Editing counts on the randomizer | XM Community
Skip to main content

I would like to edit counts on the randomizer to correct for any unequal distributions as a consequence of participants not fully completing the survey. Details of how to do this are here:
https://www.qualtrics.com/support/survey-platform/survey-module/survey-flow/standard-elements/randomizer/#EditingCounts
However, I am not clear from these instructions exactly how to make it work. Let's say we have four conditions and need 100 participants, 25 in each condition. That is, we have 4 elements and want to present 1 of these 4 elements to each participant, such that each element is seen by an equal number of participants. We look at the element counts some time during data collection and see that the numbers are unequal because some people assigned to Conditions 2 and 3 ended the survey early:
Cond. 1: 11
Cond 2: 8
Cond 3: 3
Cond 4: 11
What are the calculations involved in deciding what changes to make to the counts for elements Condition 2 and Condition 4 so that they 'catch up' and we end up with 25 participants in each condition?
Many thanks in advance.


If the current randomizer counts are all equal, reduce the Condition 2 count by 3 (11-8) and the Condition 3 count by 8 (11-3).


Thanks TomG - so, just reduce each one by the 'shortfall'. Does it matter then that after the edit, the overall number of participants who have completed will appear to be lower than it actually is when we tot up the figures for all four conditions? - the system will handle this and will not under-count the total?
(Just to check, is 'randomizer count' different from the element count?)


ruthjlee1 ,
Reducing by the shortfall will effectively put them back to even.
Caveats:

  1. They may go out of balance again as you continue to field.

  2. Some of the 'shortfall' could be partials and if those partial come back and complete then you will have over adjusted by that amount.

You could put quotas of 25 on the 4 conditions that only count completes. Then you could terminate (overquota) those that get assigned to full conditions.


Thanks TomG, quota looks like a partial solution.


Leave a Reply