Experiment Settings
*
LEARNING OUTCOMES
9–*
Basic characteristics of experiments
Experimental effects
Issues in experimental design
Manipulation of IV
Measurement of DV
Selection of test units
Different types of experiment designs
Manipulation check
Internal validity of experiments
Test-marketing
Design a basic and factorial experiment
*
Research QuestionsDoes jogging make you feel better? Does point-of-purchase advertising influence sales?Is the anti-drug campaign successful in decreasing the drug usage?What is the effect of color and lighting on shopper patronage?
Experiments
Experiments are usually conducted to investigate questions like these, with the goal to establish causal claims.
*
EXHIBIT 9.1 Experimental Conditions
in Color and Lighting Experiment
9–*
*
The Characteristics of Experiments
9–*
Subjects (a.k.a. Participants)Human respondents who provide measures based on experimental manipulation.
Experimental ConditionOne of the possible levels of an experimental (independent) variable manipulation.
Independent Variable (IV) & Dependent Variable (DV)
*
EXHIBIT 9.2 Consumer Average Patronage Scores in Each Condition
9–*
Main Effect of Color on Consumer Patronage:
Blue color attracted consumers more than did orange color.
*
EXHIBIT 9.3 Experimental Graph Showing
Interaction Effect
9–*
*
EXHIBIT 9.3 Experimental Graph Showing
Interaction Effect
9–*
The effect of color on consumer patronage is stronger when
the lighting is bright than when the lighting is soft.
*
Experimental Effects
9–*Main effect: The experimental difference in dependent variable means between the different levels of any single experimental variable. The effect of a single IV on a DV, regardless of the other IV.
Interaction effect: Differences in dependent variable means due to a specific combination of independent variable. The effect of one IV on DV depends on the level of the other IV.
*
LEARNING OUTCOMES
9–*
Basic characteristics of experiments
Experimental effects
Issues in experimental design
Manipulation of IV
Measurement of DV
Selection of test units
Different types of experiment designs
Manipulation check
Internal validity of experiments
Test-marketing
Design a basic and factorial experiment
*
All of the following are experimental design issues EXCEPT:
a. selection and assignment of subjects to treatments
b. control over extraneous variables
c. manipulation of the independent variable
d. manipulation of the dependent variable
d. manipulation of dependent variable
If I ask you to design an experiment, what would you do? What issues should you consider?
*
Basic Issues in Experimental Design
9–*Manipulation of the Independent VariableExperimental treatment: the way an experimental variable is manipulated.Categorical variables: class or quality (e.g., color)Continuous variables: quantity (level) (e.g., price)
Experimental Treatment
Experimental Group
Control Group
*
Categori ...
1. Experiment Settings
*
LEARNING OUTCOMES
9–*
Basic characteristics of experiments
Experimental effects
Issues in experimental design
Manipulation of IV
Measurement of DV
Selection of test units
Different types of experiment designs
Manipulation check
Internal validity of experiments
Test-marketing
Design a basic and factorial experiment
*
Research QuestionsDoes jogging make you feel better? Does
point-of-purchase advertising influence sales?Is the anti-drug
2. campaign successful in decreasing the drug usage?What is the
effect of color and lighting on shopper patronage?
Experiments
Experiments are usually conducted to investigate questions like
these, with the goal to establish causal claims.
*
EXHIBIT 9.1 Experimental Conditions
in Color and Lighting Experiment
9–*
*
The Characteristics of Experiments
9–*
Subjects (a.k.a. Participants)Human respondents who provide
measures based on experimental manipulation.
Experimental ConditionOne of the possible levels of an
experimental (independent) variable manipulation.
Independent Variable (IV) & Dependent Variable (DV)
*
3. EXHIBIT 9.2 Consumer Average Patronage Scores in Each
Condition
9–*
Main Effect of Color on Consumer Patronage:
Blue color attracted consumers more than did orange color.
*
EXHIBIT 9.3 Experimental Graph Showing
Interaction Effect
9–*
*
EXHIBIT 9.3 Experimental Graph Showing
Interaction Effect
9–*
The effect of color on consumer patronage is stronger when
the lighting is bright than when the lighting is soft.
*
4. Experimental Effects
9–*Main effect: The experimental difference in dependent
variable means between the different levels of any single
experimental variable. The effect of a single IV on a DV,
regardless of the other IV.
Interaction effect: Differences in dependent variable means due
to a specific combination of independent variable. The effect of
one IV on DV depends on the level of the other IV.
