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. Also known as repeated MANOVA or rMANOVA
Also known as
To investigate the effect of an independent factor (having different levels)
on a group of dependent variables
Why to Use
3. Same subjects are tested under each level of the independent
variable.
Independent variable can either be different treatment conditions
or different time points.
Features
DVs : Number of correct recalling of name, colour and shape of objects
IV : Four and Six seconds visual time
Example Which visual time is more effective for memory retention of the
object’s characteristics?
4. When individuals vary widely on the experimental variable.
When several dependent variables (DVs) measure different
aspects of some cohesive theme.
Where improvement trend needs to be investigated.
Example of DVs
personality (Extraversion, Psychoticism, Neuroticism)
health(blood pressure, heart rate, vital capacity)
product features(economy, comfort, attractiveness)
fitness(cardio respiratory endurance, flexibility,
strength)
nature(extrovert, optimism, creativity)
academic achievement(English, Maths, Commerce)
Example: To study the change in physiological status((heart rate, blood pressure and vital
capacity) of subjects while undergoing an exercise programme over a period of time.
5. 5
This Presentation is based on
Chapter 7 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. Choose DVs carefully in the study
DVs should be moderately correlation (.3 to .7)among themselves
Highly correlated DVs Weaken the power of the analysis
Uncorrelated DVs MANOVA has nothing to offer
Word of Caution
Thumb Rule
Even if
dependent
variables are
moderately
Don’t be
tempted to use
RM MANOVA
If combining
DVs can not
be justified
Consider using
series of univariate
ANOVAs
7. 1. Due to demand of research question being investigated.
2. Variables explaining latent variable are often correlated
hence separate rANOVA's will be redundant and difficult to
integrate.
3. None of the individual ANOVAs may produce a significant
effect on the DV, but if combined they might.
4. By using MANOVA, family wise error rate(α) can be
controlled.
5. The sphericity assumption in rANOVA is often violated
whereas RM MANOVA does not require this assumption.
8. To investigate as to how the personality(Extraversion,
Psychoticism and Neuroticism) transformation takes place
during one year of training in communication skill.
To investigate as to which naturopathy intervention
(pranayama, meditation and relaxation exercise) is more
effective in improving mood state(confusion, depression
and fatigue)
An educational consultant may wish to investigate
performance(numerical aptitude, reasoning and English
comprehension) trend of subjects during a training
programme for a competitive examination.
9. Data type : There should be two or more continuous DVs and one
categorical IV.
Sample Size Number of observations must be higher than the number
of DVs. Recommended sample size of at least 20.
Independence of Measurement
Missing Data This design requires complete data for all the subjects.
Outliers No outlier should exist in any group
Linearity All DVs are linearly related among themselves in each group of
the independent variable.
Normality There should be multivariate normality.
Multicollinearity There should be no multicollinearity among the DVs.
SphericityThere should be no sphericity in data.
10. Case I: Levels of the within-subjects variable are different treatment conditions
Example: To investigate the effect of naturopathy intervention in improving mood state of six
subjects
When to use One-way rMANOVA
Each subject is tested on multiple dependent variables in each treatment
condition
Issues in the Design
Carryover effect – Controlled by having sufficient gap between any two treatments
Order effect – Controlled by counterbalancing
IV : Naturopathy intervention (pranayama, meditation and relaxation exercise)
DVs : Mood state parameters(confusion, depression and fatigue)
11. S2
S5
S1
S6
S3
S4
Relaxation Exercise
First phase
testing
S2
S5
S1
S6
S3
S4
S2
S5
S1
S6
S3
S4
Second phase
testing
Third phase
testing
Testing protocol
Treatment: Naturopathy intervention
Confusion Depression Fatigue
S1
S6
S3
S4
S2
S5
S1
S6
S3
S4
S2
S5
S1
S6
S3
S4
S2
S5
Confusion Depression Fatigue
S3
S4
S2
S5
S1
S6
S3
S4
S2
S5
S1
S6
S3
S4
S2
S5
S1
S6
Confusion Depression Fatigue
MeditationPranayama
Figure 7.1 Layout design
1. Divide sample into groups
2. Randomized treatments on these groups and take measurements on all dependent variables
Designing procedure
12. S1
S2
S3
S4
S5
S6
4 week
Testing protocol
Treatment: Time
Numerical Reasoning English
Aptitude Compre
2 weekZero week
Numerical Reasoning English
Aptitude Compre
Numerical Reasoning English
Aptitude Compre
S1
S2
S3
S4
S5
S6
S1
S2
S3
S4
S5
S6
S1
S2
S3
S4
S5
S6
S1
S2
S3
S4
S5
S6
S1
S2
S3
S4
S5
S6
S1
S2
S3
S4
S5
S6
S1
S2
S3
S4
S5
S6
S1
S2
S3
S4
S5
S6
Case II: levels of the within-subjects variable are different time periods
When to use One-way rMANOVA
Example: To investigate the performance trend of subjects during a training programme for a
competitive examination.
