Presented by
Dr.J.P.Verma
MSc (Statistics), PhD, MA(Psychology), Masters(Computer Application)
Professor(Statistics)
Lakshmibai National Institute of Physical Education, Gwalior, India
(Deemed University)
Email: vermajprakash@gmail.com
To test a theory through deductive logic
To develop a theory through inductive logic
2
The exercise intensity with 65% maximum heart rate may
improve the cardio- respiratory endurance significantly.
By means of Hypothesis
Example
3
If most of the sports persons are medal winners from a
particular university their training programme
may be superior than the other universities.
By observing a phenomenon
Example
4
5
This Presentation is based on
Chapter 1 of the book
Repeated Measures Design
for Empirical Researchers
Published by Wiley, USA
Complete Presentation can be accessed on
Companion Website
of the Book
To investigate some kinds of relationship between
independent and dependent variables.
6
 lack cause and effect relationships
 Reduced internal validity
Non Experimental or Correlational
Experimental
 Explain cause and effect relationships
 Higher internal validity
7
Experimenter manipulates independent variable
to see its impact on dependent variable
by controlling extraneous factors
8
Organizing a controlled experiment to generate data
for understanding the causes of variation
9
- Ensures homogeneity in the experimental groups
- Enhances external and internal validity in the study
Randomization
Replication
- Repeating an experiment a number of times on subjects/
experimental units
- A way of reducing experimental error by including an extraneous variable
in the experiment.
- Heterogeneous experimental units are divided into homogenous blocks
- Treatments are randomly allocated in these blocks.
Blocking
10
 Used when experimental material/subjects are homogeneous
 Effect of one factor on dependent variable is investigated
Classification of Statistical Designs in Research
A. Completely Randomized Design(CRD)
5
8
12
7
3
9 10
1 2
11
6
4
6 1 11Stage 1
2 9 4Stage 2
5 12 3Stage 3
7 8 10Stage 4
T2 T1 T3
T1 T3 T2
T2 T3 T1
T3 T1 T2
Sample
Comparing effect of
three advertisements
T1,T2 andT3 on sale of
a product
Example
Figure 1.1 Layout of the completely randomized design
Fig.1.1 Layout of completely randomized design 11
 Used when experimental material/subjects are heterogeneous
 Effect of one factor on dependent variable is investigated by
introducing the blocking variable in experiment.
B. Randomized Block Design
T1 T3 T2Low IQ
Average IQ
High IQ
Block:IQ
T3 T1 T2
T2 T3 T1
Allocation of
treatments in block
Block 1
Block 2
Block 3
Subjects in block
Fig.1.2 Layout of randomized block design
To study the effect of three
different types of teaching
methodologies T1,T2 and
T3 on learning efficiency.
Example
12
 Special case of randomized block design
 Subjects are matched on some characteristics which are supposed
to affect the experimental variable
 Here each matched pair is like a block
 Only comparison of two treatments is possible
B(i). Matched Pairs Design
S1
S4
S5
S100
Subjects in each pair
Pair 1
Pair 2
Pair 3
Treatment
S2
S3
S6
S99
Exercise Placebo
. . .
. . .
Pair 50
Figure 1.3 Layout of matched pairs design
To study the effect of
exercise on strength in
100 students
Example
13
C. Latin Square Design
 In this design random variation of two factors is controlled
 Two blocking factor can be taken in this design
 Number of rows columns and treatments are required to be same in this design.
 Each treatment can occur only once in the corresponding row and column.
Low
Average
High
Block:IQ
Fig.1.2 Layout of Latin square design
To study the effect of three
different types of teaching
methodologies T1,T2 and
T3 on learning efficiency.
Example
14
T1 T2 T3
T3 T1 T2
T2 T3 T1
Block: Age
Teens
Mid
age
Old
age
If factors A(exercise intensity)
has three levels(low , medium
and high) and B(Environment)
also has three levels(hot,
humid and cold) then nine
treatment groups are
required.
To investigate the effect of two or more factors
on a dependent variable simultaneously
Example
Low
Medium
High
Hot Humid Cold
Cells
Subjects in
each cell
FactorA:MentalExercise
Factor B: Environment
Figure 1.4 Layout of 3×3 factorial experiment
in CRD
15
DependentVariable: Task efficiency
- Experimental unit on which experiment is conducted
Subject
Treatment
- Levels of the independent variable whose effect is to be seen on the
dependent variable.
