1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
|
\name{contr}
\alias{contr.treatment}
\alias{contr.sum}
\alias{contr}
\alias{contrasts}
\alias{contrasts,item-method}
\alias{contrasts,ANY-method}
\alias{contrasts<-}
\alias{contrasts<-,item-method}
\alias{contrasts<-,ANY-method}
\title{Convenience Methods for Setting Contrasts}
\description{
This package provides modified versions of
\code{\link[stats:contrast]{contr.treatment}} and
\code{\link[stats:contrast]{contr.sum}}. \code{contr.sum}
gains an optional \code{base} argument, analog to the
one of \code{contr.treatment}, furthermore,
the \code{base} argument may be the name of a
factor level.
\code{contr} returns a function that calls either
\code{contr.treatment}, \code{contr.sum}, etc.,
according to the value given to its first argument.
The \code{contrasts} method for \code{"item"} objects
returns a contrast matrix or a function to produce
a contrast matrix for the factor into which
the item would be coerced via \code{as.factor} or \code{as.ordered}.
This matrix or function can be specified by
using \code{contrasts(x)<-value}
}
\usage{
contr(type,\dots)
contr.treatment(n, base=1,contrasts=TRUE)
contr.sum(n,base=NULL,contrasts=TRUE)
\S4method{contrasts}{item}(x,contrasts=TRUE,\dots)
\S4method{contrasts}{item}(x,how.many) <- value
# These methods are defined implicitely by making 'contrasts' generic.
\S4method{contrasts}{ANY}(x,contrasts=TRUE,\dots)
\S4method{contrasts}{ANY}(x,how.many) <- value
}
\arguments{
\item{type}{a character vector, specifying the type of the contrasts.
This argument should have a value such that, if e.g. \code{type="something"},
then there is a function \code{contr.something} that produces
a contrast matrix.
}
\item{\dots}{further arguments, passed to \code{contr.treatment}, etc. }
\item{n}{a number of factor levels or a vector of factor levels names, see e.g. \code{\link[stats:contrast]{contr.treatment}}.}
\item{base}{a number of a factor level or the names of a factor level,
which specifies the baseline category,
see e.g. \code{\link[stats:contrast]{contr.treatment}} or NULL.
}
\item{contrasts}{a logical value, see \code{\link[stats]{contrasts}}}
\item{how.many}{the number of contrasts to generate, see \code{\link[stats]{contrasts}}}
\item{x}{a factor or an object of class "item"}
\item{value}{a matrix, a function or the name of a function}
}
\value{
\code{contr} returns a funtion that calls one of \code{contr.treatment},
\code{contr.sum,\dots}.
\code{contr.treatment} and \code{contr.sum} return contrast matrices.
\code{contrasts(x)} returns the "contrasts" attribute of an
object, which may be a function name, a function, a contrast matrix or NULL.
}
\examples{
ctr.t <- contr("treatment",base="c")
ctr.t
ctr.s <- contr("sum",base="c")
ctr.h <- contr("helmert")
ctr.t(letters[1:7])
ctr.s(letters[1:7])
ctr.h(letters[1:7])
x <- factor(rep(letters[1:5],3))
contrasts(x)
x <- as.item(x)
contrasts(x)
contrasts(x) <- contr.sum(letters[1:5],base="c")
contrasts(x)
missing.values(x) <- 5
contrasts(x)
contrasts(as.factor(x))
# Obviously setting missing values after specifying
# contrast matrix breaks the contrasts.
# Using the 'contr' function, however, prevents this:
missing.values(x) <- NULL
contrasts(x) <- contr("sum",base="c")
contrasts(x)
missing.values(x) <- 5
contrasts(x)
contrasts(as.factor(x))
}
|
