Weighted randomization for large number of variants

Question

Hello!
I was wondering if you have a suggestion for my problem: I have 25 variants of a survey (a planned missingness design) which each should be shown in different proportions to reach the correct sample numbers for each scale. This is is all already implemented within one survey link.
However if I create 25 codes within my randomizer item, I cannot assign weights how often each should be drawn. Some should be drawn at 4-5-6% some at 12% some at 20% probability. I have read in one of the other support responses that the easy workaround is to create a copy of the randomizer codes to the weights that I require, but with this many variants, it would be a long list - I was wondering if there's a workaround?

Answer 1

but with this many variants, it would be a long list

Not so long, probably. If you're counting the probability in percent, then it's no more than 100 entries long.

Theoretically, the random_p() funtion can draw with probabilities, but other than the random generator, it will not care for the actual proportions. That means, you would have the likely deviations from each probability within statistical likelihood. The ranom generator with 100 codes is -in my opinion - a much more reliable option.

Comments

I just double checked and indeed I have it in hundredth not thousandth so you are absolutely right. I'll go with the generator in that case, thanks a lot!