- A variable of interest
- An independent variable whose effect is to be seen on the
dependent variable
CriterionVariable
Factor
16
To see the effect of
Aerobic exercise
with different
intensity
on the Cardio
respiratory
endurance
in
Housewives
SubjectsTreatments: Intensities of aerobic exercise
Factor: Aerobic exercise
Criterion
variable
17
- Spread of Scores
Variation
- Measure ofVariation
Variance
18
11
4
3
1
16
3
14
13
6
6
3
21
4
8
17
3
2
13
How to measure variation?
Can be estimated by
Range Variance Q.D. Mean Dev.
19
 ī€ŠīƒĨ ī€Ŋ 22
x
N
1
 ī€ŠīƒĨ 

ī€Ŋ 22
xx
1n
1
S
Population variance =
Mean Square Deviation
11
4
3
1
16
3
14
13
66
3
21
4
8
17
3
2
13
Population
Sample
11
2
14
3
Whether the population variance
can be estimated correctly by the
sample variance ?
S2 is an unbiased estimate of population variance
20
 Uncontrolled error in an experiment
 Attributed to non-assignable causes
 Individual variation
Experimental Error
21
 Extent of generalizibility of findings to the population
from which sample has been drawn.
External validity
 Extent to which one can say that the variation observed in
the Dependent variable(DV) is due to the variation in the
Independent variable(IV).
InternalValidity
22
Variation among scores
 ī€ŠīƒĨ 

ī€Ŋ 22
xx
1n
1
SMean Square Deviation =
df
Variation
ī€Ŋ
df
SS
ī€Ŋ
23
Purpose of Experimental Design
īąMaximize SystematicVariance
īąControl ExtraneousVariance
īąMinimize ErrorVariance
24
Effect of 2 weeksTeaching methodology on performance
Traditional Method
T1
Flexible method
T3
Audio-visual Method
T2
6
7
5
2
9
8
7
7
5
4
3
2
Systematic variance
High IQLow IQ
Extraneous
variance: IQ
Error variance
Mixed IQ
25
26
To buy the book
Repeated Measures Design
for Empirical Researchers
and all associated presentations
Click Here
Complete presentation is available on
companion website of the book

Foundations of Experimental Design

  • 1.
    Presented by Dr.J.P.Verma MSc (Statistics),PhD, MA(Psychology), Masters(Computer Application) Professor(Statistics) Lakshmibai National Institute of Physical Education, Gwalior, India (Deemed University) Email: vermajprakash@gmail.com
  • 2.
    To test atheory through deductive logic To develop a theory through inductive logic 2
  • 3.
    The exercise intensitywith 65% maximum heart rate may improve the cardio- respiratory endurance significantly. By means of Hypothesis Example 3
  • 4.
    If most ofthe sports persons are medal winners from a particular university their training programme may be superior than the other universities. By observing a phenomenon Example 4
  • 5.
    5 This Presentation isbased on Chapter 1 of the book Repeated Measures Design for Empirical Researchers Published by Wiley, USA Complete Presentation can be accessed on Companion Website of the Book
  • 6.
    To investigate somekinds of relationship between independent and dependent variables. 6
  • 7.
     lack causeand effect relationships  Reduced internal validity Non Experimental or Correlational Experimental  Explain cause and effect relationships  Higher internal validity 7
  • 8.
    Experimenter manipulates independentvariable to see its impact on dependent variable by controlling extraneous factors 8
  • 9.
    Organizing a controlledexperiment to generate data for understanding the causes of variation 9
  • 10.
    - Ensures homogeneityin the experimental groups - Enhances external and internal validity in the study Randomization Replication - Repeating an experiment a number of times on subjects/ experimental units - A way of reducing experimental error by including an extraneous variable in the experiment. - Heterogeneous experimental units are divided into homogenous blocks - Treatments are randomly allocated in these blocks. Blocking 10
  • 11.
     Used whenexperimental material/subjects are homogeneous  Effect of one factor on dependent variable is investigated Classification of Statistical Designs in Research A. Completely Randomized Design(CRD) 5 8 12 7 3 9 10 1 2 11 6 4 6 1 11Stage 1 2 9 4Stage 2 5 12 3Stage 3 7 8 10Stage 4 T2 T1 T3 T1 T3 T2 T2 T3 T1 T3 T1 T2 Sample Comparing effect of three advertisements T1,T2 andT3 on sale of a product Example Figure 1.1 Layout of the completely randomized design Fig.1.1 Layout of completely randomized design 11
  • 12.
