You are on page 1of 20

Research Methodology

Experimental and Causal-


Comparative Designs

TOPIC 10
Learning Objectives
At the completion of this topic you should be
able to :
• Outline the importance of control in experimental
designs
• Explain the difference between independent and
dependent variables
• Give examples of experimental research designs
• Evaluate different sampling designs
Introduction
• In the descriptive research we make no attempt to
identify the causes of the phenomenon
• In experimental design we are aiming to identify
the cause-and-effect relationship
• Key steps:
- identify the range of possible factors, which may have
influence
- control all factors except those that we want to investigate
- identify their impact
Progress is relative. We measure progress by noting the amount
of change between what was and what is. And we account for
the change by identifying the dynamics that have caused it.
Ideally, we must manipulate one possible causal factor while
keeping all other possible causal factors constant; only in this
way can we determine whether the manipulated factor has an
effect on the phenomenon we are studying. To the extent that
multiple factors all vary simultaneously, we learn little about
true underlying causes.
Leedy and Ormrod, 2005, pp. 217
• Examples for experimental design – in medical
research, the impact of educational methods, or advertising
message
- Identify two groups of people that can be
considered identical by all intends and purposes
- Measure certain characteristic by certain tests
- Apply treatment
- Measure the characteristic again by using a post-
test
- Conclude whether the treatment has caused any
changes
• The difference between experiment and
experimental design is in the objective
• To determine a fact or to identify a cause-and-
effect relationship
• The consensual participation of the group
members is required
• All ethical questions must be clarified before
commencing the study
The Importance of Control
• Experimental designs have to deal with the
influence of confounding variables, which
confuse the relationship we seek to investigate
• Key question - the internal validity
• Internal validity refers to the extent to which the
research design and the data allow accurate
conclusions to be derived about the cause-and-
effect and other relationships
• Example
• In order to maximise the internal validity we have
to control the confounding variables, so that their
impact is ruled out
• Six most common strategies –
• Keep some things constant – factors that are the same
for everyone (age, sex, educational level, socioeconomic
status, etc)
• Include a control group – observe in parallel a group of
participants that are not exposed to treatment
- placebo effect and ethical considerations
• Randomly assign people to groups – there are no
identical samples, but by random assignment be can ensure
that any difference are due entirely to chance
• Assess equivalence before treatment with one or
more tests – if the groups are similar it reduces the
possibility that such variables account for the measured
differences
• Expose the participants to all experimental
conditions – by varying the order of treatment we can
eliminate the impact of different research designs
• Statistical control of confounding variables – need
to have very large samples and even then these methods do not
provide a substitute for research design controls
Independent and Dependent Variables
• Key question in experimental design is the
decision on dependent and independent variables
• Variable – is any quality or characteristic in a
research investigation that has two or more
possible values
• Independent variables/causes– those variables
that are manipulated by the researcher
• This is usually what we measure/observe, e.g.
educational method, amount of medicine, the health and safety
equipment, etc.
• Dependent variables – those variables that are
influenced by changes of other variables
• They are influenced by/depend on the independent
variables
• In the above examples
Independent Dependent

educational method, educational result


amount of medicine, health condition
health & safety equipment, health & safety record
Examples of experimental research design

• There are various research design techniques


based on the extent to which the researcher
manipulates the independent variables and
controls for the confounding variables
• Pre-experimental designs – for tentative hypothesis
formulation
• True experimental designs – high degree of internal
validity, random assignment of group members
• Quasi-experimental designs – in cases when
randomness is not possible, therefore alternative explanation
are possible
Conducting experiments on the Internet

• Internet as a source of new opportunities for


research
• Most of the studies are descriptive or survey
studies involving the completion of questionnaires
• More rarely there are some experimental studies
• Very suitable for the business and risk
management research (e.g. risk/safety perceptions,
consumers attitudes, market niche products, etc.)
• Example for experimental research
• Psychological research on patters of learning
where the participants were randomly assigned to
an experimental group and a control group
- respondents are asked to read and study a story that
is illustrated with photographs
- read three additional stories, one of which is similar
to the first one
- answer a series of questions about details in the
stories
• Drawbacks
- the sample may not be representative of the
population
- accuracy of responses
- time constraints
• Suitable for initial stages of the research, for tentative
hypothesis formation or for pilot tests
Testing hypotheses and beyond
• The research hypothesis reflects the interest of the
researcher in certain phenomenon
• It is an educated guess about how to resolve the
problem situation
• Practical purpose – acts as a temporary objective,
operational target and logical framework which
guides the researcher through the data collection
and analysis
• The statistical hypothesis - involves the
comparing of the distribution of data collected by
the researcher with an ideal, or hypothetical
distribution
• Then we investigate the variability of the actual
data and make judgements about the extend of the
variability
• Conceptual shift
• In the past experimental designs were used
predominantly in conjunction with hypothesis
testing and statistical methods
• At present – descriptive quantitative analysis has
been increasingly used for evaluation of
experimental work
• It is conceivable to produce relevant research
findings, which do not require an explicit
statement of a hypothesis
Task for the next week
• Read the article “Business social
responsibility: how are SMEs doing in
Gauteng, South Africa?” by Ladzani and
Seeletse
• Think about the strengths and the weaknesses
of this paper

You might also like