Experiments
(A) These follow the classical deductivist research
approaches involving…
•
i) Careful outlining of the problem to be
investigated via the formulation of a research hypothesis
•
ii) Careful identification of the factor whose
variation is being studied. This is the ‘theoretically dependent variable’
•
iii) Identification of factors or situations
whose variation affects, cause or explain the theoretically-dependent variable
in the manner suggested by the hypothesis. This step involves the
identification of the ‘theoretically independent variables’
•
iv) Operationalisation of the variables
•
v) Accounting for effects of ‘extraneous
variables’ on variations in the dependent variable by neutralising or eliminating such extraneous variables.
•
Use of a control experiment may be useful here.
Failure to control for extraneous variables may affect the Internal Validity
of any findings from the experiment
•
(B) Making causal links between observed changes
in the dependent variables as a result of changing the independent variables
•
(C) Note: Experiments are rarely used in
business or management research, Psychologists, however, sometimes use
experiments as a means of social research.
•
The problem is in managing to produce control
groups and eliminate all extraneous variables
Some specific problems in experimental research
•
A) Indexicality
•
B) Experimenter effects
•
C) Subjects’ mediation through interpretation
Note: The ‘Hawthorne Effect may be explained by all 3 of
these phenomena
A. Indexicality
•
The context of research has effects on subjects’
behaviour. E.g. behaviolur of a subject changes from home to classroom to
laboratory contexts
B. Experimenter effects
•
This is similar to indexicality. Here, subjects’
behaviour is affected/influenced by their perception of the experimenter(s).
•
Changes in experimenter(s) may elicit
corresponding changes in research subjects (over and above the effect of
changes introduced in the experiment itself)
Experimenter effects may result from
i)
The problem of the introduction of unintended
biases into the subject or sample group(s)
ii)
Changes in the measurement processes during the
course of the experiment
iii)
Biases due to the subjects’ reaction to the
actual context of the experiment and the processes used
iv)
Changes in subjects’ social, political,
economic, physiological states etc. during the course of the experiment
C. Subjects’ mediation through interpretation
•
Different research subjects interpret the same
environmental and/or research-induced cues differently – that is, research
subjects’ mode of meaning-making is
unique to them. Their responses to changes in cues would also be individually
differentiated. Indeed, a research subject’s ‘reading’ of an experimental
phenomenon may be closer to reality than that of the experimenter
Ecological validity
Due to:
i)
All above, and
ii)
The fact that the research context is unlikely
to be correctly extrapolated to normal situations….
iii)
This would make social experiments lack ‘ecological
validity’
Quasi – Experiments
i)
Like experiments, quasi experiments are
deductive in approach (i.e. hypothesis
–method- observation- analysis- conclusion)
ii)
They differ from pure experiments in that
research subjects are observed in their natural settings (as opposed to
laboratory settings)
iii)
The control group is also observed in its
natural settings
iv)
The group is selected to be as similar as
possible to the ‘experimental group’
v)
Changes (of the independent variable) in the
experimental group are not mirrored in the control group
Observed differences in the dependent variable between the
experimental and control groups are attributed to the changes in the
independent variable in the experimental group.
vii As far as possible, extraneous variables are eliminated.
However, causal relationship between dependent and independent variables cannot
be concluded to be totally unaffected by other extraneous variables, be they
known or unknown
Viii An example: A local authority erects CCTV in one area
of town and observes the effects of this on levels of crime over a period of
time. Causal links are made between the erection of CCTV cameras and the drop
in crime rates by comparison with a nearby area with similar demographics but
no CCTVs erected
ix Much as inferences in (viii) may be valid. However, other
explanations may be given for the observed differences – e.g. the moving away
from the CCTV area of key gang members, quite unconnected to the CCTV erection
or the coincidental rise in interest in computer games of crime age group
members, etc
Action Research
•
In this method of research, the researcher’s
role is to diagnose a problem in an organisation’s or department’s activities and procedures
•
Next a possible solution for the problem is
decided on
•
The solution is then implemented (intervention)
and the outcome of this carefully monitored and recorded
•
As a result, further adjustments are made and
the cycle repeated until a sustained outcome is arrived at
•
The research is usually via the
hypothetico-deductive approach
•
Typically, action research is engaged in by
consultants
•
Aim is to both add to existing knowledge and
help resolve actual practical organisational problems
Action Research – Some issues
i)
Difficulty in finding a control group or
situation
ii)
Difficulty in assigning changes in
organisational issues to researcher interventions
iii)
Effect of context may introduce extraneous
variables
iv)
Researcher’s own personal influence cannot be
ignored
v)
Making changes may remove the very naturalism
that the research is intended to study
vi)
Time constraints may affect findings (- hence
usually not ethnographic?)
No comments:
Post a Comment