2. Experimental Designs
Experimental design is the blue print of the procedures
that enabled the researcher to test hypothesis by relating
independent and dependant variables.
• A researcher can select their research design based on
– Purpose of the study
– Variables involved and their nature
– Controls required.
– Pure experimental studies are rare in social sciences. Often
we conduct crude and quasi-experiments in education.
3. PRE EXPERIMENTAL DESIGNS
• One shot case studies
• Simple one group design
• Static Group Comparison design
4. ONE SHOT CASE STUDIES
• One group is involved and exposed to
treatment.
• Posttest is given to measure the effect of
treatment on the dependant variable.
• No control group is involved.
5. SIMPLE ONE GROUP DESIGN
– involves a single group to which is applied an experimental
factor.
– Changes in outcome are determined by measuring the
dependent variable before and after the experimental
treatment. (I.e. a pretest and post test is given)
Limitations
– Carry over effect affects the result of the study.
– Maturation of children during the course affects the result of the study.
– Novelty of the treatment, researcher & participant’s enthusiasm affect
the experimental group.
6. • Two groups are selected and treatment is given
to one group and the other is not exposed to
treatment.
• No pretest is given.
• The difference between the posttest scores of
both the groups reveal the effect of treatment.
Limitations
– Groups are not equated so the effect cannot be purely
due to the independent variable.
STATIC GROUP COMPARISON DESIGN
7. TRUE EXPERIMENTAL DESIGNS –
These designs ensure internal validity and external
validity and highest control since the subjects are
randomly assigned.
• Posttest only control group design
• Pretest post test control group design
• Solomon four group design
8. POSTTEST ONLY CONTROL GROUP DESIGN
• Two groups are chosen randomly. One as
experimental and the other as the control group.
• Random assignment of subjects to the experimental
and control group.
• Experimental group is given treatment and the
control group does not receive treatment.
• Post test for the two groups are conducted and the
difference tested to measure the degree of change.
9. PRETEST- POSTTEST CONTROL GROUP DESIGN
• Two groups experimental and control group are
randomly selected.
• Pre test is conducted for both the groups to measure
the degree of dependant variable before treatment.
• Treatment is applied to the experimental group.
• Post test is conducted for both the groups.
• Difference in the gain scores are computed to find
the effect of the independent variable.
• Gain score = posttest - pretest
10. – Subjects are randomly selected and placed in four groups.
– Two groups are experimental groups one with pretest and
other without pretest.
– Two groups are control groups one with the pretest and the
other without pretest.
Experimental group I - T1 ----Treatment---- T2
Control group I - T1
1 -----------T21
Experimental group II - _ -------Treatment------T2
Control group II - _ -----------T21
– Difference between the posttest score of experimental and
control groups can be compared and tested for significance.
– It gives best results but since it involves large samples it is
time and energy consuming.
SOLOMON FOUR GROUP DESIGN
12. – Similar to pretest - posttest control group design.
– Here no randomization of groups is involved.
– The groups are chosen as clustered units
(Section A & Section B).
– Variables related to history, maturation or
testing may interfere with the treatment.
NON-EQUIVALENT CONTROL GROUP DESIGN
13. –This is an elaborated version of one group pretest – posttest
design.
–The subjects are repeatedly pre tested and post tested on the
dependent variable at periodic intervals both before and after
treatment.
–The purpose of the series of measurements is to show that there
is significant increase only immediately after the intervention
and not earlier or later.
Pretest
T1
Pretest
T2
Pretest
T3
Posttest
T1
Posttest
T2
Posttest
T3
Treatment
SINGLE GROUP TIME SERIES DESIGN
14. –The groups are selected as section A and B and we have
two treatments X and Y.
–Group A and B are pre tested and taught the same unit by
method X and Y respectively.
–The same procedure is adopted with the methods
interchanged for the next unit.
–Rotation group method rotates out some of the non
experimental factors that affect the dependent variable
(such as intelligence, maturation, difficulty of units, carry
over effect)
COUNTER BALANCED DESIGN OR ROTATION GROUP DESIGN
15. Modern Research Designs
Researchers have developed sophisticated
designs such as Latin squares Greek squares,
randomized blocks, Factorial designs based on
statistical analysis.
16. FACTORIAL DESIGNS
• When more than one independent variable is included we adopt factorial design.
• In a simple 2 x 2 factorial design, two independent variables say Gender and
Method of teaching) and a dependent variable may be involved.
• In such a design there will be four cells representing four sub groups such as
• Boys under method I
• Boys under method II
• Girls under method I
• Girls under method II
• For each cell, the students will be randomly assigned and the treatments given.
• Using analysis of variance the influence of the independent variables on the
dependent variable is measured.
17. IMPORTANCE OF EXPERIMENTAL METHOD
• Most satisfactory means for obtaining reliable
knowledge.
• Cause and effect relationship are analyzed with
greater speed and certainty.
• Serves as the basis for scientific generalization and
prediction.
18. LIMITATIONS OF EXPERIMENTAL METHOD
• Control of factors affecting dependent variable is
not quite possible in social sciences.
• Identifying and measuring the factors affecting
dependent variable is a tedious task.
• Therefore results of experimental method in
social science is not that much reliable as in pure
sciences
19. Parallel or equivalent group design:
– Two or more groups of subjects, equivalent in all respects in relation to
the study have to be selected. One of these parallel groups serves as the
control group while the other as the experimental group.
– If the differences obtained between pre test and posttest (gain score) is
more for the experimental group than the control group, then excess gain
is due to the introduction of experimental factor.
Equating of groups
• Co-twin technique
• Matched pair technique: significant characteristics
• Matched group technique: same mean score and standard deviation
• Randomized Group Technique: random sampling technique