Multiple Draws without Replacement

Experiments
, , ,
1

Hi!

I am creating a rather complex survey, which requires multiple layers of randomisation, and wondered if anyone could help me figure out how to do this!

I have 8 DCEs, forming 8 different treatment conditions. Each of these DCEs is built from 12 blocks of 4 choice tasks (using NGene design).

I can so far randomly assign each respondent to one of the 8 treatment conditions, and then one of the 12 blocks within this treatment condition, by using two decks one after the other and using derived values to create design offsets for each block.

However, I would like each respondent to see 3 blocks, from three different treatment conditions (e.g. block 4 from treatment group 3, block 5 from treatment group 1, block 11 from treatment group 8).

Can I use Decks to achieve this? - Essentially, I need to be able to draw 3 numbers without replacement, rather than 1 at a time, to ensure that any one respondent sees 3 different blocks from 3 different treatment groups. Or is there an easier way?

Thank you!

2

Hi Olivia,

I don’t think there is a way to do that in one go. These easiest way I can think of is to use 3 decks of size 8, 7, 6. Use the first one to determine the first treatment group, then use the second one to pick one of the 7 remaining groups and the third to pick one of the remaining 6.

You can do something similar to draw three different blocks - in case the blocks are required to be all different, too (despite coming from different experiments already).

Cheers,
Lutz