Definition of Experimental Research

1
Definition of Experimental Research

• In an experiment, the researcher changes one
  element, the explanatory or independent variable
  (IV), to observe the effect of that change on
  another element, the dependent variable (DV).

2
What is an Experiment?
Independent variable Cause Manipulation Experiment
al variable
Dependent variable Effect
Example Does advertising influence sales?
Advertising
Sales
3
Explanation of Causation

• Typically in an experiment, the researcher wants
  to assert that the change in the IV causes the
  change in the DV.
• To prove causation, the researcher must
  demonstrate three things
• Concomitant variation
• Appropriate time order
• Elimination of other possible causal factors

4
Explanation of Causation

• Concomitant variation - the IV and the DV must
  vary together in some predicable fashion
• For example, a positive relationship, such as an
  increase in disposable income together with an
  increase in sales of luxury cars
• Also possible, an inverse relationship, such as
  an increase in disposable income together with a
  decrease in sales of low quality hamburger

5
Explanation of Causation

• Appropriate time order - the change in the IV
  must precede the change in the DV
• A cannot cause B if A does not occur before B
  does

A
B

• For example
• to show that warm weather causes people to blow
  off class,
• you cant have more people missing class before
  the weather warms up than after it does

6
Explanation of Causation

• Elimination of Other Possible Causal Factors
• Nonspuriousness - the researcher should be
  able to eliminate any other potential
  explanations to account for the change in the DV
• Example Music at work?

7
Explanation of Causation

• Elimination of Other Possible Causal Factors -
  the researcher should be able to eliminate any
  other potential explanations to account for the
  change in the DV
• For example, if you want to argue that subliminal
  tapes made your employees work harder, youll
  have to show that it is not just the fact that
  you started playing music for the first time with
  the subliminal tapes

8
Explanation of Causation

• A study is considered confounded if there is
  more than one IV that could have caused the
  effect.
• Confounded variables provide skeptics with easy
  counter-explanations for the results
• In the subliminal tape example, the exposure of
  employees to subliminal messages is perfectly
  confounded with the introduction of music at the
  workplace

9
Evidence of Causation
Example Does advertising influence sales?
Sales
Advertising
TIME ORDER 1. Changes in advertising must
precede changes in sales. CONCOMITANT
VARIATION 2. Advertising changes must
covary with changes in sales
NONSPURIOUSNESS 3. Nothing else can explain
changes in sales EXCEPT changes in advertising.
10
Explanation of Causation

• Another way of stating nonspuriousness is to
  say that extraneous causal factors must be
  eliminated.
• We want to show that our explanation of the
  phenomena is the best and that there arent other
  factors involved.
• This is the most difficult thing to demonstrate.

11
Explanation of Causation

• Extraneous variables therefore need to be
  designed out of our experiments.
• Some examples of extraneous variables
• History
• Takes place between beginning and end of
  experiment but is not controlled by experimenter
• Maturation
• Changes in subjects that are a function of time
  (getting tired, hungry, older, etc.)

12
Explanation of Causation

• Extraneous variables continued
• Instrument variation
• Changes in the administration of the test
  measures (example interviewer bias)
• Selection bias
• The experimental group is significantly different
  from the population of interest or control group
• Mortality
• Respondents drop out during the course of the
  experiment

13
Explanation of Causation

• Extraneous variables continued
• Testing effect
• The process of experimentation produces its own
  effects on the observed responses
• Also called demand effect
• Regression to the mean
• There is an observed tendency of subjects with
  extreme behavior to move toward the average for
  that behavior during the course of the experiment

14
Basic Experimental Issues

• Experimental design and treatment
• In an experimental design the researcher has
  control over and manipulates one or more
  independent variables
• Four factors
• The IV, also known as treatment (manipulated)
• The subjects, both experimental and control
• The DV (what is being measured)
• The plan or procedure to deal with extraneous
  variables

15
Basic Experimental Issues

• Experimental Effects
• The term experimental effect refers to the effect
  of the treatment (independent) variable(s) on the
  dependent variables.
• The goal is to determine the effect of each
  treatment condition (level) on the dependent
  variable.

16
The Effect of Gender on Gays and Lesbians
Attitude Toward Advertising Content The Role of
Subcultural Code

• IV Varying degrees of gay advertising content
• Mainstream, coded, male-oriented,
  lesbian-oriented
• DV Attitude towards the ad
• The ad is
• Very bad Very good
• What is your opinion of the ad?
• Very unfavorable Very favorable
• Please rate how you felt about the ad
• Liked very much disliked very
  much
• H1 Gays and lesbians will have a more positive
  attitude towards ads with homosexual imagery than
  ads with mainstream imagery.

17
The Effect of Gender on Gays and Lesbians
Attitude Toward Advertising Content The Role of
Subcultural Code

• Method
• Pilot test 54 consumers
• Participants were presented with four ads (4
  degrees of gay advertising content)
• Reviewed ad, answered three-item attitude towards
  ad measure
• Results
• Results indicate an overall difference for
  attitude toward the ads
• Possible Problems
• Did not control for product category or brand
  type
• Manipulation check measure the perceived gayness
  of the ad content
• Presentation randomized to control for order
  effects
• Demographics

18
Limitations to Experiments

• Why arent experiments used more often?
• High cost
• Security issues
• Implementation problems

19
Experimental Validity

• Validity is measuring what you intended to
  measure. To do so, you must reduce systematic
  and random error.
• There are two types of experimental validity
• Internal validity
• External validity

20
Experimental Validity

• Internal validity refers to the extent to which
  competing explanations for the experimental
  results observed can be avoided.
• Avoid confounds

21
Internal Validity
Did the IV influence the DV? Other explanations
are threats to internal validity.
How do we control for these problems?
Randomization
Physical Control
Statistical Control
Design Control
22
Basic Experimental Issues

• There are four basic approaches to controlling
  extraneous causal factors
• Randomization in assigning subjects to treatment
  conditions
• Physical control of the extraneous factor -
  holding it constant
• Design control of extraneous factors through the
  specific type of experimental design used
• Statistical control through identifying and
  measuring the effects of the extraneous factors
  throughout the experiment

23
Experimental Validity

• External validity refers to whether the causal
  relationships measured in an experiment can be
  generalized to outside persons, settings, and
  times.
• A common example (in medicine not marketing) is
  Can studies on cancer in rats be extended to
  human cancers?

24
External Validity
Experimental Results
Population and Variables of interest
Does the experimental situation represent the
broader environment? Can we generalize to a wide
range of cases?
25
External Validity
Threats to External Validity
Non-representative sample
Reactive bias (subjects behave differently in
the experiment than in non- experimental
settings)
Non-generalizable laboratory setting
26
Experimental Settings

• The debate Laboratory experiments vs. field
  experiments.
• In a laboratory experiment, the experimenter can
  control more variables which helps prove
  causality
• In a field experiment (i.e., the test is
  conducted in a real world setting), the study is
  more realistic in terms of the marketplace

27
Internal and External Validity
Validity
Internal External
High
Low
Laboratory
Low
High
Natural Setting
Use techniques of randomization and control
(physical, design, and statistical) to increase
validity.