Repeated Measures

Anova in Physical
Therapy Using SPSS
By: I Putu Gde Surya Adhitya, S.Ft., M.Fis., Ph.D.

Department of Physical Therapy, College of Medicine, Universitas Udayana

Definition

•It is a technique used to test the equality of means of more

than 2 groups.
– It is used when all members of a random sample are measured
under a number of different conditions.
– As the sample is exposed to each condition in turn, the
measurement of dependent variable is repeated.
 When to Use

Measuring performance on the same variable over time

– for example looking at changes in performance during training or before and after a
specific treatment
The same subject is measured multiple times under different
conditions
– for example, first exam, midterm exam, and final exam.
The same subjects provide measures/ratings on different
characteristics
– for example the desirability of red cars, green cars and blue cars
Note how we could do some RM as regular between subjects
designs
– For example, Randomly assign to drug A or B
 Assumptions
• Continuous outcome variables and categorical
independent variables are the basic requirements
• Independent observations or, precisely, Independent and
identically distributed variables
• No significant outliers in any combination of the related
groups
• Normality: the test dependent variables follow a
multivariate normal distribution in the population
• Sphericity: the variances of all difference scores among the
test variables must be equal in the population.
– Sphericity is sometimes tested with Mauchly’s test: the variances of the
differences between condition are equal.
 The test of Sphericity

There is some deviation from sphericity:

The variance of the differences between conditions P1 and P3
(16.0) is greater than the variance of the differences between
P1 and P2 (11.9) and between P2 and P3 (1.9).
 What is the effect of
violating the assumption
of Sphericity?
•Loss of power
increased probability of a Type II error
•A test statistic (F‐ratio) that simply cannot be
compared to tabulated values of the F‐ distribution
 Assessing the severity of departure from
Sphericity
• Three ways
– Greenhouse‐Geisser (1959)
– Huynh‐Feldt (1976)
– The lower Bound estimate ( the lowest possible theoretical
value for the data)

• The G‐G and H‐F estimates can both range from the
lower bound (the most severe departure from
sphericity possible given the data) and 1 (no
departure from sphericity at all)
 Different types of RMANOVA
•One‐way
– Measured a dependent variable over two or more
time points.
•Two‐way
– Measured a dependent variable over two or more
time points.
– All subjects have undergone two or more conditions.
•Mixed
– Measured a dependent variable over two or more
time points.
– Subjects have undergone either one of conditions.
 One‐way RMANOVA

•When you only wanted to know whether there was a

difference in chronic back pain before and after a back
rehabilitation course

•You only have one factor, which is "time", where you are
comparing three time points: "before“, “midway”, and
"after" the back rehabilitation course.
 Two‐way RMANOVA

• To determine whether any change in back pain (i.e., the dependent variable) is the
result of the interaction between the "type of treatment" (i.e., the massage
program or acupuncture program) and "time" (i.e., our second factor).

• Two "within‐subjects" factors

– All subjects undergo all conditions (e.g., if the study has two conditions – a control and a
treatment – all subjects take part in both the control and the treatment
 Mixed RMANOVA

• To understand if there is an interaction between your within‐subjects

factor and between‐subjects factor on the dependent variable.

• Subjects only receive one "condition": either the control or the

treatment).
 Comparis
on
Types Factors Meaning
One‐way Time Changes in dependent variable
over time (one within‐subject
factor)
Two‐way Group x time All subjects undergo the same
conditions over time (two
within‐ subjects factors)

Mixed Group x time Subjects receive either A or B

group over time (one within‐
subject factor and one between‐
subject factor)
One‐way RMANOVA
 Exercise 1
•Sleep problem is a common complaint in
patients following brain injury.
•Participants completed sleep surveys at four times
points (1, 3, 6, and 12M) after brain injury.
•Specifically, you want to know whether their sleep
at different time points were different.
 Exercise 1
•Within‐subjects design
•Dependent variable: paticipants’ rating of
sleep.
•Use “sleep_rating_one‐way.sav”.

https://drive.google.com/file/d/1TUH_6FWCMoT0NRXr9JN3URz-sxAqK
zAn/view?usp=sharing
Each level of a within subjects factor is entered as a
separate variable (2S01l9e/9e/2p7 1, 2, 3, 4).
Click Analyze > General Linear Model > Repeated Measures...
Name and Define the Within Subjects Factors

Click “Add” to
enter each
within subjects
factor.

