Research Design

1
Research Design

• Dr. John T. Drea
• Professor of Marketing
• Western Illinois University

2
Research Design

• Defined
• a master plan specifying the methods and
  procedures for collecting and analyzing
  information.
• The purpose of the research design is to ensure
• the study will be relevant to the problem.
• the study will use economical procedures.
• It is the plan for how to do the research.

3
Types of Research Designs

• Exploratory research aimed at uncovering ideas
  and possible explanations for phenomena.
• Example What are the reasons why students might
  choose to attend a womens basketball game?

4
Exploratory Research Can Be Used to...
Gain Background Information (especially when we
dont know much about a topic)
Establish Research Priorities (identifying what
needs to be addressed first)
Define Terms (such as what does customer service
mean to our customers?)
Developing Hypotheses (identifying how
things might work)
5
Types of Exploratory Research

• Secondary data analysis (a literature search)
• Experience survey interviewing key individuals
  with insight into a problem.
• Focus group 8-12 people led in a discussion,
  interaction is encouraged.
• Can be used to generate hypotheses, refine
  questions, establish response categories, test
  new product ideas
• Case Analysis a case-study approach to
  understanding a phenomenon
• Who else has faced this situation, and what
  actions and results occurred?

6
Types of Descriptive Research

• Cross-sectional study
• One time measurement (snapshot)
• Longitudinal study
• Repeated measures of the same sample
• Traditional Panels repeat the same questions
• Omnibus Panel varies the questions

Measure Time Measure
A A A A A A A A A A B C D E F G H I
1 2 3 4 5 6 7 8 9
7
Causal Research Designs

• Causal Research is concerned with
  cause-and-effect relationships between two or
  more variables.
• This typically takes the form of an experiment.
• Experiments allow us to control sources of
  possible extraneous (outside) variance.
• Does the having attended WIU Summer Music Theater
  change a consumers attitude toward Summer Music
  Theater?

8
Causal Designs

• To infer that X causes Y, four criteria must be
  satisfied
• Time sequence (temporal sequentiality) If
  changes in X causes changes in Y, then the
  occurrence of a change in X must precede in time
  a change in Y.

Time Sequence
Covariation
Systematic Elimination
Experimental Design



9
Causal Designs

• Covariation (associative variation) If
  changes in X cause changes in Y, then changes in
  X must be systematically associated with changes
  in Y. An absence of this association is evidence
  that X and Y are not causally related.

10
Causal Designs
Time Sequence
Covariation
Systematic Elimination
Experimental Design

• Systematic elimination (Nonspurious association)
  If X causes Y, the introduction of factor Z
  shouldnt make the relationship between X and Y
  disappear.

11
Causal Designs
Time Sequence
Covariation
Systematic Elimination
Experimental Design

• Experimental Design The results should fit with
  existing knowledge and theories.
• Beware the troubles of Galileo!
• The rigor of these four criteria are why we
  typically focus on the idea of descriptive
  research and that even when we attempt causal
  research, we can only infer causality.

12
Interpreting Experimental Designs

• Xi refers to the exposure of an individual or
  group to a treatment
• Oi refers to the observation/measurement process.
• R refers to a random group assignment.

The issue in experimental design is to balance
the concerns of practicality, usefulness, and
controlling extraneous variation (error).
13
Common Experimental Designs

• After-Only Design
• Weaknesses include possible confounding problems
  of history, testing.
• Control these by reducing time and tracking
  possible alternative explanations.

X O1
14
Common Experimental Designs

• Static Group Comparison

X1 O1 O2

• There is no formal means built in to such as
  design to
• insure that the two groups are equal prior to the
  
• experiment.
• This design eliminates some of the problems
  associated
• with the after only designs.

15
Common Experimental Designs

• One Group, Before-After (pre-post) Design
• O1 X O2
• Weaknesses include possible confounding problems
  of history, maturation, testing.
• Control these by reducing time and tracking
  possible alternative explanations..

16
Common Experimental Designs

• The Before-After (pre-post) w/ Control Group
  Design

R O1 X O2 R O3 O4

• This provides for random assignment to each of
  the groups (R).
• The use of a control group allows for the
  isolation of the
• effect of the experimental treatment (X).
• However, there is still the possibility of a
  testing effect, since Group 1 is exposed to a
  test and may be sensitized to the effect of
  being in an experiment.

17
Experimental Validity

• Internal Validity
• Is the change in the observed dependent variable
  (O) because of changes in the independent
  variable (X)?
• External Validity
• Are the results generalizable to the target
  population?

How do field experiments and lab experiments
relate to these two types of validity?
18
Test Marketing

• Defined an experimental study done in a field
  setting
• testing the potential for a new product
• testing variations of a new product
• Pros An accurate way of forecasting the
  success/failure of new products and product
  modifications
• Cons Expensive, and some forms can be easily
  biased by competitors

19
Types of Test Markets

• Standard test market
• Conducting using a companys normal distribution
  channels.
• Accurate.
• Expensive.
• You lose the element of surprise.

20
Types of Test Markets

• Controlled test market
• Contracted out to a research company who obtains
  access to store shelves in desired test markets.
• Fast
• You lose the element of surprise.
• Will it represent the actual distribution system?

21
Types of Test Markets

• Electronic
• Participants carry an ID card they present when
  purchasing items.
• Purchases are automatically linked to consumer
  profiles.
• Fast, less expensive, can identify repurchasers.
• Do the participants represent the population?
  (generalizability of the results)

22
Types of Test Markets

• Simulated
• Select respondents who match target market,
  expose respondents to test ads embedded with
  other ads, then give the opportunity to
  purchase/no purchase in a real/simulated store.
• Consumers re-contacted after opportunity to
  purchase to see if they repurchased and/or how
  product was used
• Information is used to create a computer model of
  likely outcomes for the product.
• Fast, confidential and less expensive, but are
  they as accurate? (Assumes some actions by
  channel members.)