*
LEARNING OUTCOMES
9–*
Basic characteristics of experiments
Experimental effects
Issues in experimental design
Manipulation of IV
Measurement of DV
Selection of test units
Different types of experiment designs
Manipulation check
Internal validity of experiments
Test-marketing
Design a basic and factorial experiment
*
5. All of the following are experimental design issues EXCEPT:
a. selection and assignment of subjects to treatments
b. control over extraneous variables
c. manipulation of the independent variable
d. manipulation of the dependent variable
d. manipulation of dependent variable
If I ask you to design an experiment, what would you do? What
issues should you consider?
*
Basic Issues in Experimental Design
9–*Manipulation of the Independent VariableExperimental
treatment: the way an experimental variable is
manipulated.Categorical variables: class or quality (e.g.,
color)Continuous variables: quantity (level) (e.g., price)
Experimental Treatment
Experimental Group
Control Group
*
Categorical variables: color, present or absent of humor appeals
in advertising, type of movies
Continuous variables: price, age, customer satisfaction score, $
spent.
Experimental Design (cont’d)
9–*Manipulation of the Independent VariableSeveral
experimental treatment levels (different values of the
independent) may be used.More than one independent variable
6. may be examined.Cell: a specific treatment combination
associated with an experimental group.How to compute # of
cells in an experiment:
K = (T1)(T2)..(Tm)
*
Manipulation may sound negative, but it simply means that
experimenters vary the levels of IV systematically. For
example, if a marketing researcher is interested in the effect of
price on consumers’ buying behavior, s/he can vary the levels of
price. For example, s/he can decide using $1.99 as low price
and $2.99 as high price for a tube of lip balm, or $2.29 as low
and $3.99 as high. Again, s/he can manipulate the levels of IV
to go along with the study purpose.
Experimental Design (cont’d)
9–*Selection and Measurement of the Dependent
VariableSelecting dependent variables that are relevant and
truly represent an outcome of interest is crucial.Will outcomes
of the dependent variable (information or insights gained) assist
managers in decision making?
Sales
Ad Attitude
Recall
Purchase Intention
Brand Attitude
*
Very often, marketing managers choose sales as DV. In
addition, they (should) also consider other variables that lead to
or explain sales. For example, choosing brand attitude as DV.
7. The reasoning is that if consumers say they have
positive/favorable attitudes toward a product, then they are
more likely to buy it, which will increase sales. In Ch. 10, we
are going to discuss how we measure ATTITUDES, among other
constructs.
Experimental Design (cont’d)
9–*Selection and Assignment of Test UnitsTest units: the
subjects or entities whose responses to treatment are measured
or observed.
RandomizationRandom assignment of subject and treatments to
groupsDevice for equally distributing the effects of extraneous
variables to all conditions.
Repeated measuresExperiments in which individual subject is
exposed to more than one level of an experimental treatment.
*
Suppose a company has a new nicotine patch product and they
want to test if it works. They want to run an experiment to find
out. They create one experimental condition, where subjects
will use actual product and a control condition, where subjects
will use a placebo (what is it? See Slide 19) product. Now, let’s
say they have recruited 100 smokers to participate in the study.
How should they assign subjects to the 2 study conditions. Can
they assign the first 50 smokers to the experimental condition
and the rest of 50 to the control condition? Why or why not?
See next slide.
*
8. Methods of Random Selection
Tossing a perfect coinUsing computer programs that provide
random selectionUse random number table
*
The answer is NO. Because they may introduce other variables
that will confound the effect of nicotine patch on quit smoking,
such as willingness to quit. The smokers who sign up for the
study earlier may have stronger willingness to quit. The correct
procedure is random assignment. This slide discusses different
methods we can randomly assign subjects to conditions. For
example, we can flip a coin (decision rule: head for
experimental condition; tail for control condition) when a
subject signs in. If the coin is head, assign that person to
experiment condition, etc. We can also use random number
tables to do the random assignment. But first, be sure you make
a decision rule (e.g., even # for experimental; odd # for
control).
Demand Characteristics and Experimental Validity
9–*Demand CharacteristicAn experimental design element or
procedure that unintentionally provides subjects with hints
about the research hypothesis.Demand EffectOccurs when
demand characteristics actually affect the dependent variable.
*
9. Reducing Demand Characteristics
9–*Experimental disguisePlacebo – an experimental deception
involving a false treatment.Placebo effect – the corresponding
effect in a dependent variable that is due to the psychological
impact that goes along with knowledge that a treatment has been
administered.Isolate experimental subjectsUse a “blind”
experimental administratorAdminister only one experimental
condition per subject
Establishing Control
9–*Constancy of ConditionsSubjects in all experimental groups
are exposed to identical conditions except for the differing
experimental treatments.
CounterbalancingAttempts to eliminate the confounding effects
of order of presentation by varying the order of presentation
(exposure) of treatments to subject groups.
