Define
x <- data.frame(
ID=letters[1:10],
class = as.factor(c(rep(1,5),rep(2,5))),
treat = rep(0,10))
s.t.
> x
ID class treat
1 a 1 0
2 b 1 0
3 c 1 0
4 d 1 0
5 e 1 0
6 f 2 0
7 g 2 0
8 h 2 0
9 i 2 0
10 j 2 0
I have a treatment with two levels, 1 & 2. I want to assign exactly one one unit per class to each level s.t. that, after randomization, we get something like:
> x
ID class treat
1 a 1 0
2 b 1 0
3 c 1 1
4 d 1 0
5 e 1 2
6 f 2 0
7 g 2 0
8 h 2 0
9 i 2 2
10 j 2 1
s.t. units c and j get level 1 of treatment and e and i level 2.
How do I generate the treatment vector in R?
I'll assume you just want to assign one level 1 treatment and one level 2 treatment in each class. You can use the ddply function from the plyr package to do it easily:
set.seed(1)
require(plyr)
> ddply(x, .(class), transform,
treat = replace(treat, sample(seq_along(treat),2), 1:2))
ID class treat
1 a 1 0
2 b 1 1
3 c 1 0
4 d 1 0
5 e 1 2
6 f 2 0
7 g 2 0
8 h 2 1
9 i 2 2
10 j 2 0
To explain: the ddply function splits the data-frame by the class variable, and within each data-frame, it "transforms" the treat column by replacing 2 randomly chosen entries by 1 and 2. The sample(...,2) function picks two random indices in the treat column. Other variants (e.g. assign more than 1 of each treatment type) can be done similarly.