contrasts attribute of a factor variable.
contrasts(x) contrasts(x, how.many=<<see below>>) <- value
value.
The default is to make one less contrast than the number of levels in
x.
x.
This is commonly the result of one of the standard contrast functions, such as
contr.sum.
contrasts attribute.
contrasts attribute is changed when an assignment is made to it.
There are three functions that specify contrasts for factor variables in S-PLUS:
C,
contrasts, and
options.
Use the
C function to specify a contrast as you type a model formula.
The
contrasts function is used to specify a contrast attribute for a factor variable.
The
options function specifies the default choice of contrasts for all factors.
# assign polynomial contrasts to temp.
contrasts(temp) <- contr.poly(3)
# the contrast attribute from this command has 4 columns,
# which is probably not what was intended.
contrasts(treatment) <- contr.sum(5)[, 1:3]
# the contrast attribute from this command has 3 columns instead.
contrasts(treatment, how.many=3) <- contr.sum(5)[, 1:3]
# assign and extract contrasts from quality.
quality <- factor(c(rep("tested-low",10), rep("low",10),
rep("high",10), rep("tested-high",10)),
levels = c("tested-low", "low", "high", "tested-high"))
contrast.mat <- contr.helmert(4)
contrasts(quality) <- contrast.mat
contrasts(quality)