I have created a survey with 11 different question groups. The first group has equations questions, that help me randomise if a question on the follow 10 groups is displayed or not (like seen here ). All groups from 2 to11 have a set of two questions, so in principal I have 20 questions. The idea is that each participant gets a random set of 10 questions, only one of each group, which are as well presented in a random order.
I want to use ANOVA to analyse my results, hence I want to make sure that each question has almost (or exactly) the same sample size. Since questions are fetch in a random order from the groups, there is a possibility for example, that Question 1.1 is presented to users more times (e.g double) than Question 1.2.
Is there a way to write a condition in my survey that it checks how many times per group, a question has been displayed and overwrite the randomisation function. For example, an user gets from the randomisation Question 1.1 but by checking the results from previous surveys this one has been presented more times than Question 1.2, so the randomisation is overruled and instead Question 1.2 is shown.
In principal what I want is that my sample sizes are as balanced as possible even if display randomisation is used.
Unfortunately, there is no direct access to previous responses from within a survey.
I suppose you could make an AJAX call to a remote script that queries the database and set hidden relevance questions based on the results but that would require significant coding and wouldn't solve the problem of concurrent responses.
Cheers,
Tony Partner Solutions, code and workarounds presented in these forums are given without any warranty, implied or otherwise.