When setting up experiments, you can define different percentages for each experimental condition. The platform will initially allocate cases to each experimental condition according to the percentages defined.
As cases get resolved, there's likely a certain bias towards a mode or a questionnaire length (IVR assigned respondents or shorter questionnaires might get answered faster). As cases are resolved, the new cases that are added to the queue will be assigned the same percentages defined in the survey setup regardless of the experimental condition that was assigned to the resolved cases.
An example:
- You setup an experiment with 30% cases assigned to SMS and 70% cases assigned to IVR.
- The first batch of 100 cases will be assigned: 30 to SMS, 70 to IVR.
- The first 10 cases that get completed had all been assigned to IVR.
- The platform will enqueue 3 new cases for SMS and 7 for IVR.