How does editing counts with Randomizer work? | XM Community
Skip to main content
Solved

How does editing counts with Randomizer work?

  • 2 April 2021
  • 2 replies
  • 351 views

I'm running a study with 3 experimental conditions, and currently, one condition is running significantly behind in number of participants. Several times, I've attempted to edit participant counts to fix this--my understanding was that if I set the count for that condition lower than the others, that the program would send more participants to it to correct for this difference. However, the differential seems to just keep growing. The explanations I've found online for how the count works have not been clear--does making the count higher for a condition INCREASE the number of people sent to that condition, or decrease it? Further, what if I now want to send participants to one of two (of my three) conditions, so as to balance things out and not pay for any more participants from the condition I already have more than enough data for?
Thanks!

Making the count higher would decrease the number assigned to a condition.
Based on the behavior you describe, it sounds like you have qualifying conditions on the items within the randomizer and those override the even presentation. The condition for the lagging item is true less often than than others. For example:
Randomizer - Evenly present
Branch Condition 1
Branch Condition 2
Branch Condition 3
If the branch condition 3 is true less often than 1 and 2, it's count is going to be lower.

That makes sense, but I didn't set any extra qualifications for any of the conditions (at least as far as I know/can tell)--participants in each condition are just shown slightly different material before being directed to a set of questions. Conditions 2 and 3 are required to do a bit more reading than Condition 1, so it's possible, I guess, that it just has a higher dropout rate. That would make sense. Nonetheless, I'll check whether there's something else wrong. Mostly, I wanted to make sure my understanding of counts was correct, and it sounds like it was! Thanks--now I can get more participants just for the conditions I still need more of!

Leave a Reply


Terms & Conditions