Click “Define”
to define
Within
Subjects
Factors.
Defining Within Subjects Factors

Within
Subject
s
Factors

Cli
ck
O
pti
on
s
Repeated Measures:
Options
•Click the button.
•This will generate the output.
 SPSS Output: Descriptive
General Linear Model
 Assumptions: Sphericity?

Mauchly’s Test, X2(5) = 4.045, p = 0.543, did not indicate any violation of
sphericity.

The amount of sphericity is estimated by epsilon (the Greek letter ‘e’ and
written as ε). There are different ways for estimating it, including the
Greenhouse‐Geisser, Huynh‐Feldt and lower bound methods. If sphericity is
violated, these are used to correct the within‐subjects tests as we'll see below.
 SPSS Output: Within Subjects Factors

• As sphericity was not violated then the statistics was F(3,

117) = 15.372, p = .000, indicating that sleep scores were
not equal in the population.
• If Sphericity was violated, we use the adjusted values of
Greenhouse- Geisser, Huynh-Feldt, and lower bound .
 Post hoc
Tests

• Sleep score in the 1 M was significantly lower than those in the 6 and 12 M.
• Sleep score in the 3 M was significantly lower than those in the 6 and 12 M.
•2019S/9le/2e7 p scores in the 6 and 12 M were not significantly
different.
 Reporting One‐Way RMANOVA
• Mauchly’s test indicated that the assumption of
sphericity had not been violated (Chi2 = 4.045, P =
0.543).
• The results showed that there was a significant
upward changes in sleep among individuals after
brain injury (F = 15.372, P < 0.001).
• Specifically, sleep in 1 month was poorer than 6 and
12 months. Sleep in 3 month was worse than6 and
12 months.
Mixed RMANOVA
 Exercise 2
• Imagine that you are a researcher who is interested
in the effects of one‐week warm footbath on the
sleep of individuals with brain injury compared to
those of receiving sleep hygiene education.
• The participants in both groups competed the sleep
survey at 4 times points: before the treatment and 1
W, 2W, and 4W after the treatment.
• Specifically, you want to know whether the sleep
changes were different between two groups.
 Exercise 2
•Within‐ and Between‐ subjects design
•Independent variable: group (warm footbath and
sleep hygiene education)
•Dependent variable: paticipants’ rating of
sleep
•Use “sleep_rating_one‐way.sav”.
https://drive.google.com/file/d/1TUH_6FWCMoT0NRXr9JN3URz-sxAqKzAn/vi
ew?usp=sharing
 Data Entry

Each level of a within and between subjects factor is

entered as a s2e10p/9a/9r2a7 te variable. Sleep 1, 2, 3, 4
Click Analyze > General Linear Model > Repeated
Measures...
Name and Define the Within Subjects
Factors
Click “Add” to
enter each
within subjects
factor.

Click “Define”
to define both
Within and
Between
Subjects
Factors.
Defining Within & Between
Subjects Factors

Within
Subject
s
Factors

Between
Subjects
Factors
(group)
Repeated Measures: Plots
Repeated Measures:
Post hoc

Note: If your between‐subjects factor only has two groups, you will not need to run any
post hoc tests. Remember, these post hoc tests are for the main effects and not the
20i1n9t/e9/r2a7ction (i.e., they are not simple main effects). 25
Repeated Measures: Save
Repeated Measures:
Options
•Click the button.
•This will generate the output.
 Homogeneity assessment

Mauchly’s Test, X2(5) = 3.998, p = 0.55, did not indicate any

violation of sphericity.
Main analyses: repeated
measures
SPSS Output: Within and between Subjects Factors
Main effect of sleep

Main effect of sleep

Group X Sleep
interaction

• The value of F was 15.008 (p = .000) indicated that sleep scores

were not equal in the population.
• The value of F was 0.077 (p = .972) indicated that sleep scores
were equal between groups in the population.
Can you find the source of the
interaction?

