When the number of treatment conditions evenly divides the size of a block,
the conditions will be repeated that many times within the block. For example,
with three conditions, c("T1", "T2", "C")
, and a block with six units,
two units will be assigned to each condition.
When the number of treatment conditions does not evenly divide the block size,
the conditions are repeated up to the closest multiple lower than the block
size and the remaining conditions are chosen at random. For example, with the
three conditions from above and a block with four units, each condition will be
repeated once (since floor(4/3) == 1
). One additional condition is needed
to assign all units in the block, and that condition is selected at random from
c("T1", "T2", "C")
with equal probability. In a block with 8 units, each
condition will be repeated twice (floor(8/3) == 2
). Two additional
conditions are now needed, and they are chosen from c("T1", "T2", "C")
without replacement.
In all cases, the treatment conditions within a block are shuffled so that all
units have the same probability of being assigned to each condition. Units not
assigned to blocks will not be assigned treatments (indicated by NA
).