DVs : Performance parameters (numerical aptitude, reasoning and English comprehension)
IV : Time(zero week, 2 week, 4 week)
Figure 7.2 Layout design
13. Steps in One-way rMANOVA
Test assumptions of design
Describe layout design
Write research questions to be investigated
Write hypotheses to be tested
Specify familywise error rates (α)
Use SPSS to generate outputs
Descriptive
statistics
MANOVA table containing
Wilk’s Lambda
Continue …
14. IsWilk’s Lambda
Significant
Terminate
further analysis
N
Y
Apply rANOVA for each
dependent variable
Use SPSS to generate following outputs
Mauchly's test
of sphericity
F table in rANOVA
for each dependent
variable
Pair-wise
comparisons of
means for each
dependent variable.
Means plot for each
dependent variable
Steps in One-way rMANOVA
15. Test Sphericity assumption
in each rANOVA
Is
p<α/k
Test F ratio by
assuming sphericity
N
Y
Check
<.75 Test F by using Huynh-Feldt
correction
NTest F by using Greenhouse-
Geisser correction
Y
If F is significant use Bonferroni correction for
comparison of means
Report findings
k: number of DVs
16. Table 7.1 Marks obtained by the students in different subjects tested at different times of the day
_____________________________________________________________________________________
Time of the day
Morning(7 AM) Afternoon(1 PM) Evening(7 PM)
_____________________________________________________________________________________
Maths English Reasoning Maths English Reasoning Maths English Reasoning
12 12 15 15 15 11 17 14 12
13 14 16 17 13 12 16 12 10
14 10 17 18 14 14 15 15 15
13 9 15 15 14 13 16 16 12
14 8 17 14 13 11 14 13 14
15 11 15 18 12 10 16 15 15
13 10 14 17 15 9 15 13 10
12 13 15 15 12 8 13 12 13
13 12 13 16 15 11 15 16 12
15 11 14 18 16 12 16 15 13
_____________________________________________________________________________________
Objective : To see the effect of time of the day on the student’s performance
in different subjects.
- An Illustration with SPSS
17. S1
S3
S2
S4
S5
S6
Evening
First phase
testing
S1
S3
S2
S4
S5
S6
S1
S3
S2
S4
S5
S6
Second phase
testing
Third phase
testing
Testing protocol
Treatment: Time of the day
Maths English Reasoning
S2
S4
S5
S6
S1
S3
S2
S4
S5
S6
S1
S3
S2
S4
S5
S6
S1
S3
S5
S6
S1
S3
S2
S4
S5
S6
S1
S3
S2
S4
S5
S6
S1
S3
S2
S4
AfternoonMorning
Maths English Reasoning Maths English Reasoning
All subjects are tested on all the three DVs but not in a particular sequence.
S1 and S3 are tested on all DVs in the morning, S2 and S4 in the afternoon and
S5 and S6 in the evening.
Similarly treatments(time) are randomized in other phases.
Procedure
Figure 7.3 Layout of the one-way rMANOVA design in the illustration
18. Whether time of testing affects student’s academic performance
together in all the three subjects?”
Whether time of testing affects student’s performance in each of
the subject; Maths, English and Reasoning?
Which time of the day improves performance of the students in
each subject?
19. 19
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