R help - Aug 2006 - how to run ANCOVA?

Dear all,

  I would like to know how to run an analysis of covariance in R.  For
example, I have a data frame ("data") consisting of two second-degree
categorical variables ("diagnosis" and "gender"), one
continous
independent variable ("age") and one continous dependent variable
("response").

I ran a simple anova to see the effects of diagnosis and gender (and
interaction):

aov.out <- aov(response~diagnosis*gender,data)
anova(aov.out)

Now I would like to covary for age, how can I add "age" as a covariate
to this equation?

Thanks,
Sasha

>  aov.out <- aov(response~diagnosis*gender,data)
Just add it where you think it belongs in the
sequential sum of squares

To adjust the factors for the covariate use
aov.out <- aov(response ~ age + diagnosis*gender, data)

To adjust the covariate for the factors
aov.out <- aov(response ~ diagnosis*gender + age, data)

If you want to check for interaction of the factors with the
covariate, then use * instead of + in the formula.

Please note that I added spaces to your statement to improve human legibility.

Rich

> But how is that different from just a 3-way ANOVA with age, diagnosis,
> and gender as the the three effects?  Isn't ANCOVA a fundamentally
> different model?
> Thanks,
> Sasha
ANCOVA is a linear model with both factors and continuous variables
on the right-hand side of the model formula.  In pre-computer texts,
the commonly used algorithm hid that fact.

Rich