     Used whenexperimental material/subjects are heterogeneous  Effect of one factor on dependent variable is investigated by introducing the blocking variable in experiment. B. Randomized Block Design T1 T3 T2Low IQ Average IQ High IQ Block:IQ T3 T1 T2 T2 T3 T1 Allocation of treatments in block Block 1 Block 2 Block 3 Subjects in block Fig.1.2 Layout of randomized block design To study the effect of three different types of teaching methodologies T1,T2 and T3 on learning efficiency. Example 12
  • 13.
     Special caseof randomized block design  Subjects are matched on some characteristics which are supposed to affect the experimental variable  Here each matched pair is like a block  Only comparison of two treatments is possible B(i). Matched Pairs Design S1 S4 S5 S100 Subjects in each pair Pair 1 Pair 2 Pair 3 Treatment S2 S3 S6 S99 Exercise Placebo . . . . . . Pair 50 Figure 1.3 Layout of matched pairs design To study the effect of exercise on strength in 100 students Example 13
  • 14.
    C. Latin SquareDesign  In this design random variation of two factors is controlled  Two blocking factor can be taken in this design  Number of rows columns and treatments are required to be same in this design.  Each treatment can occur only once in the corresponding row and column. Low Average High Block:IQ Fig.1.2 Layout of Latin square design To study the effect of three different types of teaching methodologies T1,T2 and T3 on learning efficiency. Example 14 T1 T2 T3 T3 T1 T2 T2 T3 T1 Block: Age Teens Mid age Old age
  • 15.
    If factors A(exerciseintensity) has three levels(low , medium and high) and B(Environment) also has three levels(hot, humid and cold) then nine treatment groups are required. To investigate the effect of two or more factors on a dependent variable simultaneously Example Low Medium High Hot Humid Cold Cells Subjects in each cell FactorA:MentalExercise Factor B: Environment Figure 1.4 Layout of 3×3 factorial experiment in CRD 15 DependentVariable: Task efficiency
  • 16.
    - Experimental uniton which experiment is conducted Subject Treatment - Levels of the independent variable whose effect is to be seen on the dependent variable. - A variable of interest - An independent variable whose effect is to be seen on the dependent variable CriterionVariable Factor 16
  • 17.
    To see theeffect of Aerobic exercise with different intensity on the Cardio respiratory endurance in Housewives SubjectsTreatments: Intensities of aerobic exercise Factor: Aerobic exercise Criterion variable 17
  • 18.
    - Spread ofScores Variation - Measure ofVariation Variance 18
  • 19.
    11 4 3 1 16 3 14 13 6 6 3 21 4 8 17 3 2 13 How to measurevariation? Can be estimated by Range Variance Q.D. Mean Dev. 19
  • 20.
     ī€ŠīƒĨ ī€Ŋ22 x N 1  ī€ŠīƒĨ   ī€Ŋ 22 xx 1n 1 S Population variance = Mean Square Deviation 11 4 3 1 16 3 14 13 66 3 21 4 8 17 3 2 13 Population Sample 11 2 14 3 Whether the population variance can be estimated correctly by the sample variance ? S2 is an unbiased estimate of population variance 20
  • 21.
     Uncontrolled errorin an experiment  Attributed to non-assignable causes  Individual variation Experimental Error 21
  • 22.
     Extent ofgeneralizibility of findings to the population from which sample has been drawn. External validity  Extent to which one can say that the variation observed in the Dependent variable(DV) is due to the variation in the Independent variable(IV). InternalValidity 22
  • 23.
    Variation among scores ī€¨ī€ŠīƒĨ   ī€Ŋ 22 xx 1n 1 SMean Square Deviation = df Variation ī€Ŋ df SS ī€Ŋ 23
  • 24.
    Purpose of ExperimentalDesign īąMaximize SystematicVariance īąControl ExtraneousVariance īąMinimize ErrorVariance 24
  • 25.
    Effect of 2weeksTeaching methodology on performance Traditional Method T1 Flexible method T3 Audio-visual Method T2 6 7 5 2 9 8 7 7 5 4 3 2 Systematic variance High IQLow IQ Extraneous variance: IQ Error variance Mixed IQ 25
  • 26.
    26 To buy thebook Repeated Measures Design for Empirical Researchers and all associated presentations Click Here Complete presentation is available on companion website of the book