Closed macartan closed 6 years ago
I'm actually just going to implement this as one argument, assignment_probs
. When it's a scalar, block probs are equal, and when it's a vector of length N_blocks
, then those are the probs used in the respective blocks ("as determined by sort(unique(blocks))").
If you have two blocks, for example, assignment_probs = .5
gets you 50/50 in each, while assignment_probs = c(.1,.9)
gets you 10% treated in first and 90% in second.
Adding this. I think, given that we are in a two-arm world, we can just use block_prob though:
from randomizr help: