https://www.growthbook.io/ logo
h

handsome-library-89124

08/22/2022, 7:28 AM
Hi @here If we want to do sampling based on gender and the population on gender is not equal for example total population = 1000 male = 900 female = 100 What is the chance that female will randomize equally if we have 50/50 split of two variations?
f

fresh-football-47124

08/22/2022, 7:38 AM
you would expect to have 90/10 split in each variation if your a/b test is random
due to the small sample size, there might be some variation there, you could plot the probabilities of this is you need.
h

handsome-library-89124

08/22/2022, 7:43 AM
Yes, I am more concerned like female should divide in each variation as per 50/50 split. I am thinking like it will be not happened like 80 females in 1 variation and 20 in other.
m

miniature-agency-3305

09/01/2022, 11:56 AM
@handsome-library-89124 couldn’t you have two different samplings then? One for males and one for females?
and split those up 50/50
h

handsome-library-89124

09/01/2022, 12:10 PM
Yes, thats is another way but still for small data size its bit difficult.
2 Views