Choice of block size

The choice of block size is clearly very important - with any block size the last treatment allocation of the block is always predictable, and a maximum of half the allocations will be predictable (for example with two treatments and a block size of 10, if the first five treatments within a block are treatment A, the next five must be treatment B). To avoid this predictability

♦ never reveal the block size to participants,

♦ avoid using the smallest possible block size,

♦ consider using varying block sizes.

Bear in mind, however, that the aim of block randomization is to keep a slightly tighter control on the overall balance of treatment allocations than is possible with simple randomization - yet the larger the block size, the less the control. For two treatments, equally allocated, a block size of 8-10 is usually adequate. If for any reason a long run of the same treatment allocation needs to be avoided, then remember that the maximum run is equal to the block size, being achieved when a block in which all the allocations for a particular treatment appear at the end is followed by a block in which they all appear at the beginning.

0 0

Post a comment