Specifying terms involving covariates
main topic
 

You can specify variables to be covariates in General MANOVA. You must specify the covariates in Covariates, but you can enter the covariates in Model, though this is not necessary unless you cross or nest the covariates (see table below).

In an unbalanced design or a design involving covariates, General MANOVA's sequential sums of squares (the additional model sums of squares explained by a variable) will depend upon the order in which variables enter the model. If you do not enter the covariates in Model when using General MANOVA, they will be fit first, which is what you usually want when a covariate contributes background variability. The subsequent order of fitting is the order of terms in Model. The sequential sums of squares for unbalanced terms A B will be different depending upon the order that you enter them in the model. The default adjusted sums of squares (sums of squares with all other terms in the model), however, will be the same, regardless of model order.

General MANOVA allows terms containing covariates crossed with each other and with factors, and covariates nested within factors. Here are some examples of these models, where A is a factor.

Case

Covariates

Terms in model

test homogeneity of slopes (covariate crossed with factor)

X

A  X  A * X

same as previous

X

A | X

quadratic in covariate (covariate crossed with itself)

X

A  X  X * X

full quadratic in two covariates (covariates crossed)

X  Z

A   X  Z  X * X  Z * Z  X * Z

separate slopes for each level of A (covariate nested within a factor)

X

A  X (A)