3/23/2020

EXPERIMENTAL RESEARCH

 A type of research aimed at establishing

the possible cause and effect
relationship between variables under
study through some systematic and well
DR. AMITA BAJPAI
PROFESSOR
planned observations carried out in
DEPARTMENT OF EDUCATION controlled conditions.
UNIVERSITY OF LUCKNOW

PRINCIPLES OF EXPERIMENTAL
Conducting an experimental research DESIGNS- R. A. FISHER
 select the problem that is answerable with
available tool  PRINCIPLE OF REPLICATION-each treatment is
 Formulate a hypothesis as a tentative solution applied in many experimental units/experiment
to the problem should be repeated more than once for accurate
results.(replication itself is a source of variation)
 select participants and instrument
 selection and execution of a research
 PRINCIPLE OF RANDOMIZATION-provide
design/plan protection against the effect of extraneous factors by
 collection and organization of data randomization(chance error)
 Conclusions and generalization of hypothesis
if it is confirmed

EARLY EXPERIMENTATION
 PRINCIPLE OF LOCAL CONTROL-we can  LAW OF SINGLE VARIABLE-(Jon Stuart Mill-
eliminate the variability due to extraneous factors 1873)-
from the experimental error.(under it the If two situations are alike in every respect,
extraneous factor, the known source of and one element is added to one but not the
variability is made to vary deliberately. This other, any difference that develops is the
needs to be done in such a way that variability effect of the added element, or if one element
it causes can be measured and hence can be is removed from one but not from the other,
eliminated from the experimental error.) any difference that develops may be attributed
to that element.

1
 3/23/2020

VARIABLES DEPENDENT VARIABLE

 a concept (e.g., intelligence, height, aptitude) that  the change or difference occurring as a
can assume any one of a range of values
 INDEPENDENT- an activity of characteristic
result of the independent variable
believed to make a difference with respect to (syn.) criterion variable, assigned
some behavior
variable, effect, outcome, posttest
(syn.) experimental variable, active variable,
cause, treatment eg.- academic achievement, language
 Treatment variable-manipulated eg. Duration of development, social skills, emotions,
treatment adjustment etc.
 Attribute variable-can not be altered eg. Age, sex,
race etc.

CONTD…… CONTRLLING EXTRANEOUS VARIABLES

 CONFOUNDING VARIABLE- the fact
that the effects of the independent variable may
intertwine with extraneous variables, such that it is  Removing the variable
difficult to determine the unique effects of each  Matching cases
variable
 …(syn.) criterion variable, assigned variable, effect,  Balancing cases
outcome, posttest
 INTERVENING-anxiety, fatigue, motivation etc. can be controlled
 Randomization
through appropriate design  Analysis of covariance
 EXTRANEOUS- SES, age, competency etc. can be controlled
through proper selection of sample.

EXPERIMENTAL VALIDITY
THREATS TO INTERNAL VALIDITY
 MATURATION
 INTERNAL VALIDITY- to the extent that
 HISTORY
the factors that have been manipulated,  TESTING
actually have a genuine effect on the  UNSTABLE INSTRUMENTATION
observed consequences.  STATISTICAL REGRESSION (regression to the
mean)
 EXTERNAL VALIDITY-to the extent to  SELECTION BIAS

which the variable relationship can be  EXPERIMENTAL MORTALITY

generalized to other settings.  EXPERIMENTER BIAS (blind/double blind)

2
 3/23/2020

THREATS TO EXTERNAL VALIDITY Types of reactive arrangements…

 INTERFERENCE OF PRIOR TREATMENT …Hawthorne
Hawthorne effect (1933):
(1933) any situation in
 ARTIFICIALITY OF EXPERIMENTAL SETTING
which participants’ behavior is affected
not by the treatment per se but by their
 INTERACTION EFFECT OF TESTING
knowledge of participating in a study
 INTERACTION OF SELECTION AND TREATMENT(selected
sample rarely random)
…compensatory
compensatory rivalry:
rivalry the control group
 THE EXTENT OF TREATMENT VERIFICATION is informed that they will be the control
( Someone else delivers the treatment)
group for a new, experimental study
(“John
John Henry effect”)
effect

