8.

2 - The Multivariate Approach: One-way

Multivariate Analysis of Variance (One-way
MANOVA)
Printer-friendly version

Now we will consider the multivariate analog, the Multivariate Analysis of Variance, often
abbreviated as MANOVA.

Suppose that we have data on p variables which we can arrange in a table such as the one below:

In this multivariate case the scalar quantities, Yij, of the corresponding table in ANOVA, are
replaced by vectors having p observations.

Notation

Yijk = Observation for variable k from subject j in group i. These are collected into vectors:

Yij = ⎛⎝⎜⎜⎜⎜⎜Yij1Yij2⋮Yijp⎞⎠⎟⎟⎟⎟⎟ = Vector of variables for subject j in group i

ni = the number of subjects in i

N = n1 + n2 + ... + ng = Total sample size.

Assumptions
The assumptions here will be essentially the same as the assumptions in a Hotelling's T2 test,
only here they will apply to groups:

1. The data from group i has common mean vector μi=⎛⎝⎜⎜⎜⎜μi1μi2⋮μip⎞⎠⎟⎟⎟⎟

2. The data from all groups have common variance-covariance matrix Σ.
3. Independence: The subjects are independently sampled.
4. Normality: The data are multivariate normally distributed.

Here we are interested in testing the null hypothesis that group mean vectors are all equal to one
another. Mathematically this is expressed as:

H0:μ1=μ2=⋯=μg

The alternative hypothesis being:

Ha:μik≠μjk for at least one i≠j and at least one variable k

This says that the null hypothesis is false if at least one pair of treatments is different on at least
one variable.

Notation

The scalar quantities used in the univariate setting are replaced by vectors in the multivariate
setting:

Sample Mean Vector

y¯i.=1ni∑nij=1Yij=⎛⎝⎜⎜⎜⎜⎜y¯i.1y¯i.2⋮y¯i.p⎞⎠⎟⎟⎟⎟⎟ = sample mean vector for group i . This

sample mean vector is comprised of the group means for each of the p variables. Thus,
y¯i.k=1ni∑nij=1Yijk = sample mean vector for variable k in group i .

Grand Mean Vector

y¯..=1N∑gi=1∑nij=1Yij=⎛⎝⎜⎜⎜⎜⎜y¯..1y¯..2⋮y¯..p⎞⎠⎟⎟⎟⎟⎟ = grand mean vector. This grand

mean vector is comprised of the grand means for each of the p variables. Thus,
y¯..k=1N∑gi=1∑nij=1Yijk = grand mean for variable k.

Total Sum of Squares and Cross Products

In the univariate Analysis of Variance, we defined the Total Sums of Squares, a scalar quantity.
The multivariate analog is the Total Sum of Squares and Cross Products matrix, a p x p matrix of
numbers. The total sum of squares is a cross products matrix defined by the expression below:
 T=∑i=1g∑j=1ni(Yij−y¯..)(Yij−y¯..)′

Here we are looking at the differences between the vectors of observations Yij and the Grand
mean vector. These differences form a vector which is then multiplied by its transpose.

Here, the (k, l)th element of T is

∑i=1g∑j=1ni(Yijk−y¯..k)(Yijl−y¯..l)

For k = l, this is the total sum of squares for variable k, and measures the total variation in the kth
variable. For k ≠ l, this measures the dependence between variables k and l across all of the
observations.

We may partition the total sum of squares and cross products as follows:

T===∑gi=1∑nij=1(Yij−y¯..)(Yij−y¯..)′∑gi=1∑nij=1{(Yij−y¯i)+(y¯i−y¯..)}{(Yij−y¯i)+(y¯i−y¯..)}′∑gi
=1∑nij=1(Yij−y¯i.)(Yij−y¯i.)′
E+∑gi=1ni(y¯i.−y¯..)(y¯i.−y¯..)′ H

where E is the Error Sum of Squares and Cross Products, and H is the Hypothesis Sum of
Squares and Cross Products.

The (k, l)th element of the error sum of squares and cross products matrix E is:

∑i=1g∑j=1ni(Yijk−y¯i.k)(Yijl−y¯i.l)

For k = l, this is the error sum of squares for variable k, and measures the within treatment
variation for the kth variable. For k ≠ l, this measures the dependence between variables k and l
after taking into account the treatment.

The (k, l)th element of the hypothesis sum of squares and cross products matrix H is

∑i=1gni(y¯i.k−y¯..k)(y¯i.l−y¯..l)

For k = l, this is the treatment sum of squares for variable k, and measures the between treatment
variation for the kth variable,. For k ≠ l, this measures dependence of variables k and l across
treatments.

The partitioning of the total sum of squares and cross products matrix may be summarized in the
multivariate analysis of variance table:

MANOVA
Source d.f. SSP
Treatments g - 1 H
 N-
Error E
g
N-
Total T
1

We wish to reject

H0:μ1=μ2=⋯=μg

if the hypothesis sum of squares and cross products matrix H is large relative to the error sum of
squares and cross products matrix E