Topic: Research Design

1
Topic Research Design

• Introduction
• Any researcher who wants to investigate a
  research question or test a hypothesis faces some
  important problems that must be solved before
  he/she can start his research or study. These are
  the problems
• Which research strategy should be used?
• What kind of data will be needed to investigate
  the research questions?
• How will the data be analyzed?
• The research design is the blueprint that
  enables the researcher to come up with solutions
  to these problems and guides him/her in the
  various stages of his/her research.

2
Topic Research Design

• Sub-topic Types of Research Design
• The Experimental Design
• How is it done? Or the steps.
• Hypothesis Training improves employee
  productivity.
• Step 1 Divide your subjects into two groups the
  experimental (treatment) group and the control or
  comparison group.
• Step2 Measure the groups on the dependent
  variable. In other words, we measure each groups
  productivity.

3
Topic Research Design

• Step 3 We give the treatment (training) to the
  experimental group. The control group gets no
  treatment or training.
• Step 4 We measure the two groups on the
  dependent variable. In other words, we measure
  their productivity again.
• Conclusion If productivity of the experimental
  group increases after the treatment, we can then
  conclude that the treatment or training improves
  workers productivity.

4
Topic Research DesignsSub-topic Types of
Research Design

• Hypothesis Giving commission to employees
  improves their performance.
• Step 1
• Divide subjects into two groups Experimental and
  control groups.
• Step 2 Measuring the two groups on the dependent
  variable
• Group Performance Before Treatment
• Experimental 50 percent
• Control 50 percent
• Step 3 Give commission to the experimental
  group. The control group gets no commission.
• Step 4 Measuring the two groups on the dependent
  variable
• Group Performance After Treatment
• Experimental 70 percent
• Control 50 percent
• Results and Conclusion The performance of the
  experimental group increases with commission.
  Therefore, commission improves employee
  performance.

5
Topic Research DesignsSub-topic Types of
Research Design

• 1. Experimental Design
• Evaluation of the Experimental Design
• Does it have internal validity?
• Internal validity is the most important type of
  validity. It talks about causal relations. In
  other words, it tells us whether or not the
  independent variable (IV) causes changes in the
  dependent variable (DV). This is the purpose of
  any experiment.

6
Topic Research DesignsSub-topic Types of
Research Design

• Answer
• It will have internal validity if
• We use randomization to select members for both
  the experimental group and control group. When
  use randomization, we make sure that every member
  of a population has an equal chance of being
  selected for the group. The purpose of
  randomization is to equalize the two groups.
• We eliminate threats to internal validity
• (i) Maturation This means that people change
  overtime. If your experiment takes time (such as
  a year), the members of the experimental group
  and the control group may change, and this change
  may affect the dependent variable in your
  experiment.

7
Topic Research Design

• Example
• This is example shows the effect of maturation on
  the dependent variable.
• We want to see if piece rate pay (IV) increases
  workers productivity (DV). Suppose that this
  experiment is done for one year.
• After one year, we will not be sure whether any
  increase in productivity is due to the
  introduction of piece rate pay or to the fact
  that the workers have become mature in doing
  their jobs.
• How do get rid of maturation? Answer we have to
  use a control group.

8
Topic Research DesignSub-topic Types of
Research Design

• (ii) Mortality This means that some members of
  the experimental and/or control group drop out of
  the experiment for some reason.

9
Topic Research DesignSub-topic Types of
Research Design

• Experimental Research Design
• Ethical Issues
• In experiments, subjects need to be debriefed.
  Debriefing means that we (i.e., the researchers
  who do the experiments) are obliged to inform the
  subjects about the purpose of the experiment,
  that the information that they give you will be
  confidential, and that they will not be deceived
  (i.e., the information will be used only for
  research and not for any other purpose).
• Debriefing is usually done immediately after the
  experiment.
• What is debriefing in experimental research?

