Hi,
what would be a good way of enhancing the ancova function from the HH
package when using a 2 factor ANCOVA?
The current behaviour for the "ancova" function from package HH is:
----------------------------------------------
| P1 || P1 || P3 || P4 | | PS | <- the lattice strip
----------------------------------------------
| x|| x ||x ||x | | |
| x ||x || x || x | |superimp.| <- the lattice plot
|x || x || x || x | | |
----------------------------------------------
P1:P4 : Show an xy-plot for each factor level of the covariate P.
PS: Displays the previous plots superimposed
superimp: P1:P4 superimposed
I would like to use this function for a two factor design.
-------------------------------------------------
| Q1 || Q1 || Q1 || Q1 | | PS.Pi | <- strip1
-------------------------------------------------
| P1 || P1 || P3 || P4 | | PS.Pi | <- strip2
-------------------------------------------------
|x || x || x || x | | |
|x || x || x || x | | | <- plot
|x || x || x || x | | |
-------------------------------------------------
-------------------------------------------------
| Q2 || Q2 || Q2 || Q2 | | PS.Pi | <- strip1
-------------------------------------------------
| P1 || P1 || P3 || P4 | | PS.Pi | <- strip2
-------------------------------------------------
|x || x || x || x | | |
|x || x || x || x | | | <- plot
|x || x || x || x | | |
-------------------------------------------------
[...]
-------------------------------------------------
| Q5 || Q5 || Q5 || Q5 | | PS.Pi | <- strip1
-------------------------------------------------
| P1 || P1 || P3 || P4 | | PS.Pi | <- strip2
-------------------------------------------------
|x || x || x || x | | |
|x || x || x || x | | | <- plot
|x || x || x || x | | |
-------------------------------------------------
Here is an example:
--8<---------------cut here---------------start------------->8---
rm(list = ls(all = TRUE))
rm(list = c(ls()))
library(lattice)
library(HH)
## 1. generate data
random <- rnorm(200)
y <- abs(random)
x1.cont <- abs(random)
x2.fac <- as.factor(rep(1:5, 4)) # 4 groups
x3.fac <- as.factor(rep(1:4, each=5))# 5 groups
A <- data.frame(y, x1.cont, x2.fac, x2.fac)
## 2. plot trellis plot:
foo <- xyplot(log(y) ~ log(x1.cont) | x2.fac * x3.fac,
data = A,
type = c("p","r"),
strip = strip.custom(strip.names=TRUE),
as.table=TRUE)
plot(foo)
## 3. plot "ancova" plot using HH:
## normal usage:
ancova(y ~ x1.cont * x2.fac)
## 4. Trying to combine "ancova" with trellis:
## does not work:
ancova(y ~ x1.cont * x2.fac * x3.fac)
--8<---------------cut here---------------end--------------->8---
--
Patrick Drechsler
Department of Zoology
University of Cambridge
Downing Street
Cambridge CB2 3EJ, UK