…placebo
placebo effect:
effect the situation in which half
Questions or Comments
of the participants receive no treatment

?
but believe they are

…novelty
novelty effect:
effect the situation in which
participant interest, motivation, or
engagement increases simply because
they are doing something different

16

EXPERIMENTAL DESIGNS

 PRE-EXPERIMENTAL-least effective, it provides

either no control group or no way to equate the
groups

THANK YOU  TRUE EXPERIMENTAL- employs randomization

to provide for control of the equivalence of
groups and exposure to treatment
 QUASI EXPERIMENTAL- provides a less
satisfactory degree of control. Used only when
randomization is not feasible.

3
 3/23/2020

ANOTHER CLASSIFICATION CLASSIFICATION CONTD……

 INFORMAL DESIGNS- (less control) Single Factor Designs

before and after without control Pretest-posttest (one-group)
after only with control Pretest-posttest (control group)
before and after with control Posttest-only (control group)
 FORMAL DESIGNS- (more control) Multiple Factor Designs
Latin square design Two-way factorial
factorial design e.g., 2 x 3
randomized block design Three-way factorial
e.g., 2 x 2 x 3

PRE EXPERIMENTAL/INFORMAL PRE-EXPERIMENTAL(CONTD.)

one-shot case study
one-
static group comparison
X O
X1 O
a single group exposed to a treatment ( X ) and
X2 O
then post tested ( O )
 involves at least two groups ( X ), one
one--group pretest-
one pretest-posttest design
receiving a new, or experimental treatment (
O X O X1 ) and another receiving a traditional, or
a single group is pretested ( O ), exposed to a control treatment ( X2 ) and, then, are post
treatment ( X ) and, then, is posttested ( O ) tested ( O )

TRUE EXPERIMENTAL DESIGNS/FORMAL

 Post test only, equivalent group design-  The Solomon four group design-
R X O1
R C O2 Ra O1 X O2
Treatment effect = O1 - O2 Rb O3 C O4
Pre test- post test equivalent group design- Rc X O5
Rd C O6
R O1 X O2
Combination of above two designs
R O3 C O4 Comparison of (Ra and Rc) and (Rb and Rd)
Treatment effect = O2-O1 = A Possible to evaluate the effect of testing, history and
O4-O3 = B maturation etc.
TE = A - B Difficulty in finding enough subjects to assign randomly
to four equivalent groups

4
 3/23/2020

QUASI-EXPERIMENTAL DESIGNS

 Pre test post test non equivalent groups design-

counterbalanced design
O1 X O 2 A = O 2 - O1
O3 C O4 B = O 4 – O3 X1 O X2 O X3 O
Treatment effect = B – A X3 O X1 O X2 O
random assignment of intact groups that are X2 O X3 O X1 O
pretested ( O ), exposed to a treatment ( X ) and all of the groups receive all treatments but
then posttested ( O )
in a different order; the number of groups
Groups can be two classes or two slums etc. and treatments must be equal

TIME SERIES DESIGN Single Subject Design

OOOOO X OOOOO  Time series analysis

 a single group is pretested ( O ) repeatedly  Dependent measure
is continuous
until pretest scores are stable, exposed to
 Establish baseline
a treatment ( X ) and, then, is repeatedly
 Measure treatment
post tested ( O ) effect over time
 Control group time series
OOOOO X OOOOO Baseline Treatment

OOOOO C OOOOO
Time

Two-Way Factorial Design

A 2 X 2 factorial design…
 Studies multiple Example: 2 x 3
independent Independent Variable
variables-eg.sex
and teaching ME2 A
manipulated
B
not manipulated
method
L1 L2 L3
 Main effects (ME)
 Each with a number of O Group #1 Group #2
L1
levels (L)
Dependent Variable

ME1
 Permits study of L2
interactions
O Group #3 Group #4
 Analysis
Cells
 ANOVA

5
 3/23/2020

A 2 X 2 factorial design… A 2 X 2 factorial design…

A
A

No interaction Interacting factors

between factors
B

Questions or Comments

? THANK YOU

33