Some surveys call for prefilling certain questions with random answers. This is often done to randomly place respondents in different groups and alter their routing, to not show all respondents the same items, to show them the items in a different order, to show or not show certain answer options, to …
Our randomization takes into account which answer choice has been selected the least. Why? Because, when you flip a coin 10 times, you could get ‘heads’ 8 times. In survey, it is better if each answer choice has an equal number of responses. So how is that random? Well, the system selects a random answer choice with the fewest number of responses. An example will make it more clear: We have a radio buttons question with four answer choices (group 1, group 2, group 3, group 4). When the first respondent answers, all four answer choices have 0 respondents, so the system will randomly select 1 of the 4. Say he gets ‘group 3’. The second respondent will get a random answer from group 1, 2, or 4 because those now have the fewest number of respondents and so on.
We have 4 different kinds of randomization:
- Random – fewest respondents: selects a random answer option with fewest number of responses.
- Random – fewest completes: selects a random answer option with fewest number of responses from respondents that reached the end of the survey.
- Random – fewest per language: selects a random answer option with fewest number of responses for each language separately.
- Random – fewest completes per language: selects a random answer option with fewest number of responses from respondents that reached the end of the survey and this per language.
The spreading of the respondents may not be exactly equal. The preselection occurs based on the response counts at the moment the respondent gets to the page with this question. Their response in turn, will not be added to the response counts, until they click on ‘next’ or finish the survey if the preselection is based on completes.