10
Topic Research DesignSub-topic Types of
Research Design

• 2. Cross-Sectional Design
• Characteristics
• a. It involves more than one case. Usually, we
  study many cases. For example, we study 60
  countries to see whether there is a link between
  democracy and economic development. We need those
  many cases to see if there is variation. For
  example, if democracy is at the same in all 60
  countries, then there is no variation. Hence, we
  cannot study those cases. This means that we need
  variation (i.e., differences) on a variable such
  as democracy in order to study it.

11
Topic Research DesignSub-topic Types of
Research Design

• Country Dem Econ Dev
• Sudan 60 50
• Syria 70 60
• Lebanon 80 65
• UAE 80 70
• Iran 87 75
• This is data for a cross-sectional study. Here,
  we have 5 cases which are the countries. We also
  have enough variation. The more cases you have,
  the more the variation there will be.

12
Topic Research DesignSub-topic Types of
Research Design

• b) All the cases represent events happening at
  one point of time such as a specific year like
  2005.
• Example
• Data on Dem and Economic Dev for 5 countries for
  2005.
• __________________________________________________
  _
• Country Dem Econ Dev
• Sudan 60 50
• Syria 70 60
• Lebanon 80 65
• UAE 80 70
• Iran 87 75
• __________________________________________________
  ___

13
Topic Research DesignSub-topic Types of
Research Design

• c) When we do cross-sectional studies, we are not
  seeking causation or causal relationships. We are
  just looking for association or correlation or
  link. From cross-sectional designs, we get
  correlation, but from experimental designs we get
  causation because in experimental design we have
  control over the experimental conditions.

14
Topic Research DesignSub-topic Types of
Research Design

• Evaluation of Cross-sectional Design
• It is likely to be strong on external validity or
  generalizability. This is likely to be the case
  if the cases involved in a cross-sectional study
  are many. This means that the more cases you have
  a cross-sectional the stronger will be external
  validity. So if you have two studies one with 15
  cases and the other with 200 cases, the latter
  will have stronger external validity.
• The Cross-sectional design is not strong on
  internal validity or causality, because we do not
  control for factors that could affect the
  dependent variable.

15
Topic Research DesignSub-topic Types of
Research Design

• Exercise Question
• Contrast between experimental design and
  cross-sectional design?
• To answer this question, we have to discuss the
  differences between the two designs.
• The following are the differences
• In cross-sectional designs, we do not have
  experiments whereas we have these things in
  experimental designs
• Experimental designs are very strong on internal
  validity while cross-sectional designs are weak
  on internal validity.

16
Topic Research DesignSub-topic Types of
Research Design

• 3. Longitudinal Research Designs
• Characteristics
• a. Here the time dimension or element is
  important
• b. We collect data for events which happened at
  two points of time at least
• Example Suppose that I am studying the extent to
  which the introduction of TQM improves
  productivity in some business organization.
  Suppose further that TQM was introduced in that
  organization in 2000.
• To answer this question, I have to collect data
  on productivity in 2000 when TQM was introduced
  and also collect data on productivity after TQM
  was introduced (e.g., in 2006).

17
Topic Research DesignSub-topic Types of
Research Design

• Example
• Suppose I am studying the effect of inflation on
  business profitability in KFC. We could use a
  longitudinal research design here. My data may
  look like this
• Year Inflation
  Profitability
• 10 50
• 13 70
• 15 80
• 20 100
• 22 240

18
Topic Research DesignSub-topic Types of
Research Design

• C. Variation is established by collecting data
  for events happening at two points of time at
  least. Variation here means differences in the
  values of variables. If there is no variation,
  then we cannot do studies.
• Variation in the values of our variables is
  needed in all research designs. Remember No
  variation, no research.

19
Topic Research DesignSub-topic Types of
Research Design

• Evaluation of the Longitudinal Research Design
• It is a little strong on internal validity or
  causality. It is stronger than cross-sectional
  research design on internal validity. However, it
  is not strong as experimental design.
  Experimental design is much stronger on internal
  validity than longitudinal design, why?
• Answer In experimental designs, we have control
  over experiment. We do not have this kind of
  control in longitudinal design.
• b. In longitudinal designs, we attempt to control
  for external factors by using data analysis
  techniques such as regression analysis.