Causal Designs

1
Chapter 7

• Causal Designs

2
Issues Addressed

• conditions of causality
• what is an experiment?
• differences between a lab vs field experiment
• two types of experimental validity
• extraneous factors that affect validity
• types of experimental designs
• test marketing and types of test markets

3
Conditions of Causality

• Concomitant variation
• extent to which the cause (X) and the effect (Y)
  occur together
• e.g., advertising and sales
• Time Order of Occurrence of Variables
• X has to occur before Y
• Elimination of Effects of Other Factors
• for X to cause Y, X has to be associated with Y
  even after the effects of all other factors have
  been accounted for

4
What is an Experiment?

• is a scientific investigation in which one or
  more independent variables are manipulated and
  the effect on a dependent variable is observed.
• other potential independent variables are
  controlled
• class to come up with an experiment

5
Field versus Lab Experiments

• Lab Experiment
• create a situation with desired conditions
• manipulate some variables while controlling
  others
• examine the dependent variable
• Field Experiment
• research study in a natural setting
• manipulate some variables
• under as carefully controlled conditions as the
  situation would permit
• examine the dependent variable

6
Experimental Validity

• Internal Validity
• extent to which the researcher can attribute the
  effect observed to the experimental variable and
  not to other factors
• effect of price on sales
• other factors (ad, display space, store traffic,
  competitor prices)
• these can be maintained constant in a lab
  experiment

7
Experimental Validity (cont.)

• External Validity
• extent to which the results of the experiment can
  be generalized to other situations
• e.g. simulated shopping trip and problems

8
Extraneous Factors Affecting Validity

• History
• Maturation
• Testing
• Instrument Variation
• Selection Bias
• Experimental Mortality

9
History

• specific events, external to the experiment
• occurring at the same time as the experiment
• that may affect the dependent variable
• e.g., refrigerator pricing
• usually difficult to identify history factors

10
Maturation

• changes occurring within the test units
• which are not due to the effects of the
  experimental variable
• e.g., tired, hungry, not attentive
• effects observed may then be partly attributable
  to the changes in the test unit
• particularly relevant in longitudinal studies
• respondents may undergo changes in taste, family
  status, knowledge level

11
Testing

• the process of experimentation itself affects the
  observed response
• main testing effect
• effect of a prior observation on a later
  observation
• may be a result of respondent desire to be
  consistent
• can occur in both longitudinal and
  cross-sectional studies
• affects internal validity of the experiment

12
Testing (cont.)

• interactive testing effect
• effect of a prior measurement on the test units
  response to the independent variable
• e.g. attitude toward a brand
• sensitivity to that brands information
• measured effects are then not generalizable to
  the population
• affects external validity of the experiment

13
Instrument Variation

• effect of changes in the measuring instruments on
  what is being measured
• could also occur because of variability in
  administration
• pricing study

14
Selection Bias

• occurs when it is difficult to certify that
  groups of test units are equivalent
• matching and randomization techniques

15
Experimental Mortality

• loss of test units during an experiment
• comparison of units lost with units retained

16
Controlling Extraneous Variables

• randomization
• matching
• statistical control
• design control

17
Types of designs

• Preexperimental
• True Experimental
• Quasi-experimental

18
Pre-experimental Designs

• One Shot Case Study
• One Group Pretest-Posttest Design
• Static Group Design

19
True Experimental Designs

• Pretest-Posttest Control Group Design
• Posttest-Only Control Group Design

20
Quasi-Experimental Design

• Time Series Design

21
Market Test

• is a controlled experiment, done in a limited but
  carefully selected part of the marketplace, whose
  aim is to predict the sales or profit
  consequences ... of one or more proposed
  marketing actions
• is the final gauge of consumer acceptance
• three out of four products test marketed succeed,
  while four out of five that have not been test
  marketed fail

22
Types of Test Markets

• Standard Test Markets
• products sold thru normal distribution channels
• results monitored by a standard syndicated
  service
• Controlled Test Markets
• entire test program is done by an outside service
• Electronic Test Markets
• variant of controlled markets
• use of consumer panels to measure purchases and
  TV viewing behavior
• Simulated Test Markets
• use of a simulated shopping environment
• fast, cheap, protection from competitors