BUAD801 Research Methodology
Experimental Research
Chapter Overview
This chapter discusses:
• Experimental research
• Conditions for establishing cause-and-effect
• Key features of experiments
• Design of experiments
Experimental Research
• An experiment is a method of data collection designed
to test hypotheses under controlled conditions
– Using controlled environments allows researchers to isolate a
single stimulus (independent variable) as the cause of
measurable differences in a responses (dependent variable).
– Involves manipulating one or more independent variables
and measure their effect on a dependent one or more
dependent variables
– It is useful when the researcher wants to test cause-effect
relationships
• Two types of experiments
– True experiment
– Quasi-experiment
Conditions for Establishing Cause-and-Effect
• There are four conditions for establishing cause-and-
effect relationship in an experimental research.
– Covariation: the independent variable and the dependent
variable should covary
– Precedence: the independent variable must precede the
dependent variable
– Effect: no other factor should be a possible cause of the
dependent variable other than the independent variable
– Explanation: there should be a logical explanation about why
the independent variable affects the dependent variable
Key Features of Experiments
• There are three important features of experimental
research design:
– Manipulation of the independent variables
– Controlling extraneous variables
– Randomization
Manipulation of the Independent Variables
• Manipulation is the process by which the researchers
purposefully alter or influence an independent variable in an
experimental design
– The independent variable is the treatment or intervention variable
– This independent variable is manipulated and the effects are observed
in the dependent variable.
– Different groups are exposed to different treatments
• Manipulation may include giving:
– Treatment to one group but not the other,
– Different treatments to comparison groups, or
– Different levels of the treatment to different groups
• The different treatments of levels are called conditions
• Manipulation affects the reliability of a study
Manipulation of the Independent Variables
• Manipulation Example
– Impact of interactive virtual branding on purchase intentions
• The independent variable to be manipulated is interactive virtual
branding
– Manipulation: a key feature to manipulate is content interactivity
• Two treatment conditions are possible in relation to the manipulation
of interactivity in virtual branding
1. A non-interactive virtual reality involving the brand
2. An interactive virtual reality involving the brand
• The information content (factual and relevant product and brand data,
such as materials used, weight or care instructions or dimensions) of
the brand will be the same under the two conditions.
• The difference (manipulation) is the ability to interact with the content
(interaction gives control over how they view the product/brand)
Controlling Extraneous Variables
• An extraneous variables is anything that varies in the context of a study
other than the independent and dependent variable; examples include:
– Participant variables (individual differences)
– Situational or task variables
• Extraneous variables affect the dependent variable; making it difficult to
detect the effect of the independent variable on the dependent variable
– First, they increase variability in the data (noise)
– Second, they can become confounding variables
• They vary systematically across all levels of the independent variable
• They provide alternative explanations for the observed change in the dependent
variable
• Extraneous variables affect internal validity
• Need to control for extraneous variables experimentally or
statistically
Randomization
• Randomization involves randomly assigning subjects to different
groups
– Treatment group versus control or comparison group
• Randomizing neutralizes factors other than the independent and
dependent variables
– Reduces bias in data
– Making it possible to infer cause-effect relationships
• Randomization may take the following forms:
– Completely randomized design:
• Every subject is assigned to a treatment group at random
• Each subject has an equal chance of being assigned to aby group
– Randomized block design
• Subjects are first assigned to categories according to the characteristics
they share and then randomly assigned to treatment groups
• Randomization improves generalizability
Measuring Effects in Experiments
• There are mainly three strategies for characterizing
effects
• Different effect
– Used to explain and predict differences in a dependent variable
across different treatment condition
• Better effect
– Used to explain and predict which of two treatment conditions
results in better outcome (dependent variable) according to a
chosen criterion
• Optimal effect
– Used to explains and predict the degree to which specific
treatment condition support an optimum amount of a dependent
variable
Main and Interaction Effects
• Effects may be assessed for each specific independent
variable or for a specific combination of independent
variables
• Main effect
– Main effect refers to changes in the dependent variable that is
accounted for by one main independent variable
• i.e., the effect of one independent variable on the dependent
variable, ignoring the effects of other independent variables
• Interaction effect
– Interaction refers to changes in the dependent variable that is
accounted for by a specific combination of two or more
independent variables
• i.e., the effect of one independent variable on the dependent variable
changes depending on the level of another independent variable
Designing an Experiment
• There are mainly four steps in experimental design
Step 1: Define your research problem
– Define your main variables
– Make predictions about them
– Think about confounding factors and consider how you will
control for them
Step 2: Write down your hypothesis
– Should be specific and testable
Step 3: Decide the expected treatment
– Decide how you will manipulate the independent variable(s)
– Ensure realism of the treatment and the context
Designing an Experiment
• Step 3: Decide the expected treatment
Example
Research Problem: To examine how independent reviewers
assess a proposal by a marketing manager to continue an existing
Customer Relationship Management (CRM) Project in light of a set
of company guidelines on capital budgeting proposal
Two independent variables:
1. Consistency with guidelines
2. Preparer’s incentive
One dependent variable: Independent reviewers’ recommendation
to continue or terminate an existing project
Designing an Experiment
Step 3: Decide the expected treatment
Manipulating the independent variable 1
Designing an Experiment
Step 3: Decide the expected treatment
Manipulating the independent variable 2
• Dependent variable: project continuation recommendation
was measured on a 10-point Likert scale
–
Designing an Experiment
Step 4: Assign subjects to treatment groups
• Randomly assign subjects into groups
• Two types of group designs
– Between group design (independent measure design)
• Each group receives only one treatment
• Each subject experiences only one experiment treatment
– Within group design (repeated measure design)
• Every individual subject is observed repeatedly in different treatment
conditions and the same measure is used for the dependent variable
across all conditions
• The mean differences of the dependent variable across the different
conditions are compared
Designing an Experiment
Preparer's Incentive
Row
Step 4: Assign subjectsPresent
to treatment
Absent groups
Average
Guidelines Consistent 6.95 7.41 7.19 Main Effect of
Example:NotBetween
Consistent group
4.27 design
5.90- 2x25.07
factorial design
Guidelines
Column Average 5.58 6.67 6.13
Main Effect of Incentive
Interaction Effect
Designing an Experiment
Step 5: Manipulation checks
• Used to measure whether the manipulation had intended
effects
– One or two questions administered on the participant to
determine whether and how they perceives the manipulation
– Example: We asked participants the following questions:
Whether the marketing manager was responsible for the initial
project investment decision
Whether the project complied with company guidelines
