimental treatment groups.
Statistical
regression
Contamination created when experimental
groups are selected on the basis of their
extreme responses or scores.
Mortality
Contamination due to changing the
composition of the test subjects in the
experiment.
Ambiguity
Contamination from unclear determination of
cause–effect relationship.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-15
Threats to External Validity
Treatment vs
treatment
When test subjects in different treatment
groups are exposed to different amounts of
manipulations.
Treatment vs
testing
When the pre-measurement process sensitises
test subjects to respond in an abnormal
manner to treatment manipulations.
Treatment vs
selection
Generalising the results to other categories of
people beyond those types used in the
experiment.
Treatment vs
setting
Generalising the results to other environments
beyond the one used in the experiment.
Treatment vs
history
Using the existing functional relationship to
predict future phenomenon outcomes.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-16
Threats to Construct Validity
Inadequate
preoperationalisation
of variables
Contamination due to inadequate
understanding of the complete makeup of the
independent and dependent variables
included in the experimental design.
Mono-operation
bias
Contamination created by using only one
method to measure the outcomes of the
dependent variable.
Monomethod bias
Contamination due to assessing multiattribute
treatment manipulations (independent
variables) using single-item measuring
instruments.
Hypothesis-guessing
Contamination by test subjects believing they
know the desired functional relationship prior
to the manipulation treatment.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-17
Construct Validity
Evaluation
apprehension
Contamination caused by test subjects being
fearful that their actions or responses will
become known to others.
Demand
characteristics
Contamination created by test subjects trying
to guess the true purpose behind the
experiment, thus giving abnormal socially
acceptable responses or behaviours.
Diffusion of
treatment
Contamination due to test subjects discussing
the treatment and measurement activities to
individuals yet to receive the treatment.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-18
Improving the Internal and External
Validity of experimentation
Inclusion of control
groups
Represent the greatest strength of the
experiment and the best way to ensure internal
validity.
Time order of the
manipulation
exposure
The researcher decides which variables will
occur first. Either though pre-experiment
measures or control groups, prevents
influencing the dependent variable before the
manipulation.
The researcher may select only test subjects
Exclusion of nonsimilar test subjects who have similar and controllable
characteristics.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-19
Improving the Internal and External Validity of
Experimentation
Matching extraneous variables
By matching, the researcher
measures certain
extraneous variables
individual by individual, and
assigns those who respond
similarly to the experimental
and control groups.
Randomisation of test subjects
to treatment groups
The Researcher must ensure
randomness in a carefully
controlled manner.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-20
Types of Experimental Designs
Three groups:
1. Pre experiments
2. True experiments
3. Quasi experiments
They vary by the degree of
control that the researcher can
exercise in the design and
execution.
See exhibit 8.3
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-21
Types of Experimental Designs:
Pre-experimental
One-shot study
A single group of test subjects is exposed to the
independent variable treatment X, and then a
single measurement of the dependent variable
is taken (o1).
One-group,
pretest–posttest
First a pretreatment measure of the dependent
variable is taken (o1), then the test subjects are
exposed to the independent treatment X, and then
a post-treatment measure of the dependent
variable is taken (o2).
Static group
comparison
There are two groups of test subjects: one group
is the experimental group (EG) and is exposed to
the independent treatment, and the second group
is the control group (CG) and is not given the
treatment. The dependent variable is measured in
both groups after the treatment.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-22
Types of Experimental Designs:
True Experimental Designs
Pretest–posttest, control
group
Test subjects are randomly assigned to either the
experimental or control group, and each group
receives a pretreatment measure of the dependent
variable. Then the independent treatment is
exposed to the experimental group, after which
both groups receive a post-treatment measure of
the dependent variable.
Post-test-only,
control group
Test subjects are randomly assigned to either the
experimental or the control group. The
experimental group is then exposed to the
independent treatment, after which both groups
receive a post-treatment measure of the dependent
variable.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-23
Types of Experimental Designs:
True Experimental Designs
Solomon
Four Group
This design combines the ‘pretest–post-test,
control group’ and ‘post-test-only, control group’
designs and provides both direct and reactive
effects of testing. Not used in marketing research
practices because of complexity and lengthy
time requirements.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-24
Types of Experimental Designs:
Quasi-experimental Designs
Non-equivalent
control group
This design is a combination of the ‘static group
comparison’ and the ‘one-group, pretest–posttest’ pre-experimental designs.
Separate-sample,
pretest–post-test
Two different groups of test subjects are drawn;
neither group is directly exposed to the
independent treatment variable. One group
receives a pretest measure of the dependent
variable. Then after the insignificant independent
treatment occurs, the second group of test
subjects receives a post-test measure of the
dependent variable.
Field
experiment
This is a causal design that manipulates the
independent variables in order to measure the
dependent variable in the natural setting of the
event or test.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-25
Test Marketing
A field experiment executed in a
market setting.
For example, test markets for new
products allow the marketer to
check for its likelihood of success,
and indicates potential
improvements that need to be
made to the marketing mix.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-26
Test Marketing
Types of Test
Marketing
Comments
Traditional
test markets
Also referred to as ‘standard’ tests; these use
experimental design procedures to test a
product and/or a product’s marketing mix
variables through existing distribution
channels.
Controlled
test markets
Tests that are performed by an outside
research firm that guarantees distribution of
the test product through prespecified outlets in
selected cities.
Electronic
test markets
Tests that integrate the use of select panels of
consumers who use a special identification
card in recording their product-purchasing
data.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-27
Test Marketing
Types of Test
Marketing
Comments
Simulated
test markets
Also referred to as ‘laboratory tests’ or ‘test
market simulations’. These are quasi experiments
where test subjects are preselected, then
interviewed and observed on their purchases and
attitudes towards the test product.
Web-based TV
test markets
Similar to electronic test markets; these use
broadband interactive TV (iTV) and advances in
interactive multimedia communication technologies
to conduct the field experiment. Preselected
respondents are shown various stimuli and asked
questions online through their iTV.
Virtual
test markets
Tests that are completely computerized, allowing
the test subjects to observe and interact with the
product as though they were actually in the test
store’s environment.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-28
Advantages and Disadvantages of
Test Markets
Advantages of Test Markets
 Traditional test markets
 Conducted in actual
distributions channels.
 Can determine both
customer acceptance
and trade support.
 Controlled test markets
 Distribution is assured.
 Costs are lower.
 Competitive monitoring
is difficult.
Disadvantages of Test Markets
 Traditional test markets
 Cost, time and exposure to
competition.
 Competitors can ‘rush to
market’.
 Controlled test markets
 Limited number of markets.
 Trade support is unknown
when incentives are
removed.
 The effect of advertising is
difficult to evaluate.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-29
Advantages and Disadvantages of
Test Markets
Advantages of Test Markets
 Simulated test markets
 Cost and time savings.
 Can predict trial,
repurchase and purchase
cycle accurately.
 Minimise exposure to
competition.
 Virtual test markets
 ‘Stores’ closely resemble
reality.
 Can be changed quickly.
 No exposure to
competition
Disadvantages of Test Markets
 Simulated test markets
 Isolation from real world.
 Broad-based customer
reaction is difficult to
measure.
 Virtual test markets
 Will consumers shop as in
real stores?
 Cost of equipment to
conduct tests.
 Lack of ‘touch’.
Copyright  2007 McGraw-Hill Pty Ltd
PPTs t/a Marketing Research 2e by Hair, Lukas, Bush and Ortinau
Slides prepared by Judy Rex
8-30