How do I use the randomiser to put participants randomly into two separate groups? | XM Community

How do I use the randomiser to put participants randomly into two separate groups?

  • 11 March 2018
  • 6 replies

Hi everyone, I am trying to set up survey where once participants complete some demographic questions, they are then randomly assigned to either a control group or an intervention group, where each group then completes a separate block. I cannot work out how to do it. I've tried using the section "Using the Randomizer with Other Elements" for information. If anyone could help me with some simple steps, I would greatly appreciate it?
Thanks, Sandra

Best answer by TomG 16 March 2018, 14:15

View original

6 replies

Userlevel 7
Badge +27

Under the randomizer in the survey flow, add two branches with conditions that are always true. Then under each include your embedded variable assignment and the question block:
Hi TomG,
Apologies for the delay in responding to your post. I think I understand what you mean. I will try it and post if I still have difficulties.
Thanks for your assistance, Sandra
Userlevel 7
Badge +6
@SJH - Sandra,
The way TomG set it up works because the logic for the randomizer is to display _only 1 of the 2 branching logics_, and for them to be _evenly distributed_ (this is the check that says "evenly present elements"). This means you will get a (near perfect) 50/50 split.

The only reason you need the branching logic in TomG's example is because he went ahead and created an embedded data field (group) that populates either "control" or "intervention" for the sake of easily identifying your population. Otherwise, you could just nest the 2 question blocks under the randomizer without the branching logic.

@TomG - I could be wrong, but I don't think you need to first create the embedded data "always" and populate it with something, just to create an always true condition. Your branching logic statements would still pass, even without an embedded data field value of "always" existing.
Userlevel 7
Badge +27
@Akdashboard - No you don't have to, but I add it for clarity. For someone looking it at, it might be confusing to compare to something that doesn't exist.
Hello @TomG
Thanks for your help on that one. Is the same logic also applicable for four branches? Or how could I model four different treatments whithin one survey?

Thanks a lot for your help.
Userlevel 7
Badge +27

Yes, the same process would work for any number > 1.

Leave a Reply