Score of sleep questionnaire

SPSS Output: Between Subjects Effects
Sleep score mean at 4 time
points and the pairwise
comparisons

• Score at Time 1 (before) was significantly lower than those at Time 3 and 4 (2W and 4 W)
• Scores at Time 2 (1W) was significantly lower than those at Time 3 and 4 (2W and 4W) .
• Scores at Time 3 and Time 4 (2W and 4W) were not significant different.
 Reporting Mixed RMANOVA
• Mauchly’s test indicated that the assumption of
sphericity had not been violated (Chi2 = 3.998, P =
0.55).
• Overall, the analysis revealed a significant upward trend
in sleep (F(3, 114) = 15.008, p < 0.001). However, warm
footbath group did not have higher scores of sleep
compared to sleep hygiene education group did (F =
2.801, p = 0.102). The interaction between group and
sleep score failed to reach statistical significance (p =
0.972).
Two‐way RMANOVA
 Exercise 3
• You are interested in the effects of one‐week warm footbath
on the sleep of individuals with brain injury compared to
those of receiving sleep hygiene education.

• The participants in both groups competed the sleep survey at

3 times points: before the treatment and 1 W and 2W after
the treatment.
• All participants undergo warm footbath and sleep hygiene
education but the order in which they received this differs
(cross‐over design).
• Specifically, you want to know whether the sleep changes
were different between two groups.
 Exercise
3
• Full within‐subjects design
• Independent variable: group (warm footbath and sleep hygiene
education)
• Independent variable: gender (male and female)
• Dependent variable: paticipants’ rating of sleep
• Use “sleep_rating_two‐way.sav”.

https://drive.google.com/file/d/18SuzFzxfEti8iQTghEqs
_osuk_U8O3qP/view?usp=sharing
 Data Entry

Each level of within-subjects factors is entered as a separate variable.

Click Analyze > General Linear Model > Repeated Measures...
 Name and Define the Within Subjects Factors

Click “Add” to enter

each within subjects
factor.

Click “Define” to define

Within Subjects Factors.
Defining Within Subjects
Factors

Within
Subjects
Factors
Repeated Measures: Plots
Repeated Measures:
Save
Repeated Measures: Options
•Click the button.
•This will generate the output.
 SPSS Output: Descriptive
General Linear Model
 Homogeneity
assessment

Mauchly’s Tests (p = 0.156 and 0.321) did not indicate any

violation of sphericity regarding time and the interaction of
time by group.
Main analyses: repeated measures
• The value of F was 10.804 (p = .000) indicated that sleep scores
were not equal in the population.
• The value of F was 16.040 (p = .000) indicated that sleep scores
were not
Can you find the source of the interaction?

Score of sleep questionnaire

Warm
footbath

Sleep hygiene
education
Sleep score mean at 3 time
points and the pairwise
comparisons

• Score at Time 1 (before) was significantly lower than those at Time 2 and 3
(1W and 2 W)
•20S19c/o9/r2e7 s at Time 2 and Time 3 (1w and 2W) were not significantly
different.
53
 Reporting Two‐way RMANOVA
• Mauchly’s test indicated that the assumption of
sphericity had not been violated.
• The analysis revealed that the sleep scores were
significantly different among before and after
treatment (F = 10.804, p < 0.001). However, sleep
scores were not significantly different between
groups (F = 0.85, p = 0.772).
• The interaction between time and group (F =
16.04, p
<0.001) reached statistical significance.
Thank
You!