Professional Documents
Culture Documents
(two-group MANOVA)
Haramaya University
March 30,2109
Two-group Manova (1)
Two groups of subjects
experimental, comparison; or
male, female
Several DVs (correlated and conceptually related)
Could be different dimensions of a construct, like an
achievement test
Conceptual knowledge
Procedural knowledge
Independence
HOV: implies 1 2 p
2 2 2
t-test statistic
(Y1 Y2 ) ( 1 2 ) (Y1 Y2 )
tobs
1 1 1 1
S p2 ( ) S p2 ( )
n1 n 2 n1 n 2
where
S 2
( n 1) S 2 ( n2 1)
2
Pooled or common
Sp
2 1 1
within-groups variance
n1 n2 2
Y1 Used to estimate
1
Y2 2
H0: 1 = 2
Multivariate? Imagine 2 DVs
Yijq = ith persons score in jth group on the qth DV
DV1 DV2
y111 y112
y211 y212
Group 1
yn1,1,1 yn1,1, 2
Y( n1 n 2) x 2
y121 y122
y221 y222
Group 2
yn 2, 2,1 yn 2, 2, 2
Multivariate Predictions
DV1 DV2 DV1 DV2
Y11 Y12 11 12 Group 1
Y2 x 2 2 x 2
Y21 Y22 21 22 Group 2
• Each group has a row vector that contains the means for each DV.
• Use the transpose of the row vectors to indicate how we compare
means across the groups on the collective set of DVs.
11 21 H 0 : 1 2
H 0 :
12 22 Group 1 Group 2
Group 1 Group 2
Q DVs, 2 groups
In general, this looks
like: centroids
11 21
H 0 : 12 22 H 0 : 1 2
1Q 2Q
Each vector contains means for Q DVs. These mean column vectors
are called “centroids.”
and
nj J
SCP (Y1 , Y2 ) ( yij1 Y..1 )( yij 2 Y..2 )
i 1 j 1
SSCP for each group?
Imagine X has 3 levels (or groups)
Treatment1, treatment2, comparison
Then HOC says that partitioning the total SSCP for Y
can be used to reasonably create a “pooled” SSCP
(similar to univariate case (S2p))
Why is HOC important?
Makes detection of differences fair, in the sense that
differences in means could truly be called differences
in means and not be the result of an artifact of one
group having greater - or less - variability than another
group!
If one group has too much spread or variability relative
to the other groups, may not be able to detect
differences in means.
SSCP (groups) = Wj
• Within each group, look at differences from group means.
• Notice size (order) of W’s remains the same as the total SSCP matrix
(still only 2 DVs!)
Box-M test asks if the Wj’s are all similar (i.e., all
contribute equally to W)
Actual test uses covariance matrices instead of SSCP
matrices 1
Covariance matrix is W j ˆ j
df j
n1 n 2
n1 n 2
( y1 y2 ) S p2
1
( y1 y2 )
Hotellings T2
Multivariate Case: Still want to subtract group means but
now on several variables (subscripts refer to groups).
n1 n2 ˆ 1
T 2
(Y 1 Y 2 ) (W )(Y 1 Y2 )
n1 n2
Column vectors of means for
each group
ˆ W1 W2 W
W (note: inverse is of a matrix)
n1 n2 2 df
Note: for three or more groups, add all the Wj’s together to find W; df is
sum of (nj-1).
Test using Hotellings T2
Hotelling showed that T2 can be transformed to an exact F.
n1 n2 (Q 1) 2
Fobs T ~ FQ , N (Q 1)
(n1 n2 2) Q
Omnibus test is Hotellings T2
Need follow-up tests to see where (which DV) the specific
group differences might be
With only two groups, obvious choice are the familiar
independent t-tests
Note: some textbooks use “p” to represent the # of variables;
here I’ve used “q” to be consistent with the terminology we’ve
used for # of DVs
Omnibus test
11 12 H 0 : 1 2
H 0 :
21 22 Group 1 Group 2
Group 1 Group 2
• In this example there are only 2 DVs (because there are only 2
rows in each column vector).
• This test can be used regardless of the number of DVs (more
than 2)
• T2 Test was developed for situations of only 2 GROUPS
• Use Wilk’s Lambda as general test: 2 or more groups!