*
LEARNING OUTCOMES
9–*
Basic characteristics of experiments
Experimental effects
Issues in experimental design
Manipulation of IV
Measurement of DV
Selection of test units
Different types of experiment designs
Manipulation check
10. Internal validity of experiments
Test-marketing
Design a basic and factorial experiment
*
Basic versus Factorial Experimental Designs
9–*Basic Experimental DesignsA single independent variable
and a single dependent variable.
Factorial Experimental DesignsAllows for an investigation of
the interaction of two or more independent variables.
*
On IPA3, you will see examples for these two different designs.
You will be asked to explain IV, DV and experimental effects,
etc.
Laboratory and Field Experiments
9–*Laboratory ExperimentA situation in which the researcher
has more complete control over the research setting and
extraneous variables.
Field ExperimentsResearch projects involving experimental
manipulations that are implemented in a natural environment.
EXHIBIT 9.5 The Artificiality of Laboratory versus Field
Experiments
9–*High control Strong causal relationship Low control Weak
11. causal relationship
*
Within- and Between-Subjects Designs
9–*
*
Advantages of Between-Subjects Designs
9–*Within-Subjects DesignInvolves repeated measures because
with each treatment the same subject is measured.Between-
Subjects DesignEach subject receives only one treatment
combination.Usually advantageous although they are usually
more costly.Validity is usually higher.
*
*
Exercise: Questions
Suppose you wanted to test the effect of three different
email requests inviting people to participate in a survey posted
on the Internet. One simply contained a hyperlink with no
explanation, the other said if someone participated $10 would
12. be donated to charity, and the other said if someone participated
he or she would have a chance to win $100.
1. To design an experimental study, how many
experimental conditions will there be?
2. If 15 participants are needed for each experimental
condition, how many participants in total are needed for a
between-subjects design? Within-subjects design?
3. How many times does each participant need to be
measured for the dependent variable in a between-subjects
design? Within-subjects design?
*
Answer questions by yourself first, and then look at the key on
the next slide
Answers
3 experimental conditions.
45 people needed for between-subjects design; 15 needed for
within-subjects design.
3 times in within-subjects design; 1 time in between-subjects
design.
9–*
Internal Validity
9–*Internal ValidityThe extent that an experimental variable is
13. truly responsible for any variance in the dependent
variable.Does the experimental manipulation truly cause
changes in the specific outcome of interest?6 threats (factors) to
internal validity (SKIP pp. 230-232)
Manipulation ChecksA validity test of an experimental
manipulation to make sure that the manipulation does produce
differences in the independent variable.
*
Uses of Test-Marketing
9–*
Identifying Product Weaknesses
Forecasting New Product Success
Testing the Marketing Mix
Test- Marketing
*
3–*
EXHIBIT 3.2 Testing for Causes with an Experiment
*
14. Advantages and Disadvantages of Test-Marketing
9–*AdvantagesReal-world settingEasily communicated results
DisadvantagesCostTimeLoss of secrecy
*
Read the example “Hidden Valley Ranch” on p. 237.
PAGE
1
BUS 138 Marketing Research
Individual Project Assignment 3 (IPA 3): Ch. 9 Experiments
1. Suppose you wanted to test the effect of two different email
requests inviting people to participate in a survey posted on the
Internet. One said if someone participated, $10 would be
donated to charity, and the other said if someone participated,
he or she would have a chance to win $100.
1) What is the independent variable? (.5 pt)
______________________________________
2) What is a possible dependent variable? (.5 pt)
__________________________________
3) Make up some numbers in the table below and then explain a
possible main effect. (1 pt)
15. _____________________________________________________
___________________
_____________________________________________________
__________________
2. Suppose you wanted to study the effects of subliminal words
in a beer ad and consumers’ gender on the liking of the beer ad.
The table below shows the results.
21
Results
-point scale to indicate their
liking with “1” being “dislike” and “10” being “like.”
9375
PresenceAbsenceSubliminal Words (sex)MaleFemale
6648
1) What is the independent variable? (.5 pt)
____________________________________
2) What type of variable is gender? (.5 pt)
__________________________________
3) What is the dependent variable? (.5 pt)
_____________________________________
4) Is there a main effect? If yes, describe it below (1 pt):
5) Is there an interaction effect? If yes, describe it below (1.5
pts):
6) State the definition of main effect and interaction effect,
respectively (1 pt).
21
Results
16. Subjects were asked to use a 10-point scale to indicate their
liking with “1” being “dislike” and “10” being “like.” 9375
Presence
Absence
Subliminal Words (sex)
Male
Female
6
6
4
8